Skip to main content
NCBI home page
Ecology and Evolution logoLink to Ecology and Evolution
. 2023 Sep 25;13(9):e10567. doi: 10.1002/ece3.10567

Top‐down effects of filter‐feeding fish and bivalves moderate bottom‐up effects of nutrients on phytoplankton in subtropical shallow lakes: An outdoor mesocosm study

You Zhang 1,2, Ruijie Shen 1, Kuanyi Li 1,3, Qisheng Li 1,4, Huihui Chen 1, Hu He 1, Xiaohong Gu 1,5,✉, Zhigang Mao 1,✉, Richard K Johnson 2
PMCID: PMC10518750  PMID: 37753309

Abstract

Biomanipulation has been widely used in the ecological restoration of eutrophic lakes for decades. However, biomanipulation is prone to failure if external nutrient loads are not reduced. In order to explore the importance of filter‐feeding fish and bivalves on algal control, an outdoor mesocosm experiment was conducted using different nutrient concentrations. Four treatments simulating daily loads of nutrients in Lake Taihu were studied: current, two times, and three times average daily loads of nutrients with both fish (Aristichthys nobilis) and Asian clam (Corbicula fluminea) and as a control current daily loads without fish or bivalves. Results showed that stocking of filter‐feeding fish and bivalves (80 g m−3 bighead carp; 200 g cm−2 clams) at two times daily nutrient loads could effectively control water column Chl a concentrations and phytoplankton biomass. At higher nutrient concentrations (TN ≥ 260 μg L−1 d−1; TP ≥ 10 μg L−1 d−1), top‐down control of filter‐feeding fish and bivalves was less effective and bottom‐up effects resulted in significant increases of Chl a concentration. Thus, as phytoplankton biomass in freshwater ecosystems is determined by both the top‐down effects of predators and the bottom‐up effects of nutrients, external loadings should be controlled when filter‐feeding fish and bivalves are used for algal control to ensure the efficacy of biomanipulation.

Keywords: algae control, biomanipulation, eutrophication, food web, lake restoration

  1. A combination of filter‐feeding fish and clams suppressed phytoplankton, which could not be affected by low‐level nutrients.

  2. Bottom‐up effects at high‐level nutrients on phytoplankton overcome top‐down effects, indicating that nutrient levels should be controlled to optimize the effect of the intervention.

graphic file with name ECE3-13-e10567-g001.jpg

1. INTRODUCTION

Successful restoration of shallow eutrophic lakes is often marked by a shift from a turbid status dominated by phytoplankton to a clear water status dominated by submerged macrophytes (Jeppesen et al., 2005; Liu et al., 2018). However, there remains a competitive relationship between submerged macrophytes and phytoplankton (Lürig et al., 2021; Scheffer et al., 2001; Søndergaard et al., 2017). Submerged macrophytes can maintain a lake's clear water state by removing excess nutrients, inhibiting sediment resuspension, and reducing the biomass of phytoplankton (Han et al., 2022; Jeppesen et al., 2012; Lürig et al., 2021). Phytoplankton are prone to be limited by nutrients in the water column but have a clear competitive advantage in terms of light, while submerged macrophytes are able to obtain nutrients from the sediment, where the nutrient content is generally much higher than the water column, but are susceptible to light limitation (Chen et al., 2020; Han et al., 2022). Although submerged macrophytes are often the main primary producers in clear water lakes, the competitive advantage of phytoplankton increases with increased nutrient concentrations, resulting in lower water transparency that limits growth and even survival of submerged macrophytes, and which eventually might result in a shift back to a turbid water status (Jeppesen et al., 2012; Jin, van Leeuwen, Temmink, & Bakker, 2022; Liu et al., 2020; Zhang et al., 2022). Thus, a primary aim of biomanipulation in lake restoration is to improve light conditions for submerged macrophytes by reducing phytoplankton biomass and suspended sediments (Yu et al., 2016).

Phytoplankton biomass in freshwater ecosystems is determined by both top‐down effects of predators and bottom‐up effects of nutrients (He et al., 2021; Karpowicz et al., 2023). Traditional biomanipulation methods developed for northern temperate lakes attempt to enhance zooplankton grazing on phytoplankton by introducing piscivorous fish or by removing plankti‐benthivorous fish (Hilt et al., 2018; Jeppesen et al., 2012; Jin, van Leeuwen, Van de Waal, & Bakker, 2022). Benndorf et al. (2002) concluded that fish manipulation for top‐down control of phytoplankton was more likely to succeed in mesotrophic or slightly eutrophic lakes than in eutrophic or hypertrophic lakes. Jeppesen et al. (2012) argued that water column phosphorus concentrations should be kept less than 100 μg L−1, or the external phosphorus loads should be reduced to 0.5–2.0 g m−2 a−1 to achieve stability and successful biomanipulations. However, the higher primary productivity and more complex food webs found in lakes in tropical and subtropical regions is expected to further weaken the top‐down effect of traditional biomanipulations (Jeppesen et al., 2012). Liu et al. (2018) found strong bottom‐up but weak top‐down effects in a successful restoration of a tropical shallow eutrophic lake, indicating the importance of bottom‐up effects through reduced sediment resuspension and nutrient release, rather than the top‐down effects observed in many temperate lakes.

Lake Taihu is a typical subtropical large eutrophic shallow lake. Large‐scale basin‐wide management interventions were implemented following the 2007 water crisis (Zhou et al., 2021); however, 10 years of monitoring data (2008–2018) has shown that average annual external nutrient inputs (TN: 40–50,000 t; TP: 2000 t) have not been significantly reduced (Cheng et al., 2020; Qin, 2020). A previous study showed that the combined use of two filter‐feeding organisms with different feeding niches, that is, bighead carp and Asian clams, was effective in reducing the total phytoplankton biomass and improving water clarity (Shen et al., 2020). The efficacy of this approach was attributed to strong complementarity in terms of algae size that can be filtered, and the aggregation of small particles by bivalves, increasing the availability of food resource for filter‐feeding fish (de Rezende Ayroza et al., 2019; Shen et al., 2020). However, the stability of this biomanipulation intervention on nutrient loads and algal control is still unclear, especially when external nutrient inputs are not reduced (Parr et al., 2020).

Building on Shen et al. (2020) in this study, we focus on the interactions of filter‐feeding fish and bivalves and different nutrient concentrations to explore if nutrient thresholds exist for this form of biomanipulation. We hypothesized that when external nutrients exceed a certain threshold, bottom‐up effects of nutrients exceed top‐down control by predators, resulting in increased phytoplankton biomass. To test the hypothesis, an outdoor mesocosm experiment was conducted using different nutrient concentrations, and four treatments simulating daily loads of nutrients in Lake Taihu were studied: current average daily load (FC + 1NP), two times (FC + 2NP) and three times (FC + 3NP) average daily loads of nutrients with both fish and clams and as a control current daily loads without fish or bivalves.

2. MATERIALS AND METHODS

2.1. Experiment design

The outdoor mesocosm experiment was conducted from 12 September to 11 October 2018 at the Taihu Laboratory for Lake Ecosystem Research (TLLER), located at the northern edge of Lake Taihu. Lake Taihu (30°56′–31°34′ N, 119°54′–120°35′ E) is the third largest freshwater lake in China and a typical shallow eutrophic lake located in the south of the Yangtze River Delta. Lake Taihu has a surface area of 2338 km2, a mean depth of 1.9 m, and a mean water residence time of approximately 300 days.

Twelve polyethylene plastic tanks (82 cm height × 102 cm upper diameter × 85 cm bottom diameter) were filled with a 10 cm layer of prescreened (mesh size: 1.7 mm) sediment and 500 L filtered (mesh size: 64 μm) water collected from eutrophic Lake Taihu. A water pump (24 W) was fixed on the wall of each mesocosm 10 cm below the water surface to ensure water circulation with a run–stop ratio of 1 h:1 h. There was a 1‐week buffering and stabilization period before the beginning of the experiment.

To simulate the external nutrient loads in Lake Taihu, nitrogen (N) and phosphorus (P) were added daily to each mesocosm throughout the experiment. The mesocosm experiment comprised a total of four treatments: current average daily loads of N and P without filter‐feeding fish (F) or clams (C) (Control + 1NP) and three treatments at three nutrient concentrations with filter‐feeding fish and clams: current average daily load (FC + 1NP), two times (FC + 2NP), and three times (FC + 3NP) average daily loads (Table 1). Three replicates were run for each treatment. Nutrient concentrations used in the different treatments were based on average daily external nutrient loads of 130 μg N L−1 day−1 for nitrogen and 5 μg P L−1 day−1 for phosphorus, respectively (Paerl et al., 2011). N and P were added separately as aqueous solutions of potassium nitrate (KNO3) and potassium dihydrogen phosphate (KH2PO4), which are the main inorganic nutrients in Lake Taihu (Xu et al., 2021).

TABLE 1.

Biomass of filter‐feeding fish and clams and N and P concentrations representing different daily nutrient loads in Lake Taihu: Control + 1NP = concentrations representing current average daily load and no fish or clams, and treatments with fish and clams (FC) at nutrient concentrations representing current (FC + 1NP), two times (FC + 2NP), and three times (FC + 3NP) average daily loads.

Fish (g m−3) Clam (g m−2) Nitrogen (μg L−1 d−1) Phosphorus (μg L−1 d−1)
Control + 1NP 0 0 130 5
FC + 1NP 80 200 130 5
FC + 2NP 80 200 260 10
FC + 3NP 80 200 390 15
Open in a new tab

The clams and bighead carp were collected from a nearby aquaculture pond and had an average weight of 3.15 ± 0.23 g/ind. and 10.41 ± 0.92 g/ind., respectively. Prior to the experiment, all clams and fish were kept in a lake‐side pond and the water was changed every 3 days. The biomass levels of fish and clams used here are based on optimal biomass having filter‐feeding effects on phytoplankton from a previous study (Shen et al., 2020). Each treatment with FC had 4 fish and 38 clams, and biomasses averaged 41.63 ± 1.34 g per tank for fish and 119.54 ± 0.90 g per tank for clams, respectively. During the experiment, fish were checked daily and no deaths were noted throughout the experiment. It was not possible to check clam densities during the experiment, however at the end of the experiment all the clams were collected and counted, with mean survival rates of 93.3 ± 3.0%. At the end of the experiment, the biomasses of fish and clams averaged 48.29 ± 4.00 g per tank and 120.97 ± 3.49 g per tank, respectively.

2.2. Sampling and analyses

Water samples (5 L) were taken every 5 days from two different depths (surface and 5 cm above the sediment) using a 2.5 L plexiglass water collector and mixed. The concentrations of total nitrogen (TN), total phosphorus (TP), total dissolved nitrogen (TDN), and total dissolved phosphorus (TDP) were determined through colorimetry after digestion with K2S2O8 and NaOH solution (Rice et al., 2017). Total suspended solid (TSS) was calculated from 100 to 200 mL water samples filtered through a pre‐combusted and pre‐weighted GF/C filters, which were then dried in a muffle furnace (MF1050, Jinlan Instrument Manufacturing Co., Ltd, Shanghai, China) to a constant weight at 105°C for 4 h. Chlorophyll a (Chl a) was determined on known amounts of GF/C filter water (Whatman International Ltd., Maidstone, England) using a spectrophotometer (UV‐2450, Shimadzu Co., Ltd., Japan) (Rice et al., 2017). Filters were kept frozen until analyzed and pigments were extracted using a 90% (v/v) acetone/water solution. Prior to collecting water samples, water temperature (WT), electrical conductivity (EC), pH, and dissolved oxygen (DO) were measured using a YSI 9500 photometer (YSI Inco, Yellow Springs, OH, USA).

Phytoplankton and zooplankton were sampled on Day 0 (before bighead carp and clams were put in) and Day 30 (the end) of the mesocosm experiment. A 1 L water sample was preserved in 1% Lugol's solution and after 48 h of sedimentation, the supernatant was removed, and the residue was collected for identifying and counting rotifers and phytoplankton using a 100×–400× microscope. Species were identified according to the taxonomic classification of Hu and Wei (2006) and Wang (1961) and their biomasses were calculated from cell number and cell size measurements assuming 1 mm3 volume equals 1 mg fresh weight biomass (Zhang & Huang, 1991). Crustacean zooplankton were collected by filtering 5 L of mixed water samples through a plankton net (64 μm mesh) and then preserved in 5% formaldehyde solution. Cladocerans and copepods were counted at 40× magnification and species identification was performed according to standard methods (Chiang & Du, 1979). Zooplankton biomass was estimated based on the weight‐body size regression of Huang (1999).

2.3. Statistical analyses

One‐way analysis of similarities (ANOSIM) was used to test for differences in phytoplankton and zooplankton community structure based on Bray–Curtis similarity of log(x + 1) transformed species biomass data. In addition, we calculated two food web metrics, that is, zooplankton to phytoplankton biomass ratio (Zoop:Phyt) and Chl a to TP ratio by weight (Chl a: TP), representing the top‐down effects of filter‐feeding fish and clams and bottom‐up effects of nutrients, respectively.

Filter‐feeding fish and clams (Control + 1NP and FC + 1NP) and nutrients (FC + 1NP, FC + 2NP and FC + 2NP) were set as fixed factors to explore the effects on water quality parameters, that is, Chl a, TP, TDP, TN, TDN, the TN to TP ratio by weight (N:P), TSS, WT, EC, pH and DO, and time was incorporated as a random effect in our generalized linear mixed models (GLMM) (Bolker et al., 2009). We also applied GLMM to explore the influence of filter‐feeding fish and clams and nutrients as fixed effects, respectively, on the biomass of phytoplankton and zooplankton, Zoop:Phyt and Chl a: TP at the end of the mesocosm experiment. We assumed Gaussian error distributions for all response variables and fitted modes using function glmmTMB from R package glmmTMB (Brooks et al., 2017). For each response variable, we assessed the relative fit of the two different candidate models (fixed effect: FC and null, or Nutrient and null, respectively) and selected the best model based on the lowest AIC value from the default ANOVA function.

3. RESULTS

3.1. Nutrients and physicochemical parameters in water column

Negative effects of filter‐feeding fish and clams were found on TN, TP, TDN, TDP and TSS, and positive effects on N:P at nutrient concentrations representing current average daily nutrient loads (1NP) (Table 2, Figure 1). With the presence of filter‐feeding fish and clams, higher nutrient concentrations, positively affected N:P in both 2NP and 3NP treatments, while effects on TN, TP, TDN, TDP, and TSS were only evident in the 3NP treatment (Table 3, Figure 1). Neither filter‐feeding fish and clams nor nutrients had significant effects on WT or DO. No significant effects of filter‐feeding fish and clams were found on EC and pH, while higher nutrient concentrations positively affected EC (3NP) and pH (2NP and 3NP) (Table 3, Figure 2).

TABLE 2.

GLMM results showing the effects of filter‐feeding fish and clams on Chl a, nutrients, water physicochemical parameters, phytoplankton, and zooplankton at nutrient concentrations representing current average daily loads of N and P.

Variables Coefficients Estimate SE z‐Value p
Chl a (Intercept) 45.55 4.03 11.30 <.001
FC −11.30 5.70 −1.98 .047
TN (Intercept) 2.49 0.10 24.11 <.001
FC −0.81 0.09 −8.69 <.001
TP (Intercept) 0.12 0.01 17.17 <.001
FC −0.05 0.01 −6.39 <.001
TDN (Intercept) 1.76 0.08 21.97 <.001
FC −0.47 0.08 −6.33 <.001
TDP (Intercept) 0.05 0.00 14.83 <.001
FC −0.01 0.00 −8.09 <.001
N:P (Intercept) 20.42 0.89 22.87 <.001
FC 2.89 1.18 2.45 .014
TSS (Intercept) 49.92 4.32 11.57 <.001
FC −19.06 4.68 −4.07 <.001
WT (Intercept) 17.61 1.07 16.44 <.001
EC (Intercept) 455.49 3.62 125.80 <.001
pH (Intercept) 8.74 0.08 109.90 <.001
DO (Intercept) 7.52 0.31 24.63 <.001
Phytoplankton (Intercept) 18.63 3.45 5.41 <.001
FC −11.54 4.87 −2.37 .018
Zooplankton (Intercept) 10.55 1.59 6.61 <.001
FC −10.36 2.26 −4.59 <.001
Zoop:Phyto (Intercept) 0.57 0.12 4.56 <.001
FC −0.37 0.18 −2.10 .036
Chl a:TP (Intercept) 0.37 0.04 8.30 <.001
FC 0.09 0.05 1.69 .091
Open in a new tab

Note: Significant terms in bold. All variables were modeled as a Gaussian distribution.

Abbreviations: Chl a:TP, chlorophyll a to total phosphorus; FC, filter‐feeding fish and clams; N:P, total nitrogen to total phosphorus; Zoop:Phyt, zooplankton to phytoplankton biomass.

FIGURE 1.

FIGURE 1

Open in a new tab

Time series plots of Chl a, TP, TN, TDP, TDN, and the TN to TP ratio by weight (N:P) in four treatments with nutrient concentrations representing current (1NP), two times (2NP), and three times (3NP) average daily loads of N and P in Lake Taihu and with (FC) or without (Control) filter‐feeding fish and clams.

TABLE 3.

GLMM results showing the effects of filter‐feeding fish and clams on Chl a, nutrients, water physicochemical parameters, phytoplankton, and zooplankton at nutrient concentrations representing two times (2NP) and three times (3NP) average daily loads of N and P in Lake Taihu.

Variables Coefficients Estimate SE z‐Value p
Chl a (Intercept) 34.25 7.76 4.41 <.001
2NP −1.00 10.98 −0.09 .928
3NP 70.47 10.98 6.42 <.001
TN (Intercept) 1.68 0.22 7.61 <.001
2NP 0.34 0.21 1.59 .111
3NP 2.71 0.21 12.79 <.001
TP (Intercept) 0.08 0.01 6.47 <.001
2NP 0.00 0.01 −0.03 .980
3NP 0.07 0.01 8.69 <.001
TDN (Intercept) 1.28 0.22 5.90 <.001
2NP 0.38 0.21 1.75 .080
3NP 2.44 0.21 11.37 <.001
TDP (Intercept) 0.03 0.00 10.37 <.001
2NP 0.00 0.00 1.28 .201
3NP 0.02 0.00 6.48 <.001
N:P (Intercept) 23.31 3.10 7.51 <.001
2NP 6.45 2.21 2.92 .003
3NP 9.56 2.21 4.33 <.001
TSS (Intercept) 30.87 7.17 4.30 <.001
2NP −0.97 4.05 −0.24 .811
3NP 12.29 4.05 3.03 .002
WT (Intercept) 17.62 1.07 16.44 <.001
EC (Intercept) 452.82 3.26 138.86 <.001
2NP 0.21 3.98 0.05 .957
3NP 15.57 3.98 3.92 <.001
pH (Intercept) 8.73 0.09 96.20 <.001
2NP 0.09 0.04 2.43 .015
3NP 0.28 0.04 7.35 <.001
DO (Intercept) 7.35 0.19 38.59 <.001
Phytoplankton (Intercept) 7.09 4.45 1.60 .111
2NP −4.46 6.29 −0.71 .478
3NP 12.31 6.29 1.96 .050
Zooplankton (Intercept) 0.19 0.18 1.05 .292
2NP 0.09 0.25 0.37 .712
3NP 1.47 0.25 5.83 <.001
Zoop:Phyto (Intercept) 0.17 0.05 3.31 .001
Chl a:TP (Intercept) 0.45 0.07 6.51 <.001
2NP −0.01 0.08 −0.13 .899
3NP 0.29 0.08 3.54 <.001
Open in a new tab

Note: Significant terms in bold. All variables were modeled as a Gaussian distribution.

Abbreviations: Chl a:TP, chlorophyll a to total phosphorus; N:P, total nitrogen to total phosphorus; Zoop:Phyt, zooplankton to phytoplankton biomass.

FIGURE 2.

FIGURE 2

Open in a new tab

Time series plots of physicochemical parameters in four treatments with nutrient concentrations representing current (1NP), two times (2NP), and three times (3NP) average daily loads of N and P in Lake Taihu and with (FC) or without (Control) filter‐feeding fish and clams.

3.2. Chlorophyll a concentrations

Filter‐feeding fish and clams negatively affected Chl a concentration at nutrient concentrations representing current normal daily nutrient loads (1NP) (Table 2, Figure 1). With the presence of filter‐feeding fish and clams, effects of higher nutrient concentrations on Chl a concentration were only evident in the 3NP treatment, but not in the 2NP treatment (Table 3, Figure 1). Specifically, the concentrations of Chl a at the beginning of the experiment ranged from between 37.6 and 45.6 μg L−1, with no differences among treatments. At the end of the experiment, the mean Chl a concentration in Control +1NP was 51.51 ± 18.09 μg L−1, that is, 11x higher than the FC + 1NP (4.50 ± 1.13 μg L−1) treatment. Moreover, mean Chl a concentration in the FC + 3NP treatment was 146.99 ± 65.60 μg L−1, 33× and 22× higher than the FC + 1NP and FC + 2NP (6.83 ± 2.13 μg L−1) treatments, respectively (Figure 1).

3.3. Phytoplankton biomass and community composition

Phytoplankton abundance and biomass at the start of the mesocosm experiment averaged 4.5 ± 1.5 × 106 ind. L−1 and 1.8 ± 0.5 mg L−1, respectively, and did not differ among treatments. Cyanophyta and Chlorophyta contributed to >90% of the phytoplankton abundance, with Microcystis sp. (Cyanophyta) and Scenedesmus sp. (Chlorophyta) predominating, while Scenedesmus sp. (Chlorophyta) and Euglena sp. (Euglenophyta) predominated in terms of biomass.

At the end of the experiment, phytoplankton biomass was lower in the FC + 1NP and FC + 2NP treatments and higher in the FC + 3NP treatment (Tables 2, 3, Figure 3). Although total phytoplankton biomass did not differ in the Control + 1NP and FC + 3NP treatments (p > .05), phytoplankton composition did (p < .05). The Control + 1NP treatment was dominated by Cyanophyta (mainly Microcystis sp.), accounting for 61.8% of the total biomass and Chlorophyta (37.4%), while the FC + 3NP treatment was dominated by Chlorophyta (mainly Scenedesmus dimorphus and Pediastrum duplex), accounting for 73.5% of the total biomass and Bacillariophyta (18.6%). The FC + 1NP and FC + 2NP treatments had similar phytoplankton assemblages and were mainly dominated by Chlorophyta and Bacillariophyta (Figure 3).

FIGURE 3.

FIGURE 3

Open in a new tab

Phytoplankton (a) and zooplankton (b) biomass at the end of the experiment. Four treatments with nutrient concentrations representing current (1NP), two times (2NP), and three times (3NP) average daily loads of N and P in Lake Taihu and with (FC) or without (control) filter‐feeding fish and clams.

3.4. Zooplankton biomass and community

Zooplankton biomass at the beginning of the mesocosm experiment averaged 1.8 ± 0.5 mg L−1 and did not differ among treatments. Cladocerans dominated in all mesocosms (reaching 84.7 ± 27.3%), with Bosmina sp. and Ceriodaphnia cornuta as the predominant species.

At the end of the mesocosm experiment, cladocerans still dominated in the Control + 1NP treatment. Zooplankton biomass in FC + 1NP treatment decreased significantly compared with Control + 1NP treatment, while the zooplankton biomass in FC + 3NP treatment increased significantly compared with FC + 1NP and FC + 2NP (Figure 3). Rotifers, mainly Keratella sp. and Brachionus sp., had the highest biomass in the FC + 3NP treatment (41.4%), followed by branchiopods (33.2%) and copepods (25.4%) at the end of the mesocosm experiment (Figure 3).

3.5. Top‐down and bottom‐up effects

Zoop:Phyt was lower in all treatments with filter‐feed fish and clams, and GLMM revealed no significant effects of nutrients (Tables 2, 3, Figure 4). However, at high nutrient concentrations (3NP) we found significantly positive effects on Chl a:TP, while effects of filter‐feeding fish and clams on Chl a:TP were not significant at current average daily nutrient loads (1NP) (Tables 2, 3, Figure 4).

FIGURE 4.

FIGURE 4

Open in a new tab

Top‐down effects of filter‐feeding fish and clams and bottom‐up effects of nutrients on zooplankton to phytoplankton biomass (Zoop:Phyt; a) and Chl a to TP (Chl a: TP; b) at the end of the experiment. Four treatments with nutrient concentrations representing current (1NP), two times (2NP) and three times (3NP) average daily loads of N and P in Lake Taihu and with (FC) or without (control) filter‐feeding fish and clams. Values represented mean ± SE (n = 3). Different letters show significant differences (p < .05).

4. DISCUSSION

Consistent with our hypothesis, we found that when nutrient concentrations exceeded twice the current average daily loads of Lake Taihu, the bottom‐up effect of nutrients overrode the top‐down control of predators, resulting in an increase in phytoplankton biomass. This finding indicates that the top‐down effects of filter‐feeding fish and bivalves on phytoplankton can moderate the bottom‐up effects of high nutrient loads.

The findings that the combination of filter‐feeding fish and clams resulted in lower water column biomass of phytoplankton, as well as the concentrations of nutrients, TSS and Chl a, are consistent with our previous work (Shen et al., 2020). Furthermore, these results indicated that the combination of filter‐feeding fish and clams could effectively control the phytoplankton biomass at current average daily nutrient loads in Lake Taihu, but not at three times daily nutrient loads. Although zooplankton biomass (Bosmina sp.) increased significantly in the control treatment without fish and clams, total phytoplankton biomass and the proportion of cyanobacteria also increased. These results could be explained by the volume‐efficiency hypothesis proposed by Brooks and Dodson (1965). For example, while Bosmina sp. can feed on particle sizes of 70–80 μm, they are unable to feed on cyanobacterial colonies, most of which were larger than 80 μm (Burns, 1968; Xiao et al., 2018). Hence, feeding on smaller algae might contribute to cyanobacterial blooms by reducing competition with other algae. In contrast, filter‐feeding fish and clams have different filter‐feeding preferences, with bighead carp feeding more efficiently on large‐celled or colonies of algae while clams prefer smaller algae (Shen et al., 2020; Wang et al., 2018). Therefore, their combination and interactions extend the top‐down control for phytoplankton.

High external nutrient loadings could also significantly affect the efficacy of biomanipulations when using filter‐feeding fish and clams (Karpowicz et al., 2023). In our treatments with filter‐feeding fish and clams, Chl a concentrations and phytoplankton biomass were highest in nutrient concentrations representing three times average daily nutrient loads, but no differences were found at lower nutrient concentrations (FC + 1NP and FC + 2NP). The increase in phytoplankton biomass in the FC + 3NP treatment may be due to excessive nutrients that negated top‐down control of filter‐feeding fish and clams resulting in an increase in Chl a and phytoplankton biomass (Mao et al., 2020; Rettig & Smith, 2021; Shen et al., 2021). Although Microcystis aeruginosa blooms occurred in the control treatment, Chlorophyta predominated at the highest nutrient concentrations (3NP), a finding that might be explained by N:P in the water column (Chen et al., 2018; He et al., 2021; Shatwell & Köhler, 2019). Smith (1983) concluded from an analysis of growing season data from 17 lakes throughout the world that Cyanophyta tend to dominate at N:P < 29 while the relative proportion of Cyanophyta decreases at N:P > 29. This was consistent with our results that the N:P at the highest nutrient concentration (3NP) was much greater than 29, while that in the no‐fish‐clam treatment (Control + 1NP) N:P was close to this value.

It is well known that filter‐feeding fish are able to strongly suppress zooplankton biomass, especially larger sized, for example, cladocerans (Yin et al., 2022) and that bivalves can compete with zooplankton for food resources resulting in reduced zooplankton biomass (Atkinson et al., 2021; Rojas Molina et al., 2012). This assumption was also supported by our mesocosm results. Zooplankton biomass in the treatments with filter feeders (FC + 1NP and FC + 2NP) were at very low at the end of the experiment and much higher in FC + 3NP treatment. Furthermore, the highest nutrient concentrations (3NP) had a positive effect on zooplankton and phytoplankton biomass, while no significant effects were noted at the two lower nutrient concentrations with filter‐feeding fish and clams. This is mainly because the highest nutrient concentration studied here supported higher growth of phytoplankton, providing food for small‐sized rotifers, and predation by fish was likely weaker than that of cladocerans and copepods (Mao et al., 2020). Moreover, it is plausible that the high phytoplankton biomass in the FC + 3NP treatment resulted in lower water transparency, which may also have hindered fish predation on zooplankton (Chen et al., 2021; He et al., 2020).

5. CONCLUSION

Our results indicated that when external nutrient loads were twice the current average daily external loads of Taihu Lake (TN ≤ 260 μg L−1 d−1; TP ≤ 10 μg L−1 d−1), the combined stocking of filter‐feeding fish and bivalves (80 g m−3 bighead carp; 200 g cm−2 clams) could effectively control water column Chl a concentrations and phytoplankton biomass. However, at higher nutrient loads the top‐down control of phytoplankton by filter‐feeding fish and bivalves might not be as effective, resulting in an increase in the bottom‐up effects and increased of phytoplankton biomass. Thus, in order to ensure the efficacy of biomanipulation interventions using filter‐feeding fish and bivalves for algal control efforts should focus on controlling nutrient loads.

AUTHOR CONTRIBUTIONS

You Zhang: Conceptualization (equal); data curation (lead); formal analysis (lead); funding acquisition (equal); methodology (equal); software (lead); writing – original draft (lead). Ruijie Shen: Conceptualization (equal); data curation (equal); formal analysis (equal); methodology (equal); software (equal); writing – original draft (supporting). Kuanyi Li: Conceptualization (equal); funding acquisition (equal); writing – review and editing (equal). Qisheng Li: Methodology (equal). Huihui Chen: Methodology (equal). Hu He: Data curation (equal); formal analysis (supporting). Xiaohong Gu: Funding acquisition (equal); writing – review and editing (equal). Zhigang Mao: Conceptualization (equal); writing – original draft (supporting). Richard K. Johnson: Writing – review and editing (equal).

CONFLICT OF INTEREST STATEMENT

We declare there is no conflict of interest associated with any of the decisions made or components of this study.

Supporting information

Data S1.

Click here for additional data file. (28.4KB, xlsx)

ACKNOWLEDGMENTS

We thank the Taihu Laboratory for Lake Ecosystem Research (TLLER) for providing the experimental site for this study. This study was financially supported by the National Natural Science Foundation of China (Grant 31930074; 32071575), the National Key Laboratory of Water Disaster Prevention (2021491711), “Pioneer” and “Leading Goose” R&D Program of Zhejiang (2022C02038) and Young Scientist Foundation of Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences (2021NIGLAS‐CJH01).

Zhang, Y. , Shen, R. , Li, K. , Li, Q. , Chen, H. , He, Hu , Gu, X. , Mao, Z. , & Johnson, R. K. (2023). Top‐down effects of filter‐feeding fish and bivalves moderate bottom‐up effects of nutrients on phytoplankton in subtropical shallow lakes: An outdoor mesocosm study. Ecology and Evolution, 13, e10567. 10.1002/ece3.10567

Contributor Information

Xiaohong Gu, Email: xhgu@niglas.ac.cn.

Zhigang Mao, Email: zgmao@niglas.ac.cn.

DATA AVAILABILITY STATEMENT

All data are provided as Supporting Information.

REFERENCES

  1. Atkinson, C. L. , Halvorson, H. M. , Kuehn, K. A. , Winebarger, M. , Hamid, A. , & Waters, M. N. (2021). Filter‐feeders have differential bottom‐up impacts on green and brown food webs. Oecologia, 195, 187–198. [DOI] [PubMed] [Google Scholar]
  2. Benndorf, J. , Böing, W. , Koop, J. , & Neubauer, I. (2002). Top‐down control of phytoplankton: The role of time scale, lake depth and trophic state. Freshwater Biology, 47, 2282–2295. [Google Scholar]
  3. Bolker, B. M. , Brooks, M. E. , Clark, C. J. , Geange, S. W. , Poulsen, J. R. , Stevens, M. H. H. , & White, J. S. S. (2009). Generalized linear mixed models: A practical guide for ecology and evolution. Trends in Ecology & Evolution, 24, 127–135. [DOI] [PubMed] [Google Scholar]
  4. Brooks, J. L. , & Dodson, S. I. (1965). Predation, body size, and composition of plankton. Science, 150, 28–35. [DOI] [PubMed] [Google Scholar]
  5. Brooks, M. E. , Kristensen, K. , van Benthem, K. J. , Magnusson, A. , Berg, C. W. , Nielsen, A. , Skaug, H. J. , Machler, M. , & Bolker, B. M. (2017). glmmTMB balances speed and flexibility among packages for zero‐inflated generalized linear mixed modeling. R Journal, 9, 378–400. [Google Scholar]
  6. Burns, C. W. (1968). The relationship between body size of filter‐feeding cladocera and the maximum size of particle ingested. Limnology and Oceanography, 13, 675–678. [Google Scholar]
  7. Chen, J. , Liu, Z. , Xiao, S. , Chen, R. , Luo, C. , Zhu, T. , Cao, T. , Ni, L. , Xie, P. , Su, H. , & Zhang, M. (2020). Effects of benthivorous fish disturbance on chlorophyll a contents in water and the growth of two submersed macrophytes with different growth forms under two light regimes. Science of the Total Environment, 704, 135269. [DOI] [PubMed] [Google Scholar]
  8. Chen, K. , Li, Q. , Han, Y. , Li, W. , Zhang, Y. , Li, K. , & He, H. (2021). Effects of Carassius auratus on water turbidity, nutrient levels and plankton biomass in shallow lakes: A mesocosm experiment. Journal of Lake Science, 33, 397–404. [Google Scholar]
  9. Chen, M. , Ding, S. , Chen, X. , Sun, Q. , Fan, X. , Lin, J. , Ren, M. , Yang, L. , & Zhang, C. (2018). Mechanisms driving phosphorus release during algal blooms based on hourly changes in iron and phosphorus concentrations in sediments. Water Research, 133, 153–164. [DOI] [PubMed] [Google Scholar]
  10. Cheng, L. , Kattel, G. , Xue, B. , Yao, S. , Li, L. , & Liu, J. (2020). Application of subfossil Bosmina and its δ13C values in tracing the long‐term food web dynamics of shallow eutrophic lakes: A case in Taihu Lake, Southeast China. Science of the Total Environment, 730, 138909. [DOI] [PubMed] [Google Scholar]
  11. Chiang, S.‐c. , & Du, N. (1979). Fauna sinica; crustacea: freshwater cladocera. Science Press. [Google Scholar]
  12. de Rezende Ayroza, D. M. M. , do Carmo, C. F. , Camargo, A. F. M. , de Oliveira, M. D. , & Petesse, M. L. (2019). Net cages enhance golden mussel (Limnoperna fortunei) larval density and condition factor. Freshwater Biology, 64, 1593–1602. [Google Scholar]
  13. Han, Y. , Zhang, Y. , Li, Q. , Lürling, M. , Li, W. , He, H. , Gu, J. , & Li, K. (2022). Submerged macrophytes benefit from lanthanum modified bentonite treatment under juvenile omni‐benthivorous fish disturbance: Implications for shallow lake restoration. Freshwater Biology, 67, 672–683. [Google Scholar]
  14. He, H. , Chen, K. , Du, Y. , Li, K. , Liu, Z. , Jeppesen, E. , & Søndergaard, M. (2021). Increased nitrogen loading boosts summer phytoplankton growth by alterations in resource and zooplankton control: A mesocosm study. Frontiers in Environmental Science, 9, 772314. [Google Scholar]
  15. He, H. , Jeppesen, E. , Bruhn, D. , Yde, M. , Hansen, J. K. , Spanggaard, L. , Madsen, N. , Liu, W. , Søndergaard, M. , & Lauridsen, T. L. (2020). Decadal changes in zooplankton biomass, composition, and body mass in four shallow brackish lakes in Denmark subjected to varying degrees of eutrophication. Inland Waters, 10, 186–196. [Google Scholar]
  16. Hilt, S. , Alirangues Nunez, M. M. , Bakker, E. S. , Blindow, I. , Davidson, T. A. , Gillefalk, M. , Hansson, L. A. , Janse, J. H. , Janssen, A. B. G. , Jeppesen, E. , Kabus, T. , Kelly, A. , Kohler, J. , Lauridsen, T. L. , Mooij, W. M. , Noordhuis, R. , Phillips, G. , Rucker, J. , Schuster, H. H. , … Sayer, C. D. (2018). Response of submerged macrophyte communities to external and internal restoration measures in north temperate shallow lakes. Frontiers in Plant Science, 9, 194. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Hu, H. J. , & Wei, Y. X. (2006). The freshwater algae of China‐systematics, taxonomy and ecology. Science Press. [Google Scholar]
  18. Huang, X. (1999). Survey, observation and analysis of lake ecology, standard methods for observation and analysis in Chinese ecosystem research network, Series V. Standards Press of China. [Google Scholar]
  19. Jeppesen, E. , Søndergaard, M. , Lauridsen, T. L. , Davidson, T. A. , Liu, Z. , Mazzeo, N. , Trochine, C. , Özkan, K. , Jensen, H. S. , Trolle, D. , Starling, F. , Lazzaro, X. , Johansson, L. S. , Bjerring, R. , Liboriussen, L. , Larsen, S. E. , Landkildehus, F. , Egemose, S. , & Meerhoff, M. (2012). Chapter 6 – biomanipulation as a restoration tool to combat eutrophication: Recent advances and future challenges. In Woodward G., Jacob U., & O'Gorman E. J. (Eds.), Advances in Ecological Research (pp. 411–488). Academic Press. [Google Scholar]
  20. Jeppesen, E. , Søndergaard, M. , Mazzeo, N. , Meerhoff, M. , Branco, C. C. , Huszar, V. , & Scasso, F. (2005). Lake restoration and biomanipulation in temperate lakes: Relevance for subtropical and tropical lakes. In Reddy M. V. (Ed.), Restoration and Management of Tropical Eutrophic Lakes (pp. 341–359). CRC Press. [Google Scholar]
  21. Jin, H. , van Leeuwen, C. H. A. , Temmink, R. J. M. , & Bakker, E. S. (2022). Impacts of shelter on the relative dominance of primary producers and trophic transfer efficiency in aquatic food webs: Implications for shallow lake restoration. Freshwater Biology, 67, 1107–1122. [Google Scholar]
  22. Jin, H. , van Leeuwen, C. H. A. , Van de Waal, D. B. , & Bakker, E. S. (2022). Impacts of sediment resuspension on phytoplankton biomass production and trophic transfer: Implications for shallow lake restoration. Science of the Total Environment, 808, 152156. [DOI] [PubMed] [Google Scholar]
  23. Karpowicz, M. , Feniova, I. Y. , Sakharova, E. G. , Gorelysheva, Z. I. , Więcko, A. , Górniak, A. , & Dzialowski, A. R. (2023). Top‐down and bottom‐up control of phytoplankton communities by zebra mussels Dreissena polymorpha (Pallas, 1771). Science of the Total Environment, 877, 162899. [DOI] [PubMed] [Google Scholar]
  24. Liu, H. , Zhou, W. , Li, X. , Chu, Q. , Tang, N. , Shu, B. , Liu, G. , & Xing, W. (2020). How many submerged macrophyte species are needed to improve water clarity and quality in Yangtze floodplain lakes? Science of the Total Environment, 724, 138267. [DOI] [PubMed] [Google Scholar]
  25. Liu, Z. W. , Hu, J. , Zhong, P. , Zhang, X. , Ning, J. , Larsen, S. E. , Chen, D. , Gao, Y. , He, H. , & Jeppesen, E. (2018). Successful restoration of a tropical shallow eutrophic lake: Strong bottom‐up but weak top‐down effects recorded. Water Research, 146, 88–97. [DOI] [PubMed] [Google Scholar]
  26. Lürig, M. D. , Best, R. J. , Dakos, V. , & Matthews, B. (2021). Submerged macrophytes affect the temporal variability of aquatic ecosystems. Freshwater Biology, 66, 421–435. [Google Scholar]
  27. Mao, Z. G. , Gu, X. H. , Cao, Y. , Zhang, M. , Zeng, Q. F. , Chen, H. H. , Shen, R. J. , & Jeppesen, E. (2020). The role of top‐down and bottom‐up control for phytoplankton in a subtropical shallow eutrophic lake: Evidence based on long‐term monitoring and modeling. Ecosystems, 23, 1449–1463. [Google Scholar]
  28. Paerl, H. W. , Xu, H. , Mccarthy, M. J. , Zhu, G. , Qin, B. , Li, Y. , & Gardner, W. S. (2011). Controlling harmful cyanobacterial blooms in a hyper‐eutrophic Lake (Lake Taihu, China): The need for a dual nutrient (N & P) management strategy. Water Research, 45, 1973–1983. [DOI] [PubMed] [Google Scholar]
  29. Parr, T. B. , Vaughn, C. C. , & Gido, K. B. (2020). Animal effects on dissolved organic carbon bioavailability in an algal controlled ecosystem. Freshwater Biology, 65, 1298–1310. [Google Scholar]
  30. Qin, B. (2020). Shallow Lake limnology and control of eutrophication in Lake Taihu. Journal of Lake Science, 32, 1229–1243. [Google Scholar]
  31. Rettig, J. E. , & Smith, G. R. (2021). Relative strength of top‐down effects of an invasive fish and bottom‐up effects of nutrient addition in a simple aquatic food web. Environemental Science and Pollution Research, 28, 5845–5853. [DOI] [PubMed] [Google Scholar]
  32. Rice, E. W. , Baird, R. B. , & Eaton, A. D. (2017). Standard methods for the examination of water and wastewater. American Public Health Association. [Google Scholar]
  33. Rojas Molina, F. , José de Paggi, S. , & Frau, D. (2012). Impacts of the invading golden mussel limnoperna fortunei on zooplankton: A mesocosm experiment. Zoological Studies, 51, 733–744. [Google Scholar]
  34. Scheffer, M. , Carpenter, S. , Foley, J. A. , Folke, C. , & Walker, B. (2001). Catastrophic shifts in ecosystems. Nature, 413, 591–596. [DOI] [PubMed] [Google Scholar]
  35. Shatwell, T. , & Köhler, J. (2019). Decreased nitrogen loading controls summer cyanobacterial blooms without promoting nitrogen‐fixing taxa: Long‐term response of a shallow lake. Limnology and Oceanography, 64, S166–S178. [Google Scholar]
  36. Shen, R. , Gu, X. , Chen, H. , Mao, Z. , Zeng, Q. , & Jeppesen, E. (2020). Combining bivalve (Corbicula fluminea) and filter‐feeding fish (Aristichthys nobilis) enhances the bioremediation effect of algae: An outdoor mesocosm study. Science of the Total Environment, 727, 138692. [DOI] [PubMed] [Google Scholar]
  37. Shen, R. J. , Gu, X. H. , Chen, H. H. , Mao, Z. G. , Zeng, Q. F. , & Jeppesen, E. (2021). Silver carp (Hypophthalmichthys molitrix) stocking promotes phytoplankton growth by suppression of zooplankton rather than through nutrient recycling: An outdoor mesocosm study. Freshwater Biology, 66, 1074–1088. [Google Scholar]
  38. Smith, V. H. (1983). Low nitrogen to phosphorus ratios favor dominance by blue‐green algae in lake phytoplankton. Science, 221, 669–671. [DOI] [PubMed] [Google Scholar]
  39. Søndergaard, M. , Lauridsen, T. L. , Johansson, L. S. , & Jeppesen, E. (2017). Nitrogen or phosphorus limitation in lakes and its impact on phytoplankton biomass and submerged macrophyte cover. Hydrobiologia, 795, 35–48. [Google Scholar]
  40. Wang, J. (1961). Fauna sinica, rotifer. Science Press. [Google Scholar]
  41. Wang, L. , Liu, P. , Sun, J. , Zhang, Y. , Zhou, Q. , Wu, Z. , & He, F. (2018). Comparison and combination of selective grazing on natural seston by benthic bivalves (Hyriopsis cumingii) and pelagic fish (Hypophthalmichthys molitrix). Environmental Science and Pollution Research, 25, 33423–33431. [DOI] [PubMed] [Google Scholar]
  42. Xiao, M. , Li, M. , & Reynolds, C. S. (2018). Colony formation in the cyanobacterium Microcystis . Biological Reviews, 93(3), 1399–1420. [DOI] [PubMed] [Google Scholar]
  43. Xu, H. , McCarthy, M. J. , Paerl, H. W. , Brookes, J. D. , Zhu, G. , Hall, N. S. , Qin, B. , Zhang, Y. , Zhu, M. , Hampel, J. J. , Newell, S. E. , & Gardner, W. S. (2021). Contributions of external nutrient loading and internal cycling to cyanobacterial bloom dynamics in Lake Taihu, China: Implications for nutrient management. Limnology and Oceanography, 66, 1492–1509. [Google Scholar]
  44. Yin, C. , He, W. , Guo, L. , Gong, L. , Yang, Y. , Yang, J. , Ni, L. , Chen, Y. , & Jeppesen, E. (2022). Can top‐down effects of planktivorous fish removal be used to mitigate cyanobacterial blooms in large subtropical highland lakes? Water Research, 218, 118483. [DOI] [PubMed] [Google Scholar]
  45. Yu, J. , Liu, Z. , Li, K. , Chen, F. , Guan, B. , Hu, Y. , Zhong, P. , Tang, Y. , Zhao, X. , & He, H. (2016). Restoration of shallow lakes in subtropical and tropical China: Response of nutrients and water clarity to biomanipulation by fish removal and submerged plant transplantation. Water, 2016, 438. [Google Scholar]
  46. Zhang, P. , Zhang, H. , Wang, H. , Hilt, S. , Li, C. , Yu, C. , Zhang, M. , & Xu, J. (2022). Warming alters juvenile carp effects on macrophytes resulting in a shift to turbid conditions in freshwater mesocosms. Journal of Applied Ecology, 59, 165–175. [Google Scholar]
  47. Zhang, Z. , & Huang, X. (1991). Methods in freshwater plankton study. Science Press. [Google Scholar]
  48. Zhou, Y. , Li, H. , Wu, W. , Zhang, L. , Zhu, H. , & Naren, T. (2021). Mapping Taihu Basin research: A bibliometric analysis. Environmental Reviews, 29, 391–400. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Data S1.

Click here for additional data file. (28.4KB, xlsx)

Data Availability Statement

All data are provided as Supporting Information.

Articles from Ecology and Evolution are provided here courtesy of Wiley

RESOURCES