Fish were collected at two locations in Czechia, which contained either just C. carassius or both native and invasive Carassius species (C. carassius: Ujezdec pond, 49.8991464N, 14.9512814E; C. gibelio: Vrábče, 49.5444106N, 14.6012644E). Fish were maintained in the laboratory in glass tanks (64 × 60 × 34 cm) filled with 130 L of clean, oxygenated water for 1 week at 20 °C under a 12L:12D photoperiod. Fish were fed daily with 1% of their body weight in commercial feed (C-3 Carpe F, Skretting, Stavanger, Norway). Fish species were identified according to their morphological characters (Szczerbowski 2002; Kottelat and Freyhof 2007). The main identification characters were the serration of the last unbranched spine of the dorsal fin, gill raker count, and the shape of the head and fins.

To provide conditions similar to natural habitats in the mesocosm experiment, we collected zooplankton, phytoplankton, and macrophytes (Limnophila aquatica and Ceratophyllum demersum) in water bodies near Vrábče and in the Motovidlo fishpond near České Budějovice (48°92.5521'N, 14°37.5078'E and 48°59.9982'N, 14°22.9186'E). They were kept separately in 500 L outdoor circular vats before being used in the mesocosm experiment. Macrophytes were soaked in sugar water (1 kg/100 L) overnight to remove other organisms (snails, oligochaetes, etc.) and then transferred to clean water for rinsing.

The experiment ran for 84 days from May 20 to August 12, 2022. It was conducted in 18 glass aquaria (64 × 60 × 34 cm), each filled with 130 L of clean, oxygenated water in the laboratory at the Institute of Hydrobiology, Biology Centre of the Czech Academy of Sciences, České Budějovice, Czechia. The aquaria were equipped with an aerator and maintained under a 12L:12D photoperiod. Each aquarium received 10 fish (5 individuals of each species; a total of 90 C. carassius and 90 C. gibelio). Total initial biomass in each aquarium was kept as close to 40 g as possible (mean ± SD = 42.6 ± 3.3 g), although some aquaria deviated from this value (min–max range = 37.7–52.8 g) due to the size of available fish. Initial C. carassius standard lengths were 48.3 ± 7.1 mm, 46.3 ± 6.7 mm, and 47.7 ± 7.6 mm in the low, medium, and high feed treatments, respectively, while lengths of C. gibelio were 50.7 ± 2.9 mm, 49.3 ± 2.0 mm, and 50.3 ± 2.2 mm in the low, medium, and high feed treatments, respectively.

Individual fish length and weight were recorded every 14 days throughout the experiment. Fish identity in aquaria was established by clipping. Aquaria were divided into three groups based on daily food supplementation, calculated as 0.5%, 1%, or 2% of total fish biomass in each aquarium, quantified every 14 days from repeated measurements of all individuals. At the end of the experiment, all surviving fish were measured; C. carassius were released back to their source site, while invasive C. gibelio were overdosed with MS-222.

The mesocosm experiment ran for 104 days from June 8 to September 20, 2022. It was conducted in 24 outdoor mesocosms at the Institute of Hydrobiology, Biology Centre of the Czech Academy of Sciences, České Budějovice, Czechia. Each tank (UV-resistant high-density polyethylene, 98 cm length × 91 cm height × 90 cm width) received 800 L of aged tap water and 7.5 kg of washed river sand (grain size 1–4 mm; SP Bohemia, k.s., České Budějovice). Four weeks before the introduction of fish, macrophytes (wet weight 100 g each of L. aquatica and C. demersum) and 2 L of a well-mixed phytoplankton and zooplankton inoculum were added to each tank. Tanks were randomly divided into three nutrient-level groups based on the amount of added feed pellets (C-3 Carpe F, Skretting, Stavanger, Norway; low: 0 g, medium: 25 g, high: 50 g). The mesocosms were then allowed to stabilize and release nutrients from the pellets.

The manipulated factors included fish assemblage composition (intraspecific vs. interspecific) and nutrient levels (low, medium, and high). The intraspecific treatment (three replicates per nutrient level) received 10 individuals of C. carassius per tank, while the interspecific treatment received 5 C. carassius and 5 C. gibelio per tank (five replicates per nutrient level), i.e., a total of 165 C. carassius and 75 C. gibelio were used in this experiment. Fish identity was established by clipping. The intraspecific treatment of C. gibelio was not included due to the study’s main focus on the impact of invasive C. gibelio on native C. carassius and the limited number of tanks.

Initial total fish biomass in each tank was kept as close to 40 g as possible (mean ± SD [min–max range], intraspecific treatment: 41.2 ± 0.9 g [40.2–43.0 g]; interspecific treatment: total = 39.8 ± 1.8 g [36.4–42.7 g], C. carassius = 20.3 ± 1.2 g [17.6–22.3 g], C. gibelio = 19.5 ± 0.8 g [18.1–21.1 g]). Initial C. carassius standard lengths were 49.6 ± 5.4 mm, 49.7 ± 5.7 mm, and 50.9 ± 4.7 mm in the low, medium, and high experimental treatments and 51.2 ± 4.2 mm, 52.4 ± 4.1 mm, and 47.2 ± 3.9 mm in the control treatment, respectively. Initial C. gibelio standard lengths were 48.2 ± 2.9 mm, 49.5 ± 3.7 mm, and 48.7 ± 2.3 mm in the low, medium, and high nutrient treatments, respectively.

At the end of the experiment, the length of each surviving fish was measured to the nearest 0.1 mm and weighed to the nearest 0.1 g (SI-132-3 scales; Excell, New Taipei City, Taiwan). C. carassius were released back to their source site, while invasive C. gibelio were overdosed with MS-222. Water parameters and zooplankton density were measured at 5-week intervals on four sampling dates during the experiment. Conductivity (µS·cm⁻¹) was measured with a WTW Multi 3430 multimeter (WTW, a Xylem brand, Weilheim, Germany). Chlorophyll-a (relative fluorescence units, RFU) was measured with a YSI Pro DSS multimeter (YSI, a Xylem brand, Yellow Springs, USA). Water transparency (cm) was measured with a Sneller tube (Kolar et al. 2021). Zooplankton was sampled with a tube sampler (diameter 7 cm, length 100 cm) equipped with a one-way valve at the bottom. Four whole-column samples were collected from four regularly spaced locations within each mesocosm to reduce the effects of plankton patchiness and pooled in a 10 L bucket. After mixing, a 3 L subsample was collected and filtered through a 250 µm zooplankton net. Filtered zooplankton was collected in 50 ml vials and preserved with 4% formaldehyde for subsequent counting and identification. Additionally, six data loggers (HOBO Pendant Temperature/Light 64 K Data Logger, Onset Computer Corporation, Bourne, Massachusetts, USA) were placed in the middle of the water column in six random mesocosms to continuously monitor water temperature during the experiment (mean ± SD = 19.5 ± 0.5 °C).

For total phosphorus (TP) determination, water samples were taken from all tanks at four sampling dates: the beginning (day 1), two intermediate measurements, and day 104 at the end of the experiment. Samples were filtered through a 40 µm polyamide sieve to remove zooplankton and/or other coarse particles, followed by acid digestion. TP concentration was then measured using the semi-micro modification method (Kopáček and Hejzlar 1993). We did not measure total nitrogen (TN) because our specific aim was to investigate the direct effects of phosphorus-driven eutrophication on competitive interactions between species, as phosphorus availability is a key limiting factor in many aquatic ecosystems (Sterner and Elser 2003).

All individuals of both fish species survived in the laboratory experiment. Length- and weight-based growth increments of C. carassius were mostly negative in the 0.5% food ration treatment (mean ± SD = -0.24 ± 2.69 mm.month^-1 and -0.46 ± 0.63 g.month^-1), near zero in the 1% food ration treatment (1.03 ± 3.60 mm.month^-1 and -0.22 ± 0.63 g.month^-1), and positive only in the 2% food ration treatment (1.14 ± 2.60 mm.month^-1 and 0.33 ± 0.59 g.month^-1). This contrasted with higher growth increments of C. gibelio, which were around zero only in the 0.5% food ration treatment (0.27 ± 2.46 mm.month^-1 and -0.07 ± 0.30 g.month^-1) and almost always positive in the 1% food ration treatment (2.09 ± 3.19 mm.month^-1 and 0.35 ± 0.60 g.month^-1) and in the 2% food ration treatment (3.07 ± 2.94 mm.month^-1 and 1.43 ± 0.69 g.month^-1).

Model selection identified two plausible models for length-based growth and three plausible models for weight-based growth. Food ration affected growth according to all five of these models. Four models also included species-specific dependence of growth on food ration, with an overall steeper response to increased food levels in C. gibelio than in C. carassius (Fig. 1, Suppl. material 1: tables S1, S3, S4).

Figure 1.

Predictions of length-based (top row) and weight-based (bottom row) growth of C. carassius and C. gibelio in the laboratory experiment, based on the best model. Predictions are illustrated for two initial sizes (standard length and weight) given in panel headings. Symbols = mean estimates for each species; error bars = 95% confidence intervals based on fixed effects. Total initial biomass adjusted at TW_ini = 42.6 g. SL = standard length. The dotted horizontal line in each panel corresponds to zero growth.

The length-based growth was size independent in C. carassius (95% confidence intervals of the size slope estimate always overlapped with b = 1 in both plausible models), while initially larger C. gibelio individuals grew less in length (b < 1; Suppl. material 1: fig. S1, table S3). That is, within each food ration treatment, C. gibelio reached a significantly larger final length than same-sized C. carassius only when their initial length was below ca. 50 mm, while individuals of both species above that size (i.e., the largest individuals of C. gibelio used in the experiment) reached a similar final size (top row in Fig. 1, Suppl. material 1: fig. S1). At each food ration level, weight-based growth declined with initial body size in both species (95% confidence intervals of the size slope estimate d were always < 1), and C. gibelio always reached a larger weight than same-sized C. carassius according to the best model (Fig. 1, Suppl. material 1: fig. S1). Finally, we detected a tendency towards positive, rather than negative, density-dependent growth in length of both species (mean estimates: +0.14 mm per 1 g increase in initial biomass, Suppl. material 1: table S3), but the growth in weight was density independent (mean estimates: -0.01 g per 1 g increase in initial biomass, Suppl. material 1: table S4).

The proportion DD of surviving fish ranged widely between 0.3 and 1.0 (mean ± SD: 0.85 ± 0.20) in individual tanks, and survival differed markedly between both species. While only two out of 75 C. gibelio died, 34 out of 165 C. carassius individuals did not survive until the end of the experiment. The quasibinomial GLM model revealed no statistically significant effects of syntopy (χ²(1) = 4.59, P = 0.27), nutrient levels (χ²(2) = 13.64, P = 0.17), or their interaction (χ²(2) = 0.58, p = 0.93) on C. carassius survival. We did not test the effect of experimental treatments on C. gibelio survival.

Growth increments of C. carassius were near zero in the intraspecific treatment (0.14 ± 1.43 mm.month^-1 and -0.01 ± 0.34 g.month^-1) and mostly negative in the interspecific treatment (mean ± SD across all three nutrient levels: -0.25 ± 1.17 mm.month^-1 and -0.28 ± 0.24 g.month^-1), while the growth increments of C. gibelio in the interspecific treatment (3.39 ± 1.33 mm.month^-1 and 0.99 ± 0.42 g.month^-1) were always positive except for one large individual in the medium nutrient level treatment. In each nutrient level treatment, C. gibelio reached larger final sizes than C. carassius of the same initial size (Suppl. material 1: figs S2, S3).

Model selection identified only one plausible model for both length- and weight-based growth data constrained to the interspecific treatment (Suppl. material 1: table S2). While C. gibelio of the two sizes illustrated in Fig. 2 was always predicted to grow (i.e., the entire 95% confidence interval of the final size was above the initial size), C. carassius of the same sizes was predicted to significantly shrink or not grow (i.e., the entire 95% confidence interval was below or overlapped the initial size). According to these models, both species also responded differently in length and weight growth to increasing nutrient level treatments and initial body size. The differences in predicted final lengths of both species were constant across all three nutrient levels and decreased with initial length (Fig. 2, upper row) due to the significantly shallower slope b in C. gibelio than in C. carassius (Suppl. material 1: table S5). On the other hand, differences in the predicted final weights of C. carassius and C. gibelio did not change with initial weight due to the almost equal slopes d in both species (Suppl. material 1: table S6), but the differences in predictions increased from the low to high nutrient level treatment (Fig. 2, bottom row). We found no evidence for density-dependent growth in length or weight in the data constrained to the interspecific treatment (Suppl. material 1: fig. S2, tables S5, S6).

Figure 2.

Predictions of length-based (top row) and weight-based (bottom row) growth of C. carassius and C. gibelio in the interspecific treatment in the mesocosm experiment, based on the best model. Predictions are illustrated for two initial sizes (standard length and weight) given in panel headings. Symbols = mean estimates for each species; error bars = 95% confidence intervals based on fixed effects. Total initial biomass adjusted at TW_ini = 42.6 g. SL = standard length. The dotted horizontal line in each panel corresponds to zero growth.

With the data constrained to C. carassius in both syntopy treatments, three models were plausible for each measure of growth (Suppl. material 1: table S2). The best models predicted that the growth of C. carassius in length or weight did not vary significantly between nutrient levels within each syntopy treatment, i.e., the 95% confidence intervals of the final size overlapped between treatments (Fig. 3, Suppl. material 1: fig. S3). Moreover, syntopy had a significant effect on the growth of C. carassius in weight but only a limited effect on their growth in length. Individuals of the same initial weight were always predicted to reach significantly larger final weights in the intraspecific than in the interspecific treatment, and the predicted difference was the same for all initial sizes and nutrient treatments (Fig. 3, Suppl. material 1: fig. S3, bottom row). However, the effect of syntopy on length-based growth was more subtle: while individuals with SL_ini = 45 mm were always predicted to grow significantly in the intraspecific treatment, their growth in the interspecific treatment was never significantly different from zero irrespective of the nutrient level. This difference disappeared for larger individuals (SL_ini = 55 mm) due to the shallower size slope estimate b in the intraspecific treatment; these individuals were predicted to significantly shrink in length in the low and medium nutrient level treatments and remain at the initial length in the high nutrient level treatment (Fig. 3 Suppl. material 1: fig. S3, upper row). Finally, the best models identified no trend in final length and a tendency towards larger final weight of surviving fish in tanks with a lower proportion of survivors (Suppl. material 1: fig. S3, tables S5, S6).

Figure 3.

Predictions of length-based (top row) and weight-based (bottom row) growth of C. carassius in the intraspecific and interspecific treatment in the mesocosm experiment, based on the best model. Predictions are illustrated for two initial sizes (standard length and weight) given in panel headings. Symbols = mean estimates for each treatment; error bars = 95% confidence intervals based on fixed effects. SL = standard length. Proportion of surviving individuals adjusted at DD = 0.9. The dotted horizontal line in each panel corresponds to zero growth.

Mesocosm experiment: environmental responses

Total phosphorus (TP) concentrations were higher in the medium (mean ± SD = 62.2 ± 32.6 µg.L^-1) and, especially, high nutrient level treatments (101.7 ± 110.1 µg.L^-1) than in the low nutrient level treatment (61.5 ± 41.6. µg.L^-1) at the start of the experiment, with no clear differences between fish identity treatments until Day 68 (Suppl. material 1: fig. S4). Over time, TP concentrations declined in the interspecific treatment, most notably in the high nutrient level treatment, while they remained relatively stable in the intraspecific treatment, particularly under low nutrient conditions (Suppl. material 1: fig. S4). On Day 68, TP concentrations were higher in the intraspecific than in the interspecific treatment at each nutrient level. Thereafter TP remained stable in most treatments but increased slightly in the intraspecific, low-nutrient-level treatment. The final TP concentrations were similar across all nutrient levels but were higher in the interspecific (mean ± SD = 62.3 ± 1.6 µg.L^-1) than in the intraspecific treatment (35.8 ± 7.9 µg.L^-1). These values were close to the initial values in the low nutrient level treatment (Suppl. material 1: fig. S4). Model selection identified two plausible models describing TP concentrations (Suppl. material 1: tables S7, S8), which revealed treatment-specific temporal changes of TP concentrations in the mesocosm experiment.

Water transparency, conductivity, and chlorophyll-a concentration also changed significantly over time, with differences observed between treatments (RDA: pseudo-F = 31.2, P = 0.001, explained variation = 88.62%, Fig. 4a, Suppl. material 1: table S9). The first RDA axis was associated with conductivity, which showed higher values at the start of the experiment in all treatments. Initially, the mesocosms displayed similar environmental conditions, but both treatments, especially fish identity, influenced chlorophyll-a (mainly along the second axis) and water transparency (along both the second and third axes), while water conductivity decreased over time similarly across all mesocosms (Fig. 4a). At the end of the experiment, water transparency was slightly higher and chlorophyll-a concentrations were lower in mesocosms with intraspecific C. carassius treatments compared to interspecific treatments.

Figure 4.

Redundancy analyses (RDA) diagrams showing interaction between time and treatments in the mesocosm experiment for a. Water transparency, conductivity, and chlorophyll-a and b. Zooplankton community composition (Cladocera and Copepoda). Yellow circles and lines = intraspecific treatments (with only C. carassius), green diamonds and lines = interspecific treatments (with C. carassius and C. gibelio), dotted lines = low nutrient treatment, dashed lines = medium nutrient treatment, and solid lines = high nutrient treatments. Symbols indicate the start of the experiment.

Finally, zooplankton densities and the ratio between Cladocera and Copepoda also changed over time with differences between treatments (RDA: pseudo-F = 1.8, P = 0.019, explained variation = 32.85%, Fig. 4b, Suppl. material 1: table S10), although no clear patterns separating the treatments could be identified due to high initial random variability between replicates. Cladocerans, which showed a gradient along the first RDA axis, initially dominated the zooplankton community, especially in mesocosms with low nutrient levels, but then rapidly declined irrespective of the nutrient levels. Copepods showed a gradient along the second RDA axis but did not show clear temporal trends. By the end of the experiment, zooplankton had nearly disappeared from the mesocosms irrespective of the treatment.

Our experimental results show that invasive C. gibelio grows comparatively faster than native C. carassius under nutrient-rich conditions, demonstrating the ability of the invasive species to better exploit nutrient-enriched environments. Invasive species with lower trophic positions and broad diets can thrive in nutrient-rich environments, often at the expense of native species (Gido and Franssen 2007). Invasive C. gibelio has been shown to grow faster than C. carassius in natural water bodies as well as in mesocosms (Tapkir et al. 2022), and eutrophication may further boost C. gibelio’s competitive abilities.

When invasive species are competitively superior to native species in exploiting resources, the likelihood of coexistence decreases, often leading to the invasive species dominating native species (Leger and Espeland 2010). Exploitative competition between C. gibelio and C. carassius led to a significant weight reduction and increased mortality of C. carassius in the mesocosm experiment. This phenomenon is likely a contributing factor to the local disappearance of C. carassius populations (Šmejkal et al. 2024). Furthermore, understanding how different levels of eutrophication affect C. carassius and its ability to coexist with invasive Carassius species can guide C. carassius conservation efforts (Navodaru et al. 2002; Tarkan et al. 2016).

In our mesocosm experiment, weight increments of C. carassius were higher under intraspecific rather than interspecific competition, suggesting that C. carassius may acquire or allocate more resources toward growth when competing with conspecifics rather than heterospecifics (Fuller 2016). This could be due to differences in intra- and interspecific competitive interactions—in particular, C. gibelio is the stronger competitor, leading to fewer per capita resources for C. carassius in the interspecific treatment. In established relationships between species, intraspecific competition is usually more intense, as individuals compete directly for identical resources, potentially driving individuals to invest more in growth to outcompete peers (Gurevitch et al. 1992; Adler et al. 2018). In contrast, interspecific competition is usually less direct since species often occupy different ecological niches, leading to resource partitioning and reduced direct competition (Simberloff 2005; Adler et al. 2018; Eurich et al. 2018). Because the interaction between invasive and native Carassius species is relatively new, it appears that native C. carassius struggles in interspecific competition, as its weight-based growth increments were always lower in the interspecific than in the intraspecific treatment in our mesocosm experiment. When comparing different nutrient-level treatments in syntopy, C. gibelio showed a stronger growth response to increased food availability than C. carassius, which aligns with the nutrient-driven hypothesis.

When matched for size, C. carassius always had lower weight increments compared to C. gibelio, indicating that C. gibelio dominated resource utilization when both species co-occurred. Similarly, initial body size and food availability determined the growth rates of C. gibelio and C. carassius in the laboratory experiment, and the results were qualitatively identical and quantitatively similar to the patterns observed in the mesocosm experiment. All else being equal, C. carassius had a lower growth rate than C. gibelio, suggesting a species-specific intake rate or food conversion efficiency of the pellets used in the experiment. Finally, the slower growth of larger individuals of both species can be attributed to size-related differences in metabolic rate and energy allocation, since larger individuals may require more energy to maintain their body size, leaving less energy available for growth (Tsoumani et al. 2006; Dmitriew 2011). Growth rate strongly depends on food intake (Jobling and Baardvik 1994) and typically declines in larger individuals in most fish species unless they undergo an ontogenetic diet shift (Ohlberger 2013).

Results from the laboratory experiment should be interpreted with caution, as food accessibility in aquaria is high and competing species have limited opportunities for niche partitioning under such conditions. Both study species have flexible life-history characteristics and can adapt their growth and reproductive investment to current biotic interactions in a given water body (Brönmark and Miner 1992; Holopainen et al. 1997; Emiroğlu et al. 2012). Thus, it may be useful to study their competition under different biotic interactions, such as with varying piscivorous species (de Meo et al. 2023). Future studies incorporating a broader body size range of invasive and native species, representing more natural interactions with multiple age cohorts, could address this issue. Furthermore, varying biomass of the invasive species from the initial to the final phase of invasion could help clarify the mechanisms behind competitive exclusion between the two Carassius species. Finally, an experimental setup with higher structural complexity and larger tanks could also provide better insight into competition.

The mesocosm experiment highlights the significant role of nutrient levels and fish identity in shaping aquatic ecosystem dynamics. Total phosphorus (TP) was initially higher in medium- and high-nutrient treatments, but it declined rapidly in high-nutrient conditions and remained relatively stable in low-nutrient treatments, leading to similar TP levels across treatments by the end of the experiment. Species composition affected chlorophyll-a and transparency, with intraspecific treatments showing higher transparency and lower chlorophyll-a. Zooplankton, initially dominated by cladocerans, declined over time and eventually disappeared, highlighting how nutrient inputs and species interactions shape long-term nutrient cycling and ecosystem function.

Elevated nutrients can provide a favorable environment for the rapid establishment of species inocula, leading to an increased impact of invasion (Byers 2002; González et al. 2010; Gallardo et al. 2016; Preston et al. 2018). Increased nutrient levels, particularly phosphorus and nitrogen, promote rapid growth of phytoplankton and subsequent algal blooms (Smith and Schindler 2009; Dodds and Smith 2016; Rolton et al. 2022). Therefore, eutrophication stimulates primary productivity and enhances the growth of algae and/or submerged macrophytes, leading to easily accessible resources for invasive species with a broad dietary range (David et al. 2017).

Increased nutrient input likely promotes higher fish biomass through increased food availability (Lancaster and Drenner 1990), which can provide a competitive advantage to invasive species over native species (Havel et al. 2015; Fernández-Alías et al. 2022). In the present mesocosm and laboratory study, nutrient-rich conditions and higher food intake appeared to favor the invasive C. gibelio and played a significant role in its interspecific competition with C. carassius. We conclude that invasive C. gibelio grew faster than C. carassius because it benefited from additional plant-based food resources promoted by eutrophication. Previous studies have also shown that C. gibelio exhibits superior resource utilization efficiency compared to native C. carassius, resulting in faster growth rates (Tapkir et al. 2022, 2023), which is consistent with our nutrient manipulation experiments.

Eutrophication also changes the physicochemical characteristics of water bodies, which may favor invasive species that thrive in nutrient-rich environments (Alexander et al. 2017). Recent studies have shown that C. gibelio tends to exploit underused food resources by occupying a lower trophic niche, leading to a competitive advantage over native C. carassius (Özdilek and Jones 2014; Tapkir et al. 2023) by capitalizing on the abundance of easily accessible primary producers. Our study further suggests that chlorophyll-a concentrations were higher and water transparency lower when both C. gibelio and C. carassius were present compared to mesocosms with only C. carassius. Taken together, this indicates that invasive C. gibelio can contribute to the feedback loop of eutrophication, in which increased nutrient availability promotes algal growth and provides more food to C. gibelio (Crivelli 1995; Tarkan et al. 2012b), whose presence further aggravates eutrophication symptoms.

Europe has been invaded by several entities from the Carassius auratus complex, including various mitochondrial lineages associated with the taxa C. auratus, C. gibelio, and C. langsdorfii (Wouters et al. 2012; Kalous et al. 2013; Gu et al. 2022), as well as cytotypes exhibiting different ploidy levels. It is not possible to distinguish these entities without genetic analysis. As they may possess distinct biological characteristics, including gill raker counts (Hensel 1971; Papoušek et al. 2008; Tesfaye et al. 2025), their growth responses to eutrophication may differ accordingly.

The authors have declared that no competing interests exist.

The field sampling and experimental protocols used in this study were performed under the guidelines and permission of the Experimental Animal Welfare Commission under the Ministry of Environment of the Czech Republic (ref. no. CZ 01679). Fish were kept under the permission of the Ministry of Agriculture (ref. no. 4253/2019-MZE-17214).

No use of AI was reported.

The collection of data resulting in this manuscript was supported by the Programme of Regional Cooperation of the Czech Academy of Sciences (R200962201), the regions of South Bohemia and Vysočina, the Zoological and Botanical Garden Plzeň, and the Research Programme Strategy AV21 Water for Life.

S.T., K.T., Y.S., and M.Š. participated in experimental work; M.Š., D.B., V.K., and C.D. designed the study; S.T., M.Š., L.K., K.T., and Y.S. prepared the data for analysis; M.Š., S.T., D.B., V.K., and C.D. conducted the data analysis; and S.T., M.Š., and D.B. wrote the first draft. M.Š., D.B., V.K., L.K., and C.D. edited the manuscript. All authors contributed substantial feedback during manuscript preparation.

Sandip Tapkir https://orcid.org/0000-0003-3456-8406

David Boukal https://orcid.org/0000-0001-8181-7458

Lukáš Kalous https://orcid.org/0000-0001-5518-1505

Kiran Thomas https://orcid.org/0000-0002-4318-1732

Yevdokiia Stepanyshyna https://orcid.org/0009-0005-4682-3779

Vojtech Kolar https://orcid.org/0000-0001-6144-317X

Claire Duchet https://orcid.org/0000-0002-6542-650X

Marek Šmejkal https://orcid.org/0000-0002-7887-6411

Data will be made available on the authors' ResearchGate profile upon publication of this paper.