Introduction

Shallow freshwater hypertrophic lakes and ponds are becoming the most common waterbodies worldwide (Scheffer, 2004; Meerhoff and Jeppesen, 2009). Their increasing fishery exploitation and nutrient loading gradually deteriorate their ecological stability and other ecosystem services and result in highly unpredictable functioning of such ecosystems (Meerhoff & Jeppesen, 2009). Fishponds are shallow artificial water bodies of different sizes (< 1 ha to several hundred ha) that have supported fish rearing activities in central Europe since the Middle Ages. During the centuries, they gradually transformed into ecosystems resembling natural shallow lakes that provided alternative habitats for many aquatic and wetland organisms in the overall agricultural landscape (e.g. Šusta, 1898; Kořínek et al., 1987; Francová et al., 2019a). Early studies of Czech fishponds documented their oligo-mesotrophic state and abundant zooplankton throughout the vegetation season (Šusta, 1888; Frič & Vávra, 1895). During the last century, however, fishery management underwent dramatic changes and fishponds have become over-exploited by intensification of fish production that resulted in an overall shift towards hypertrophy (Pechar, 1995, 2000; Pokorný et al., 2015; Francová et al., 2019a). Although abundant and diverse, fishponds have so far received unproportionally less attention regarding linking of classical and microbial food webs that are fundamental for carbon transfer to the ultimate trophic level (Pechar, 2000; Potužák et al., 2007a; Šimek et al., 2019, 2022).

Until the late 1970s, fishpond seasonal plankton patterns corresponded to the following alternating predictable scenarios for shallow lakes: low-transparency (turbid) state dominated by phytoplankton (primary producers) or high-transparency (clear-water) state (Fott et al., 1980). The latter state was maintained by the filtration activity of large zooplankton (e.g. Daphnia spp.) and/or dominated by macrophytes or filamentous algae (cf. alternative shallow lake equilibria in Scheffer et al., 1993; Vanacker et al., 2016). This period was characterised by a significant increase in fish production, which indicated a very good efficiency of naturally driven production processes in the 1960s (Kořínek, 1967; Potužák et al., 2007a). The efficiency of fishpond productivity then depended mainly on plankton structure. Zooplankton, especially large Daphnia, played a key role in matter and energy transfer from phytoplankton to fish. Thus, the natural fish production relied on the density of rapidly growing large Daphnia species (Kořínek, 1967). In addition, seasonal plankton dynamics were closely related to the style of fishery management (Fott et al., 1980) and the ecosystem was primarily driven by the fish stocking regime (Kořínek et al., 1987).

Since the 1980s, however, intensive fish production practices changed both the plankton structure and overall dynamics of most Czech fishpond ecosystems. More intensive fish breeding, higher fish stock densities, and enormous nutrient loads (either directly from fishery management or indirectly from sewage waters and agricultural catchments, see Vystavna et al., 2017) have resulted in a eutrophic to hypertrophic state of most fishponds (Pechar, 2000; Pechar et al., 2002). More recent results from highly eutrophic and hypertrophic fishponds (Potužák et al., 2007a, b, 2008; Duras et al., 2015) have suggested increasing ecosystem stress. The marked rise in the fish stock and both mineral nutrients and organic matter loads have led to unpredictable seasonal variability and decrease in production efficiency and overall system stability (Baxa et al., 2021). Insufficient fish nutrition and frequent local summer anoxia lead to a drop in production process efficiency (Potužák et al., 2007a, b, 2008) compared to its optimal status in the 1960s/1970s (Kořínek, 1967; Fott et al., 1980; Kořínek et al., 1987). Consequently, the current Czech fishponds have become less cost-efficient, more stochastic, and considerably unstable systems, which resemble the most eutrophic temperate shallow lakes according to Scheffer & van Nes (2007). They represent hypertrophic, turbid ecosystems resembling the ‘high fish predation’ scenario of the revisited Plankton Ecology Group (PEG) model (Fig. 2 in Sommer et al., 2012).

Ptacnik et al. (2010b) hypothesised that interactions of various environmental factors including enhanced turbidity and light-limited high phytoplankton biomass become more important with increasing productivity of freshwaters and can severely bottleneck the carbon transfer efficiency. While strict nutrient limitation by nitrogen (N) and phosphorus (P) are common in oligotrophic lakes, a wide spectrum of biotic and abiotic constraints, such as pH fluctuation, light and periodic nutrient limitations, predation pressure, or oxygen depletion, may intersect at the same time in eutrophic lakes (Maberly et al., 2020; Chorus & Spijkerman, 2021). Such unstable conditions with frequent perturbation of environmental factors should theoretically increase niche dimensionality and eventually lead to short-lived events of enhanced microbial diversity (Ptacnik et al., 2010b). Therefore, major drivers maintaining ecosystem functions may diversify plankton communities in an unpredictable manner as productivity increases (Ptacnik et al., 2010b). On the other hand, available data suggest that hypertrophic freshwaters, in particular the fishponds with densely stocked common carp (Cyprinus carpio Linnaeus, 1758), lean towards an overall loss in diversity in zooplankton, macroinvertebrates, submersed macrophytes, or waterfowl (Musil, 2006; Kloskowski, 2011; Šorf et al., 2015; Francová et al., 2019b). This is likely different for the microbial food webs flourishing in the absence of nutrient limiting conditions under low zooplankton grazing control (e.g. Ptacnik et al., 2010b; Šimek et al., 2019, 2022).

Sommer et al. (2012) emphasised the importance of both microbial food webs and resource stoichiometry for seasonal plankton patterns in the PEG model. Moreover, the revisited PEG model predicts different top-down effects by distinct fish stocks on the plankton dynamics in eutrophic lakes (Sommer et al., 2012). Accordingly, an extreme phytoplankton biomass and considerably reduced biomass of large zooplankton were observed in many hypertrophic fishponds with high carp stock (Pechar et al., 2002; Šimek et al., 2019). On the contrary, in some fishponds densely stocked with carp, large Daphnia spp. were present, but apparently failed to efficiently control the phytoplankton, often resulting in bloom formation and turbid water states (Potužák et al., 2007b). Similarly, Sinistro (2010) reported a weak phytoplankton control by zooplankton under the conditions of high nutrient supply in warm temperate systems. Moreover, particularly in the turbid waters, large Daphnia act as keystone species that could modulate matter and energy transfer from other microbial plankton to fish by filtering nutrient-rich detrital particles occupied by bacteria (e.g. Lind et al., 1992).

In this study, we investigated the structure and seasonal dynamics of plankton in hypertrophic fishponds under the conventional fishery management based on auxiliary feeding. This fishery management typically represents fertilisation with ca. 2 t ha−1 of manure, stocking of market carps up to 800–1000 kg ha−1, and cumulative doses of ca. 2 t ha−1 of cereals applied to main ponds during a vegetation season (for overview of fishery managements, see Francová et al., 2019a). Consequently, low digestibility of phytate from cereals by carp (Roy et al., 2020a) further increases internal phosphorus (P) loading, while manure decomposition contributes to common oxygen depletions (Rutegwa et al., 2019a, b) and possible nitrogen (N) losses by denitrification (Pokorný et al., 1999). Such a stoichiometric imbalance, i.e. low mineral N:P ratio (< 10; Chorus & Spijkerman, 2021), could favour cyanobacterial growth among primary producers (e.g. Arrigo, 2005) and, in turn, any stoichiometric resource imbalance would affect consumers’ growth and/or consumer-driven nutrient regeneration (e.g. Sterner & Elser, 2002).

The study evaluated the synergistic effects of the fish stock (top-down control), the legacy of eutrophication and the importance of stoichiometric imbalance in macronutrient availability (bottom-up control) on the zooplankton structure, functioning of the planktonic food webs (e.g. Sterner & Elser, 2002), and consequences for the functioning of the shallow hypertrophic ecosystems. We hypothesised that (i) the high stock of carps exerts enormous top-down effects on the plankton community disabling the herbivorous zooplankton–fish link, while (ii) the hypertrophic conditions lead to stoichiometric resource imbalance within planktonic food webs. The cascading top-down effect may thus reduce the natural, zooplankton-driven transfer of ample nutrients, as well as fish production efficiency, and support high growth rate and biomass of microbial assemblages relaxed from top-down pressure.

Methods

Sampling and plankton analyses

We selected nine shallow fishponds in South Bohemia (Czechia). All fishponds are polymictic, eutrophic to hypertrophic waterbodies (area 2–228 ha) with a controlled fish stock. Our sampling covered two vegetation seasons from April to September in 2017 and 2018. Four main fishponds were sampled usually monthly in both years, other five fishponds were sampled less frequently (April, June, and August) only in 2017 (cf. Table 1). The fishponds were used for rearing of fish fry (3 nursery ponds: BR—Beranov, RK—Roubíček, ZB—Zběhov) and for ongrowing common carp (Cyprinus carpio) to market size (5 main ponds: DH—Dehtář, KL—Klec, KV—Kvítkovický, PM—Posměch, PT—Potěšil) in 1- or 2-year production cycles (for a review of fishery management types, see Francová et al., 2019a); one fishpond (RD—Rod) is a nature reserve with specific fishery management (for details, see Table S1).

Table 1 Environmental variables and stoichiometric ratios (median and range) during 2 years of the fishpond monitoring
Full size table

Mixed water samples from the surface layer were taken by a van Dorn sampler (length of 1 m, 6.4 l volume) between 7 and 10 a.m. and collected into washed 50 l plastic containers by integrating seven sampling points along a linear transect of the open water zone of each fishpond. Subsamples for plankton and chemical analyses were dispensed from the same mixed water sample. A subsample was filtered through a 200-µm mesh size into a plastic bottle for chemical and seston analyses (ca. 3–5 l) and transported to laboratory within 2 h. Other unfiltered subsamples were taken for analysing of bacterial, heterotrophic nanoflagellate (HNF) and ciliate density in fixed samples and processed as described in Šimek et al. (2019).

Another unfiltered subsample was fixed with Lugol solution for phytoplankton analysis. Samples were counted in Utermöhl’s sedimentation chambers on an inverted microscope (Olympus IMT-2) (Utermöhl, 1958). Cell volumes of phytoplankton species were estimated by approximating the shape to known geometric forms and the results were expressed as fresh biomass (i.e. biovolume in mm3 l−1; Hillebrand et al., 1999). The taxonomic keys by Komárek and Anagnostidis were used for determination of cyanobacteria (Komárek, 1999, 2013; Komárek & Anagnostidis, 2005).

Crustacean zooplankton were concentrated by filtering a total volume of 30 l of the remaining mixed sample through a 200-μm plankton net into another clean container. In the last step, rotifers were concentrated from 5 or 15 l of this filtrate by filtering through a 20-μm mesh, which resulted in the size fraction between 20 and 200 μm. Rotifers were preserved by formaldehyde (4% final concentration). Crustacean samples were gently preserved in a mixed sugar-formaldehyde solution at the final concentration of 4.5% (wt/vol) of formaldehyde. In laboratory, both the planktonic crustaceans and rotifers were identified to the species level whenever possible and counted in subsamples.

Laboratory analyses

Total organic carbon (TOC) and dissolved organic carbon (DOC) concentrations were determined with the FormacsHT Analyzer (Skalar, The Netherlands) using the catalytic combustion of water samples at 950°C and total inorganic carbon (TIC) or dissolved inorganic carbon (DIC) by injecting the samples into a low-temperature reactor containing acid. Then, TOC = TC − TIC and DOC = DC − DIC. Samples for the determination of DOC were filtered through Whatman GF/C filters (nominal porosity of 1.2 µm). The TOC was measured in samples that have passed through 200-µm sieves. Particulate organic carbon (POC) was calculated by subtraction of DOC from TOC.

The NH4–N, NO3–N and dissolved reactive phosphorus (DRP) were determined using a flow injection analyser (FIA Star 5010, Tecator, Sweden; Růžička & Hansen, 1988) in samples filtered through Whatman GF/C filters. The alkalisation of sample by 0.1 M NaOH and gas diffusion method was used for the determination of NH4–N (Tryzell & Karlberg, 1995). Nitrate (NO3–N) was analysed as nitrite after reduction of the sample in a Cd–Cu column. The reaction of nitrite with sulphanilamide and N-(l-naphthyl)-ethylenediamine was used for determination of NO2–N. The standard phosphomolybdenum complex method was used for the determination of DRP (Parsons et al., 1984).

Persulphate oxidation method for the simultaneous determination of total nitrogen (TN) and total phosphorus (TP) (Valderrama, 1981) converting them to NO3–N and DRP, respectively, was applied to both unfiltered and Whatman GF/C filtered samples. The TN, TP, dissolved N (DN), and dissolved P (DP) were then analysed as NO3–N and DRP, respectively, using the flow injection analyser. Concentrations of particulate N (PN) and particulate P (PP) were calculated as: PN = TN − DN and PP = TP − DP. Respective molar POC, PN, and PP concentrations were used to calculate the elemental ratios in seston.

Concentration of chlorophyll a was determined spectrophotometrically at 664 nm after the extraction of samples collected on the GF/C filters with a hot mixture of 90% acetone:methanol (5:1 by volume; Pechar, 1987).

Fish stock estimate

To assess the potential effect of overall fish predation, we calculated the fish stock index (FSI), which integrates both the biomass and density of fishes (Musil, 2016), as a proxy of fish top-down pressure:

$$\text{FSI} = \left(\text{fish biomass}\right) \times \sqrt{\text{fish density}},$$

where fish biomass (in 102 kg ha−1) and fish density (ind. ha−1) were interpolated for each sampling day from stocking and harvest data provided by commercial fisheries using a specific growth rate (SGR) for each fishpond in the given year (for details, see Eqs. 1–3 in Supplementary Information).

The stocked fish FSI (denoted as FSIS) was determined for the dominant fish species, common carp, which accounted for 82–98% of harvested fish biomass in most fishponds. The exceptions were RD in 2018, the nature reserve with atypical experimental stocking of tench as the dominant fish, and ZB where only pike was stocked in 2017 (for details, see Table S1). In addition, non-intentionally stocked planktivorous fish (hereafter referred to as undesirable fish) were sampled in 2018. Fish estimates of their density and biomass were used to determine undesirable fish FSI (FSIU; see Supplementary Information).

Statistical analyses

Proposed hypotheses were tested using the multivariate approach. The top-down effect was represented by the FSI estimate, while bottom-up effects by environmental variables, mainly various forms of nutrients. The redundancy analysis (RDA) was used to determine the relationships among main plankton groups (bacteria, HNF, ciliates, rotifers, copepods, cladocerans—adult and juvenile Daphnia and other Cladocera, and chlorophyll a concentration as a proxy of phytoplankton biomass) and environmental variables in both years. Explanatory variables were preselected based on Spearman rank order correlations (significant correlations above 0.4 were excluded; see Table S3). The final selection was done using the forward selection procedure revealing factors significantly affecting plankton. Explanatory variables were transformed by log (x + 1). Fishpond identity (fishpond code) was used as a covariate. Sampling dates of each fishpond were ordered in time series, and fishponds were not allowed to permute.

In addition, four fishponds sampled monthly in 2018 were evaluated by RDA including main zooplankton groups and both FSIS and FSIU. Samples from April were not included due to missing estimates of undesirable fish. Analysis details (data transformation, permutation design) were the same as the above-described RDA. The RDA analysis was done using Canoco 5.12 (ter Braak & Šmilauer 2012), Spearman correlations using Statistica 13 (TIBCO Software Inc., 2018).

Results

Trophic state, plankton, and nutrient availability

Our results, resulting from a total of 54 samplings, covered four fishponds monitored in April–September of 2017 and 2018 and another five fishponds monitored only during 2017 (Table 1). All studied fishponds were classified as eutrophic to hypertrophic freshwaters with high total nutrient concentrations (mean TN: 3.2 mg l−1; mean TP: 0.26 mg l−1), resulting in high phytoplankton biomass (average chlorophyll a concentration: 137 µg l−1) and low transparency (mean Secchi depth: 47 cm; Tables 1 and 2). Overall high trophic states and DOC concentrations (mean: 18.0 mg l−1) resulted in high densities of bacteria (mean: 18.3 × 106 cells ml−1), HNFs (mean: 9.5 × 103 cells ml−1), and ciliates (mean: 263 cells ml−1; Table 2).

Table 2 The overview of plankton organisms and fish stock of the studied fishponds over 2 years of the monitoring (median and range)
Full size table

Despite the overall hypertrophic state of the studied fishponds, we detected rather low DRP concentrations (median DRP: 11 µg l−1) indicating a moderate P deficiency for phytoplankton in a half of sampling dates. However, ambient DP concentrations suggested possibly higher bioavailable P (median DP: 112 µg l−1). On the other hand, surprisingly low concentrations of dissolved inorganic N (DIN = NH4–N + NO3–N; median DIN: 29 µg l−1) were found in three quarter of the samples (Table 1) indicating strong N limitation for primary producers. Thus, the DIN:DP ratio ranged from 0.01 to 30 (median: 0.50; Table 1). Although most DIN:DP values were below the optimum ratio (~ 10) of mineral nutrients for autotrophic growth, the seston N:P ratio did not indicate N deficiency (median of 22 was above the Redfield ratio of N:P = 16; Fig. 1). Seston stoichiometry revealed high variability both on spatial and temporal scales. For instance, the seston C:P ratio (mean: 260; median: 208) varied between 71 and 940 across all fishponds and samplings (Table 1). High seston N:P values often corresponded with high proportion of cyanobacteria in the summer maxima of phytoplankton biomass (Fig. 1). These cyanobacterial blooms were usually dominated by N-fixing species from the order Nostocales (Anabaenopsis spp. V. V. Miller 1923, Aphanisomenon spp. Morren ex Bornet & Flahault 1888, Dolichospermum spp. (Ralfs ex Bornet & Flahault) P. Wacklin, L. Hoffmann & J. Komárek 2009, and Sphaerospermopsis reniformis (Lemmermann 1898) Zapomělová et al., 2010), sometimes in co-occurrence with non-fixing Chlorococcales (Microcystis spp. Kützing 1833) or Oscillatoriales (Planktothrix agardhii (Gomont 1892) Anagnostidis & Komárek 1988).

Fig. 1
figure 1

Nutrient stoichiometry in fishponds: A Element stoichiometry of dissolved nutrients in fishpond water (DIN:DP) and in seston (N:P); the vertical solid line indicates the optimum for primary producers (DIN:DP = 10), the horizontal dashed line stands for the Redfield ratio (seston N:P = 16). B Potential N and P limitations of the phytoplankton at DIN:TP < 2.0 and > 5.1, respectively; the grey bar indicates a transient area (according to Ptacnik et al., 2010a). Solid symbols indicate the samples with a high proportion of cyanobacteria (> 40% of phytoplankton biovolume)

Full size image

The most common species out of 39 rotifer taxa in total were omnivorous Keratella cochlearis (Gosse, 1851) and Polyarthra dolichoptera Idelson, 1925, which occurred in more than 40 samples. Other common rotifer taxa (found in > 20 samples) were Anuraeopsis fissa (Gosse, 1851), Asplanchna spp., Brachionus angularis Gosse, 1851, Conochilus/Conochiloides spp., Keratella quadrata (Müller, 1786), Pompholyx sulcata Hudson, 1885, and Trichocerca pusilla (Jennings, 1903). Annual median of rotifer densities varied between 269 ind. l−1 in RD (2017) and 13,373 ind. l−1 in KV (2018; Table 2). Cladocerans comprised of 17 taxa, from which Bosmina longirostris (O. F. Müller, 1776), Daphnia gr. longispina, and Chydorus sphaericus (O. F. Müller, 1776) were present in majority of samples. The genus Daphnia, represented by four taxa, was found in most fishponds and sampling dates (49 out of 54 samples) in densities up to 630 ind. l−1 (mean: 113 ind. l−1, including juveniles). Out of six copepod species, the most common were Acanthocyclops trajani Mirabdullayev & Defaye, 2004, Cyclops vicinus Uljanin, 1875, and Eudiaptomus gracilis (G. O. Sars, 1863). Copepods were mainly represented by their juvenile stages, nauplii and copepodites (Table 2).

Fish stock

Fish stock in the studied fishponds was mostly composed of common carp in wide range of sizes from carp fry through 2-year-old carp to market size fish, or pikeperch and tench, and undesirable fish (for more details, see Table S1). Common carp were reared to market size in most of the fishponds (DH, KL, KV, PM, PT), and two fishponds (BR, RK) were used for rearing carp fry (BR, RK). One fishpond (ZB) was stocked exclusively with pike fry. The nature reserve (RD) was stocked with low biomass of different fishes: with common carp in 2017, but with pikeperch and tench in 2018 (Table S1). In some fishponds, mainly in KV and RD, the presence of topmouth gudgeon [Pseudorasbora parva (Temminck & Schlegel, 1846)] further considerably increased grazing pressure and thus also modified the trophic structure of studied ecosystems. Biomass of undesirable fish was not considered in 2017 but estimated in 2018.

In general, the FSIS values in fishponds stocked predominantly with carp were much higher in nursery fishponds (mean: 2,191) compared to main fishponds with fish reared to market size (mean: 146). Fish farming covered a 1-year cycle of fish production with the exceptions of DH and PM, where a 2-year cycle was used. Therefore, FSIS values were very similar in annually stocked KL (mean: 103 and 113) and KV (mean: 135 and 112) but differed in DH in either year (mean: 155 and 243, respectively, for 1-year-old fish in the first year and 2-year-old fish in the second year of the production cycle). The FSIS values in RD also differed between years, reflecting the different fishes and density ( Table 2).

Plankton structuring factors

Multivariate analysis revealed TN (fraction < 200 µm) and DIN concentrations to be the most important variables explaining plankton composition in fishponds in both years studied (Fig. 2, Table 3). The TN explained 22.1% of variability (pseudo-F = 12.5, P = 0.006), DIN added 8.9% (pseudo-F = 5.5, P = 0.003) to the total variability explained by the RDA model. Although both parameters are part of N cycle, they are not significantly correlated to each other (Table S3). Chlorophyll a concentration, bacteria, HNFs, ciliates, rotifers, and copepodites and adults of cyclopoids were positively associated with the TN concentration corresponding to overall plankton production. Total N explained 22.1% of the variability when acted both separately and alongside all other variables (Table 3). Crustacean zooplankton except for cyclopoids occupied the left side of the ordination biplot (Fig. 2).

Fig. 2
figure 2

Redundancy analysis summarising relationships among main plankton groups and environmental variables selected based on their significant contribution by forward selection procedure explains 27.8% of total variability (for detailed statistics, see Table 3). Data based on sampling of all fishponds in 2017 and 2018; TN total nitrogen (in seston < 200 µm), DIN dissolved inorganic nitrogen, HNF heterotrophic nanoflagellates; Cyclopoida consists of adult and copepodites, nauplii are not included

Full size image
Table 3 Summary of simple and conditional effects based on RDA
Full size table

A pronounced seasonal impact of fish stock on zooplankton was remarkable in the four fishponds sampled in 2018 (Fig. 3). The FSI as a proxy for fish predation intensity explained 39.1% of variability within main zooplankton groups (test on all axes: pseudo-F = 6.1, P = 0.001). While FSIU was not significant, FSIS itself explained 38.4% (pseudo-F = 9.3, P = 0.004; Fig. 3). Generally, FSIS divided biplot in the similar way as TN in the above-described analysis. The positive relationship of rotifers and cyclopoids with FSIS on the right side of the ordination biplot contrasted with cladocerans and calanoids occupying the left side. The closely correlated adults and juveniles of Daphnia were more affected by fish predation than other cladocerans (Fig. 3).

Fig. 3
figure 3

Redundancy analysis summarising the relationships among zooplankton groups and stocked fish selected based on their significant contribution by forward selection procedure (pseudo-F = 9.3, P = 0.004, 38.4% of explained variability). Data based on the sampling of four fishpond in 2018; Cyclopoida consists of adult and copepodites, nauplii are not included; FSIS fish stock index of stocked fishes (see text for details)

Full size image

Seasonal pattern

In 2018, we sampled four selected fishponds more frequently and estimated undesirable fish density and biomass. The seasonal patterns of FSIU reflected a continuous slight increase in populations of ruffe [Gymnocephalus cernua (Linnaeus, 1758)], perch (Perca fluviatilis Linnaeus, 1758), and roach [Rutilus rutilus (Linnaeus, 1758)] in KL. In DH, however, undesirable fishes, particularly perch and common rudd [Scardinius erythrophthalmus (Linnaeus, 1758)], peaked in July and slightly decreased towards September. On the other hand, abundant topmouth gudgeon remarkably boosted in KV and RD, where FSIU values were about tenfold higher than in DH and KL in late summer (Table S3). The FSIS increased during the season in DH, KV, and KL but remained relatively stable in RD where fish biomass was low during the whole year (Fig. 4). The half of the carp stock (in density) was caught off from KV in July and ca. one third of the carp stock was caught off from KL in June; yet faster growth of remaining carps clearly compensated for the losses. Therefore, the modelled biomass and corresponding FSIS still increased (Table S2).

Fig. 4
figure 4

Seasonal patterns of densities of zooplankton main groups (fraction > 200 µm) and fish stock estimates in 2018. Fish are expressed as fish stock indices of stocked (FSIs) and undesirable (FSIU) fish; fishpond codes: DH Dehtář, RD Rod, KV Kvítkovický, KL Klec

Full size image

Adult copepods were present in similar densities in all four fishponds throughout the season while copepodites did so mainly at the beginning of the season (April) and then at the end (August, September). Daphnia density in the sampled fishponds was generally higher in spring (May) but remained low during the whole season. In the second year of fish production cycle in DH, daphnids were found in relatively high number compared to other fishponds. Smaller-sized cladocerans were observed throughout the season in DH and at the end of season in KV, their presence in other fishponds was diminutive (Fig. 4). Annual mean seston C:P ratio (162) in KV likely supported higher zooplankton density compared to nearly twofold annual mean C:P ratio in KL (305; cf. values of respective 2018 parameters in Tables 1 and 2).

The scatterplot of main zooplankton groups revealed a seasonal development ordered by sampling dates in all fishponds except for KV, where an increase in small cladocerans (B. longirostris) at the end of the season changed the position of August and September samples within the ordination biplot (Fig. S1).

Discussion

Nitrogen and oxygen depletions

Our results indicated a stoichiometric imbalance of available macronutrients influencing hypertrophic fishponds. Multivariate analysis revealed that all main plankton groups were significantly correlated with TN, which explained 22% of plankton variability. As conditional effects, TN and DIN concentrations were significant explanatory variables (Table 3, Fig. 3) indicating that both primary production and plankton communities of the studied fishponds could be frequently limited by N. Another evidence of N deficiency was indicated by the low DIN:TP ratio in water (Fig. 1B), which was previously recommended as the best indicator for inferring in situ N vs. P limitation from chemical data (DIN:TP < 2.0 vs. > 5.1, respectively; Ptacnik et al., 2010a). Not surprisingly, the maximum summer phytoplankton biomasses were usually dominated by Nostocales cyanobacteria (Fig. 1B) that, besides many bacteria, could compensate for N deficits owing to N2 fixation (Smith, 1983; Hillebrand et al., 2013). These nitrogen-fixers thus may have ultimately increased the seston N:P ratio (Fig. 1A; cf. Arrigo, 2005 and references therein; Quigg et al., 2011; Hillebrand et al., 2013; Xiao et al., 2022). These results primarily confirm our hypothesis that the hypertrophic conditions forced stoichiometric imbalance within the planktonic food webs towards N deficiency.

Several recent studies well documented both great spatial and temporal variability in dissolved oxygen concentrations in the studied fishponds, as well as their overall tendency to heterotrophy and oxygen deficits (Rutegwa et al., 2019a, b) and diurnal and seasonal variability (Baxa et al., 2021). In general, both excessive nutrient loading and high fish stocking resulted in the high phytoplankton and microbial biomass in the studied fishponds. Consequently, phytoplankton photosynthesis and microbial respiration led to pronounced diurnal oxygen fluctuations (Baxa et al., 2021), or even its ultimate depletion and consequent nitrogen losses via denitrification (Pokorný et al., 1999; Rutegwa et al., 2019b; Wu et al., 2020). In 76% cases, DIN concentrations were well below a threshold for the N limitation of phytoplankton (< 100–130 µg l−1; Chorus & Spijkerman, 2021).

Moreover, the oxygen depletions, frequently recorded above the bottom sediment (data not shown; for more details and examples from BV, ZB, RK, DH, KV, and PM, see Rutegwa et al., 2019a, b; or from RD, see Baxa et al., 2021), likely resulted in internal P loading from the bottom. The P-rich sediment represents a characteristic nutrient (P) legacy that may drive these hypertrophic ecosystems to a strong stoichiometric imbalance of resources under frequent hypoxic episodes (Baxa et al., 2021), typically with frequent cyanobacterial blooms (Potužák et al., 2007b). Similarly, Zhang et al. (2018) highlighted that eutrophication process could fuel potential denitrification rate in shallow hypertrophic lakes.

Nutrients and stoichiometric imbalance

The molar DIN:DP ratio, a proxy of mineral nutrient availability (e.g. Conde-Porcuna et al., 2002), was extremely low (mean: 3.81; median: 0.5) indicating a strong tendency to N limitation among primary producers, as the theoretical optimum of DIN:DP should lay around 10 (Arrigo, 2005; Chorus & Spijkerman, 2021). The DIN:TP ratio, the best index of N vs. P limitation (Ptacnik et al., 2010a), suggested the same N limitation (mean: 2.27; median: 0.26; Fig. 1B). Under such circumstances, cyanobacteria usually dominated summer phytoplankton (> 40% biomass in July–September, see solid symbols in Fig. 1). Accordingly, the seston N:P ratio usually laid well above the Redfield ratio under such circumstances (Fig. 1A). Such a high N:P ratio is indeed characteristic for green algae and N-fixing cyanobacteria (~ 27 and > 40, respectively; Arrigo, 2005; Quigg et al., 2011). Nutrient enrichment experiments in shallow eutrophic lakes (Kolzau et al., 2014) also documented predictable, frequent, and persistent N limitation of natural phytoplankton below the molar DIN:DRP ratio of 22 (at DIN < 100 µg l−1). Chorus & Spijkerman (2021) recently highlighted the Reynold’s concept of carrying capacity: “the maximum possible plankton biomass that a given waterbody can attain depends on the total concentration of the limiting nutrient and the minimum intracellular amount with which a cell can function, i.e. the cell quota. … In temperate climates the resource limiting carrying capacity typically shifts from light in winter to TP in spring and often throughout summer, potentially interrupted by phases of N limitation during summer (Fig. 1)”. While some species have a competitive advantage at low concentrations of P, others at low concentrations of N (Hillebrand et al., 2013; Chorus & Spijkerman, 2021) as clearly illustrated summer cyanobacterial blooms in the studied hypertrophic fishponds. Moreover, Xiao et al. (2022) identified critical knowledge gaps in cyanobacterial adaptations to low nutrient concentrations.

On the other hand, despite overall low DRP concentrations, we observed only few cases when extremely high seston C:P ratios (> 400) indicated a potential P deficiency for consumers (Table 1). Yet only in three cases, these high seston C:P values coincided with the total absence of adult Daphnia and/or Eudiaptomus (data not shown). Moreover, filter-feeding Daphnia could utilise abundant P-rich detrital particles occupied by bacteria in the turbid fishponds (cf. Lind et al., 1992).

On the other hand, according to a recent study (Strandberg et al., 2022), crustacean zooplankton likely might be further stressed, besides the macronutrients’ imbalance, by an adverse stoichiometry of polyunsaturated fatty acids in the phytoplankton under summer hypertrophic conditions.

Planktonic food webs

Herbivorous zooplankton, especially large daphnids, are critical in mediating both energy and matter transfer through the planktonic food web (Carney, 1990). Data from the long-term fishponds monitoring confirmed that the strength of the food web interaction was higher at moderate eutrophic conditions (Potužák et al., 2007a). In the absence of keystone Daphnia species, the high trophic state, high primary production, and excessive DOC supported complex microbial food webs, extreme densities of bacteria, ciliates, HNF, and even rotifers in the fishpond plankton (Table 2). Notably, the numbers of bacterivorous flagellates and omnivorous ciliates approached the levels typical for moderately loaded wastewater treatment plants (Madoni, 2011). Consequently, the missing link of large crustacean zooplankton in a food chain diminished the efficiency of the natural fish production (Potužák et al., 2007a). This ‘dead end’ in carbon flux (i.e. uncoupling the microbial and classical food webs) resulted in both accumulation of microbial biomass and frequent anoxia. Intensive protistan grazing, with the dominance of omnivorous ciliates and fast bacterial turnover in hours to days, has been described in detail elsewhere (Šimek et al., 2019, 2022). In addition, abundant omnivorous rotifers were relaxed from cladoceran competition (MacIsaac & Gilbert, 1989). Rotifer density followed microbial dynamics, i.e. the inverse seasonal pattern to crustacean zooplankton in this study. The seasonal pattern of plankton succession in the studied fishponds generally matched the ‘high fish predation’ scenarios of the revised PEG model for phytoplankton, heterotrophic protists, and metazoan zooplankton in eutrophic waters (Sommer et al., 2012; Cremona et al., 2014). Overall plankton composition of Czech fishponds resembled other temperate shallow and eutrophic ecosystems, such as Estonian Lake Võrtsjärv (Cremona et al., 2014) or French fishponds (Vanacker et al., 2016).

Top-down pressure

Although we cannot fully approve our hypothesis regarding direct top-down effect of high fish stock on the plankton community which was not significant, the FSIS itself explained 38% of zooplankton variability in main ponds in 2018 (Fig. 3). Surprisingly, the high fish stock of market carp (associated with high FSIS) did not always inevitably suppress large zooplankton in main ponds, whereas a high density of smaller fishes usually did eliminate not only large Daphnia species, but also almost all crustaceans in some cases (cf. ZB or RD in Table 2). The high FSI values reflected the heavy top-down pressure caused by small but abundant planktivorous fish, either carp fry or undesirable fishes, namely topmouth gudgeon, resulting in the plankton community dominated by small cladoceran species, developmental stages of copepods, and rotifers. Such a visually orienting planktivorous fish could efficiently reduce population growth of large daphnids by eliminating females bearing advanced embryos (Zemanová et al., 2020). The adverse effects of topmouth gudgeon on zooplankton structure and carp production were indeed documented by Musil et al. (2014), Šindler (2017), and Kajgrová et al. (2022). Our preliminary results also confirmed that cladocerans made > 60% up to 100% of ingested food of 70–90% of gut contents of the topmouth gudgeon specimens inspected (L. Všetičková, unpublished results).

Fish stock and fishery management

Common carp, the dominant fish in our study, is a food opportunist that switches among available food sources; moreover, its dietary preferences change during ontogenetic development (Sibbing et al., 1986; Rahman et al., 2010). Carp is planktivorous size-selective feeder, but its ability to retain small food particles declines with age (Sibbing, 1988). Carp fry stocked at high density may exert stronger top-down effect on large-bodied Daphnia populations in nursery ponds (BR, RK—data not shown) compared to market carp foraging on zooplankton in main ponds (Fig. 4). Large market carps may act as ecosystem engineers that disturb bottom sediments and control macrophyte vegetation (Adámek & Maršálek, 2013; Francová et al., 2019a, b).

Fertilisation with manure and supplemental feeding with cereals have been common practices by market carp production in Czech fishponds (Pokorný et al., 2015; Rutegwa et al., 2019a, b; Roy et al., 2020a). During the period covered by the present study, cumulative doses of 1–1.5 t ha−1 of manure and 1–2 t ha−1 of cereals were applied to main ponds during a vegetation season (except for the nature reserve RD). Considering low digestibility of phytate P from cereals by carp (Roy et al., 2020a), we may expect remarkable eutrophication potential of the supplemental feed (Roy et al., 2020b). Breeding of semiwild mallards (with auxiliary feeding of cereals at KV and PM; Table S1) could be an additional eutrophication source. Moreover, abundant waterfowl populations foraging on macrophytes and macrozoobenthos could be responsible for additional nutrient input.

Total nutrients (TN, TP), water transparency, and phytoplankton biomass (chlorophyll a) in the fishponds apparently remained pretty much the same since the early 1990 (Pechar, 2000; Pechar et al., 2002), while summer phytoplankton biomass became increasingly dominated by cyanobacteria (Fig. 1; Potužák et al., 2007b). The overall fish production in studied ecosystems levelled in the past three decades as well (Table 4; Pokorný et al., 2015). The current annual doses of manure and cereals obviously have driven the studied ecosystems to intensive microbial growth and high respiration not only in bottom sediments but also within the water column (Cremona et al., 2014; Šimek et al., 2019, 2022; Baxa et al., 2021). These auxiliar nutrients apparently did not improve the desired energy and matter transfer to the ultimate fish consumers (Potužák et al., 2007a). Such a situation likely supported enhanced N losses via denitrification (Pokorný et al., 1999) and could maintain seasonal N limitation as Wu et al. (2020) recently described for a hypertrophic shallow lake. Thus, excessive nutrient loading (Maberly et al., 2020) and/or distinct redox-dependent P and N retention in fishpond subhabitats (Sterner, 2008) could result in a strong N limitation typical globally for wetlands or eutrophic lakes (Elser et al., 2007).

Table 4 Total production of market fish in Czechia during 1950–2020 (modified from Pokorný et al., 2015 and http://www.cz-ryby.cz/produkce-ryb/produkce-a-trh-ryb)
Full size table

Therefore, the overall plankton N deficiency observed in this study, leading to enhanced proportion of N-fixing cyanobacteria in the phytoplankton, seems to be an inevitable consequence of current fishery management (also see Ivanova et al., 2022). Perhaps surprisingly or counterintuitively, the manure application does not improve N availability for natural fish production but, on the contrary, increases the N losses from the fishponds via enhanced denitrification rates (Pokorný et al., 1999). On the other hand, intense nutrient regeneration promoting phytoplankton growth could support crustacean zooplankton, even large daphnids, important for natural carp nutrition (Roy et al., 2020a, b). Yet survival and reproduction of daphnid populations are only possible under conditions of moderate top-down pressure of planktivorous fish (Zemanová et al., 2020). Detailed population growth of Daphnia spp., adverse effects of undesirable fish on both plankton structure and carp production, and an optimised carp nutrition for semi-intensive aquaculture in hypertrophic fishponds deserve further thorough investigations, yielding considerable revision of major steps of the current fishery management.