Nontraditional biomanipulation: A powerful ecotechnology to combat cyanobacterial blooms in eutrophic freshwaters

Cyanobacterial blooms, occurring frequently in eutrophic freshwaters worldwide, are considered as potential hazards to ecosystems and human health, and it is often difficult and expensive to control their outbreaks in large lakes through reduction of nutrient loadings. Biomanipulation, launched as an ecology-based solution in the 1970s, was once believed to be an effective way to counteract cyanobacterial blooms. It is divided into traditional biomanipulation (TB) and nontraditional biomanipulation (NTB) that use filter-feeding Daphnia and filter-feeding fish, respectively. There have been numerous reviews on the former, yet few on the latter. Here, we first revisit the debate on the digestibility of cyanobacteria in silver and bighead carp. Then, we review 42 experiments that clearly mention cyanobacterial changes and reveal substantial reductions in cyanobacterial abundance by filter-feeding carp in 88% of the cases. In particular, in a whole-lake experiment in Lake Donghu, increased stock of silver and bighead carp effectively decreased Microcystis blooms from a coverage of 87% in 2021 to 0% in 2022. Finally, we discuss possible factors related to NTB��s effectiveness that depends not only on standing stock, niche divergence and shape preference of the fish but also on trophic status of the waterbodies. Particularly, silver and bighead carp feed more effectively on colony-forming Microcystis than on filamentous cyanobacteria, but are capable of increasing small-sized algae. NTB can be used to prevent or diminish cyanobacterial blooms that are poorly grazed by Daphnia, providing an effective and sustainable in-lake ecotechnology to combat heavy cyanobacterial blooms in eutrophic waterbodies.



INTRODUCTION
Problem of cyanobacterial blooms in freshwaters was noticed by George Francis as early as ca.150 years ago. 1 Over the past 50 years, incidences of toxic cyanobacterial blooms (especially Microcystis, Anabaena, Aphanizomenon, or Oscillatoria) have increased in frequency, magnitude and duration globally [2][3][4][5] due to increased eutrophication resulting from urbanization, industrialization, and intensification of agricultural activities. 63][14][15][16][17][18] Cyanobacterial blooms are highly dynamic, with growth and loss processes operating on a time scale of days and (scums) appearing and disappearing on even shorter time scales. 2Blooms of cyanobacteria in lakes result not only from external influences (e.g., nutrients, light, temperature, and hydrological conditions) 19 but also from complex interactions within an aquatic ecosystem (e.g., competition between cyanobacteria and macrophytes, grazing by herbivores such as Daphnia, phytoplanktivorous fish, bivalves, and bacteriophage). 20Therefore, regulation of external physico-chemical influences and/or internal ecological interactions can be employed to alleviate cyanobacterial blooms. 21,22Both cyanobacterial blooms and internal phosphorus loading can reinforce each other, 23,24 forming a positive feedback loop.
The reduction of external and internal nutrient loadings to levels at which the development of cyanobacterial blooms is restricted can offer the most effective and persistent solutions, but this approach is very expensive, can take years or decades, and targets the whole watershed, requiring national or even international efforts. 3,25For example, to reduce nutrient loadings of Lake Taihu (surface area 2338 km2 ), more than 100 billion CNY was spent during the past fifteen years, yet coverage of cyanobacterial blooms reached as large as 1582 km 2 on May 16, 2017. 26The use of various algicides or harvesting of surface scum can only be executed as a provisional measure. 22Biological control mainly includes the use of microbial antagonists (viruses, pathogenic bacteria or fungi) and various filter-feeding animals, such as Daphnia, mollusks, and phytoplanktivorous fish.Among these methods, manipulating of an entire food web by the removal of zooplanktivorous fish or introduction of piscivores to increase the abundance of Daphnia has been a popular method since the 1970s. 27In contrast, little attention has been devoted to the use of the phytoplanktivorous silver carp (Hypophthalmichthys molitrix) and bighead carp (Aristichthys nobilis), 28 probably due to the scarcity of such species in North America and Europe.
31][32][33][34][35] After a short introduction to the origin and challenge of TB, we focus on the validation of NTB, from long debates on digestability of cyanobacteria in carp to experimental evidences from 42 case studies (mostly from China) conducted in enclosures, ponds, aquariums, and presedimentation ponds of waterworks, and one whole-lake experiment.All reviewed experimental studies clearly mentioned the influences of silver and/or bighead carp on cyanobacteria, whereas cases only reporting the effects on total algal abundance or chlorophyll a (Chl-a) concentration were excluded.

THE PATH TO BIOMANIPULATION Origin
The idea of manipulating the food web to increase large zooplankton populations for the control of excessive algal growth can be traced back to the late 1950s.Before that time, limnological studies focused only on the effects of physicochemical factors on phytoplankton, which cascaded upward through the food chain (from zooplankton to fish), and fish were considered only response variables and not driving variables. 36,37In the late 1950s and early 1960s, pioneering works by Hrbacek et al. 38 and Brooks & Dodson 39 made limnologists realize that top predators such as fish can also have profound impacts on lower trophic level organisms such as phytoplankton.Later, Hurlbert et al. 40 found through a mesocosm experiment that Gambusia affinis reduced rotifers and crustaceans and caused an increase in phytoplankton and a decrease in transparency, indicating that manipulation of the fish population (especially through increasing piscivores) can decrease the abundance of phytoplankton.Then, the term "biomanipulation" was coined by Shapiro et al. 27 to refer to the management of aquatic communities through a series of interventions of the biota, and of their habitats to reduce algal biomass, and in particular, of cyanobacteria.On the other hand, this term is frequently used in the restricted sense favored in Europe as "topdown" manipulation, i.e., an increase in piscivores results in decreased planktivorous fish, increased herbivorous zooplankton, and decreased phytoplankton.The large-bodied Daphnia are often considered the most suitable zooplankters to control algal biomass.Fish manipulation occurs through addition or removal of piscivores or planktivores, fish poisons, and so on. 413][44] Yet, biomanipulation is completely natural in its components, with no chemicals or machineries needed, although it is often criticized that alterations of algal composition and biomass distributions by "biomanipulation" are not "biological control of eutrophication". 45

From hope to hopelessness
Attempts to control cyanobacterial blooms started in the 1970s and 1980s when the negative consequences of eutrophication could no longer be ignored 2 and studies on the trophic cascading effects of fish through zooplankton predation on phytoplankton were intensively carried out; the success rates of these measure were limited.Biomanipulation was once believed to be an effective way to counteract algal blooms when nitrogen and phosphorus levels could not be effectively reduced in the short term.Numerous laboratory, pond, enclosure and whole-lake studies were conducted to test the effectiveness of "top-down" manipulation. 37,46However, although these efforts appeared to be successful, that was not always the case, and it was challenging to make the effects of biomanipulation more stable, that is, last longer. 41Although often hailed as an optimal approach, "top-down" manipulation has not received the unequivocal support that might be expected, as most biomanipulation measures that involved the introduction of piscivores could not reduce algal abundance. 47Few studies reported a successful decline in cyanobacterial biomass after piscivore stocking was used as the only restoration measure. 20Filter-feeding mussels are also a candidate for biomanipulation, nevertheless, it is difficult to use them in practice to ingest cyanobacterial blooms, as these invertebrates are benthic.Biomanipulation seems at best to be an ephemeral substitute for nutrient reduction.
Many biomanipulation measures have produced spectacular initial results, but only a few have succeeded in maintaining low algal biomasses for several years. 48There are several reasons for the instability of biomanipulation using Daphnia.It is not only dependent on the quantitative relations between predators and prey-either piscivores and planktivores or planktivores and herbivores-but is also affected by the perpetuation of such relationships.In attempts to control planktivores, great variability occurs from year to year and from lake to lake, resulting in variability in herbivore abundance and size distribution and ultimately in algal abundance, species composition and water transparency. 41Even initially successful manipulations may show declines in effectiveness due to species substitutions of zooplankton and phytoplankton, reinvasion by planktivore-benthivores, proliferation of ungrazable algae and changes in nutrient recycling; therefore, in most cases, some maintenance is required. 49Variation may occur due to stochastic changes in hydrodynamics. 503][54] The lifespan of Daphnia is very short, and its seasonal succession is rapid.Filtration of crustaceans could potentially alter a phytoplankton community in favor of difficult-to-process filamentous or colonyforming cyanobacteria, increasing toxic bloom formation after biomanipulation. 20,55,56Small-sized cladocerans develop stronger tolerance against micro-cystins (MCs) than large-sized cladocerans when both groups are exposed to toxic Microcystis, explaining why dominant Daphnia species are usually replaced by small-sized cladocerans when cyanobacteria bloom occur in summer in eutrophic lakes. 57Given the above mentioned facts, it is clear that large Daphnia are neither able to reduce the populations of inedible cyanobacteria nor tolerate cyanotoxins.

CYANOBACTERIA IN SILVER AND BIGHEAD CARP
In the late 1990s, biomanipulation was subdivided into traditional and nontraditional approaches, and the former refers to biomanipulation using zooplankton to reduce the amount of phytoplankton, while the latter refers to biomanipulation using filter-feeding fish, such as silver carp and bighead carp, to reduce the amount of colony-forming cyanobacteria. 58Planktivorous fish use two distinct behaviors to feed on plankton: particulate feeding and filter feeding.Particulate feeders visually attack individual planktonic prey, being "active-visibility selective predators" of zooplankton, whereas filter feeders intake a volume of water and retain prey by passing this volume of water over gill rakers, serving as "passive escape-selective predators" of zooplankton plus "passive size-selective grazers" of phytoplankton. 59Silver and bighead carp are major commercial fish in China, feeding on zooplankton, phytoplankton and fine detritus.However, filter-feeding fish outside of Asia can seldom be used to counteract cyanobacterial blooms.Tilapia in the tropics have the potential to graze on cyanobacteria, however, it is difficult for them to survive winter in temperate or subtropical regions.Outside Eastern Asia, few successful attempts to control cyanobacterial blooms with filterfeeding fish have been reported. 20

What do they ingest?
Silver and bighead carp have been cultured since the Tang Dynasty (618-907 AD) in polyculture ponds for domestic Chinese carp.According to "Wings to the Erya" by Yuan Luo in the Southern Song dynasty, silver and bighead carp were believed to feed on feces of grass carp, and this belief lasted until the 1950s.This is because during polyculture, food was given to black and grass carp but not to silver and bighead carp, yet they were able to grow, which made people mistakenly believe that they feed only on the feces of grass carp. 60espite the long history of cultivating silver and bighead carp, people did not know their planktivory until the 1950s.Scientific research on the feeding habits of silver and bighead carp began in the early 1950s in China.Ni and Chiang 61 investigated the food habits of silver and bighead carp in natural waters by using microscopic analysis of intestinal contents, cultures of intestinal contents near the anus, and microscopic observation of the structure of gill rakers.The food habits of silver and bighead carp were found to be somewhat different: bighead carp fed mainly on zooplankton (ratio of zooplankton to phytoplankton = 1:4.5),while silver carp fed mainly on phytoplankton (ratio of zooplankton to phytoplankton = 1:248).Based on a δ 15 N mass balance model, the contributions of zooplankton to the diet of silver carp and bighead carp were 54% and 74%, respectively. 62ecause of their lack of ability to select food, bighead carp also eats some phytoplankton, and silver carp also eats some zooplankton, and their difference in feeding habits was mainly due to different structures in their gill rakers, i.e., the gill rakers of bighead carp are relatively sparse, while those of silver carp are relatively dense (Table 1).For example, as its gill raker spacings range from 20 to 25 µm, 64 silver carp feed selectively on particles larger than 20 µm. 85The intestine of silver carp was also longer than that of bighead carp, in agreement with their difference in food habits, as prolonged exposure to digestive enzymes along a long digestive tract favors effective digestion of algal cells.

What do they digest?
The types of algae that can or cannot be digested by silver and bighead carp has been a topic of debate for decades.A prerequisite to utilizing algal cell contents is the breakdown of algal cell walls.There are three possible mechanisms: acid hydrolysis, enzymatic digestion, and mechanical trituration. 868][89] For stom-achless silver carp, the pH value of its gut fluids is usually >6.The absence of cellulase in the gut fluids also indicates its difficulty in breaking down cellulose cell walls by enzymatic digestion. 61,90The digestibility of algae in silver and bighead carp has been debated for decades.The results from gut contents and digestive enzyme analysis suggest poor utilization, while measurements of food assimilation using radiolabeled isotope techniques indicate reasonable assimilation efficiency.
In a paper by Ni and Chiang, 61 the conclusions on the digestibility of algae in silver and bighead carp were not accurate.According to the culture of intestinal contents near the anus, they suggested that Chrysophyceae, dinoflagellates, Cryptophyceae, and diatoms were easy-to-digest species, while cyanobacteria, green algae and Euglena were difficult-to-digest species, as these species have fibrous cell walls or gelatinous envelopes and silver and bighead carp lack the enzymes required to decompose fiber or gelatin.
Subsequent studies either supported or opposed this view.Oscillatoria granulata and Anabaena werrzeri passed undamaged through the guts of silver carp, whereas diatoms and Protococcales were digested. 91Large amounts of dead and live cells were detected by epifluorescence microscopy in the excrements of silver carp fed toxic Microcystis aeruginosa. 92No differences were observed between the structures of filamentous cyanobacterial cells (mainly colony forms of Aphanizomenon flos-aquae, selectively ingested) in the final part of the gut of silver carp and in the plankton of four Masurian Lakes, demonstrating that these algae were not digested. 67ecomposing algae were better assimilated than live algae by silver carp. 93he digestibility of algae in silver and bighead carp was investigated by microscopic comparison of algal appearance before and after gut passage, e.g., Scenedesmus appeared microscopically identical in the foregut and hindgut of silver carp, indicating that this green alga was not digested, and Euglena was even more mobile in the hindgut than in the foregut. 90In an in vitro experiment, after 9 h of incubation with the gut juice of silver carp, only 9.5% of the proteins of the filamentous A. flos-aquae were detected as free amino acid. 9497][98][99][100][101] How do they digest?
The disruption of a centric diatom, Cyclotella, on its passage through the esophagus and intestine of silver and bighead carp was studied in the field.75-77 Cyclotella was chosen as an indicator of the digestion process of silver carp, as this diatom has a hard, less digestible and silicated frustule and is thus easily recognized under a microscope.As high as 50-60% of intact Cyclotella were broken or damaged upon passing through the esophagus, while this figure declined to only 15-20% from proximate to distal ends of the intestine, and 20-30% of Cyclotella remained intact in the feces, indicating that disruption of algal cell walls occurs principally by the pharyngeal teeth.A similar result was also reported by Dong 102 using the green alga Scenedesmus.The proportion of broken Scenedesmus greatly increased from 6.4% in the water to 18.8% in the foregut and further to 22.0% in the hindgut.When sand was added, these proportions declined to 4.4%, 11.4% and 14.1%, respectively, indicating that sand disturbed digestion efficiency, also demonstrating the importance of mechanical trituration of the pharyngeal teeth.
Mechanical trituration of the pharyngeal teeth helps to explain the contradictory conclusions on the digestibility of algae in silver and bighead carp.On passage through the intestine, some algal cells remain intact, leading to the erroneous conclusion by some authors 61,86,90,94 that phytoplankton are poorly utilized.Broken algal cells are actively assimilated in the intestines even though the proportion of intact algal cells may change little upon passage through the intestines, explaining the active assimilation of cyanobacteria and green algae by carp observed via isotopic techniques.Observations of intact or mobile algae in the hind gut or feces 61,94,103 do not necessarily mean that this species is indigestible.Food digestion is balanced by both the gain and expenditure of energy during digestion, and incomplete digestion may reflect an adaptive strategy for stomachless, filter-feeding fish that ceaselessly feed on plankton as well as organic detritus of low nutritional value. 76,77n they digest (toxic) cyanobacteria?Using the 14 C method, Panov et al. 97 found that silver and bighead carp willingly consumed cyanobacteria, as Anabaena and Aphanizomenon were better assimilated than several green algae.Using the 32 P method, Shi et al. 98 found that the average utilization rate of A. spiroides by silver carp fingerling was as high as 71.3%, and Zhu and Deng 101 observed that the assimilation rate of M. aeruginosa was 35-48% by silver carp and 23-38% by bighead carp.Based on the 14 C method, 17% of ingested A. flos-aquae were found to be assimilated by silver carp fry.
Because of the buoyancy of cyanobacteria, the incomplete digestion strategy of silver and bighead carp means that their feces will float in water and be consumed repeatedly, achieving thorough digestion.In a laboratory experiment by Chen and Liu, 104 the digestion rates of feces (excreted by fish feeding on Microcystis) by silver and bighead carp were 68.6% and 58.5%, respectively, and were much higher than the ingestion rates of Microcystis (only 29.5% and 26.1%, respectively), 105 suggesting that attached bacteria might play a positive role in enhancing the digestion of feces.
When stocked in a large net cage (1.1 km 2 ) located in Lake Taihu where dense toxic cyanobacterial blooms occurred, silver carp not only effectively ingested toxic Microcystis cells (up to 84.4% in total phytoplankton) but also showed fast growth, i.e., from 141 g to 1759 g in mean weight in one year. 106 similar result was found for bighead carp. 107A filed study shows that during toxic cyanobacterial blooms in Lake Taihu, the phytoplanktivorous silver and bighead carp displayed only slight ultrastructure changes in liver, while the carnivorous fish presented the most serious injury of hepatocytes such as swollen endomembrane system and deformation of the nuclear outline. 108ilver and bighead carp are strongly resistant to toxic cyanobacteria, as they not only had a fast growth, but also accumulated only low levels of MCs in the liver, a target organ for these toxins, perhaps due to efficient detoxification in liver and active excretion through the conjugation of MCs with glutathione (GSH) or cysteine (Cys) in the kidney. 7,109Therefore, it is quite possible to use these two fish to counteract toxic cyanobacterial blooms.

VALIDATION OF NONTRADITIONAL BIOMANIPULATION Early debates abroad
Silver and bighead carp are endemic to East Asia, spawning with semibuoyant eggs in river. 110,111The native range of silver carp is from the Heilongjiang River in the north to the Red River in the south, and that of bighead carp is from the Haihe River in the north to the Pearl River in the south.Currently, both fish have been introduced to Africa, North America, South America, New Zealand and Europe.Their introductions can be traced back to the end of the 19th century, but mainly to the 1960s-1980s. 112efore the introduction of Chinese carp, few phytophagous fish were of commercial importance in Europe.The first introduction of silver and bighead carp to Europe was likely related to common carp (Cyprinus carpio) cultured in ponds.Common carp, native to China, Japan and Central Asia and an important food for humans, is probably the first fish species to be introduced worldwide.During a period beginning from the Middle Ages, carp was spread nearly globally through a series of introductions and is present in most areas where climatic conditions permit its survival. 113n Europe, pond aquaculture has been dominated by carp monoculture for centuries.Common carp has a very wide spectrum of food, eating almost all invertebrates consumed by other fish, and therefore, historical attempts to culture indigenous fish with carp have been unsuccessful in Europe.While polyculture was practiced in Chinese ponds, monoculture was dominant for carp in Europe, which presented problems related to the improvement of pond water quality and the full utilization of food resources to increase fish yield.During the early period of the introduction of silver and bighead carp in Europe, scientists and land managers showed great interest in these carp for the following reasons.First, there were no indigenous filter-feeding planktivorous fish in Europe, and second, carp served as food; moreover, the introduc-tion of silver carp can both improve water quality and lead to the harvest of other edible fish. 114fter silver and bighead carp were introduced from the USSR to Poland from 1964-1966, 114 they became subjects of interest for increasing fish production through the utilization of insufficiently exploited sources of food; moreover, since silver carp feed on phytoplankton, it was hoped that silver carp could be used to counteract algal blooms and improve water quality. 115roduction increases through polyculture was easily confirmed, but debates remained as to the influence of carp on water quality improvement (at least in terms of phytoplankton abundance).Silver carp was introduced to North America in the early 1970s mainly for the purpose of improving water quality in ponds containing channel catfish and prawns.Excessive phytoplankton growth and its decomposition in fish ponds is a problem because of nighttime oxygen deficits, fish kills, and chemicals that ruin the flavor of fish. 116owever, attempts to control phytoplankton in fertilized aquaculture ponds in eastern Europe and America during the 1960s-1980s using filter-feeding fish produced unsatisfactory or even negative results, probably due to inadequate filtration and strong disturbance of sediments by fish in small shallow water bodies, where phytoplankton production was extremely high and the phytoplankton community was driven by bottom-up forces, i.e., nutrient recycling by carp outperformed their filtration of phytoplankton.For example, during April and October 1968, Janusko 117 conducted experiments in 26 common carp ponds co-cultured with three one-year-old phytophagous fishes to evaluate their effects on phytoplankton communities.With the application of fertilizer, a stock of 1500 ind.ha −1 silver carp increased 10% of algal biomass (from 38 to 42 mg L −1 ), while a stock of 3000 ind.ha −1 silver carp decreased 10% of algal biomass.A stock of 1500 ind.ha −1 bighead carp led to an over 100% increase in algal biomass (from 27 to 66 mg L −1 ).Silver carp increased the proportion of diatoms, but bighead carp greatly increased the proportion of filamentous cyanobacteria.However, Janusko suggested that the 10% decrease in algal biomass in the 3000 ind.ha −1 silver carp ponds was not caused by fish grazing but by other factors.Opuszynski 118 investigated the influences of silver carp at varied densities on water quality in carp ponds.The carp density was 4000 ind.ha −1 , and silver carp densities were 0, 4000, 8000 and 12000 ind.ha −1 .When silver carp were stocked, the net production of phytoplankton greatly increased, and the biomass and total production of phytoplankton and Chl-a also showed small but significant increases, whereas both the density and biomass of zooplankton showed some declines.Phytoplankton biomass was significantly higher in channel catfish ponds containing bighead and silver carp. 119Silver carp was used to improve the water quality of aquaculture ponds (0.4-hectare) for the production of the freshwater prawn Macrobrachium rosenbergii, yet total Chl-a amounted to almost 40% in ponds containing as few as 30 silver carp.Silver carp consistently decimated net-plankton (>10 µm) Chl-a, but nanoplankton (<10 µm) and total Chl-a increased in prawn ponds containing free-roaming silver carp. 120

Evidences from mesocosm experiments
There were several classic enclosure experiments to show the ability of filter-feeding carp to control cyanobacterial blooms.The first experiment was in the pond-type Warniak Lake (Poland), where Kajak et al. 115 conducted an enclosure experiment in 1973.Silver carp, at densities of 30 and 90 g/m 3 , decreased zooplankton biomass by an average of 4.5 and 16 times, respectively, the algal biomass declined 4.5-fold at both fish densities, and the biomass share of colony-forming cyanobacteria (mainly M. aeruginosa) also decreased.The inhibitory effects were striking, but the explanations sound farfetched: "the intensity of feeding of these fish was rather small, as can be concluded from the small increase in their weight…Much lower biomass of plankton including nanoplankton, which was not consumed by fish, and slightly smaller amounts of organic and inorganic substances in water, undoubtedly were the result of sedimentation; the seston consumed by silver carp was excreted later as its feces".The second experiment was in Paranoa Reservoir (Brazil), where Starling & Rocha 121 conducted an enclosure experiment in 1988.A high density (125 g/m 3 ) of silver carp decreased the biomass of a filamentous cyanobacterium, Cylindrospermopsis raciborskii, although the decline was not as large as expected.The third set of experiments was in eutrophic Lake Donghu (Wuhan, China), where enclosure

REVIEW
124] To date, there have been ample experimental studies, mostly in China, that confirmed the ability of silver and bighead carp to control cyanobacterial blooms.These were mainly enclosure experiments, also included a few pond, presedimentation pond and aquarium experiments.The phytoplankton in the experimental systems (particularly in the control) listed were all dominated by colony-forming or filamentous cyanobacteria.88% of these experiments showed that silver and bighead carp can effectively control cyanobacterial (especially Microcystis) blooms beyond a certain density (Table 2).
A few experimental studies are cited as examples of failures or bottlenecks in NTB measures for the control of cyanobacteria. 20In an enclosure experiment in Villerest Reservoir (France), silver carp was stocked at densities of 0, 8, 16, 20 and 32 g/m 3 , and the densities of cyanobacteria (dominated by M. aeruginosa) increased in all treatments, with the largest increase in the 8 and 16 g/m 3 enclosures and the lowest increase in the 32 g/m 3 enclosures. 140The negative results may be attributed to the very short experimental period (only 28 days).If the stocking density of silver carp is at a medium or low level, a short-term experiment may not produce a significant result.Of course, if the density is high enough, the fish introduced for biomanipulation can also have significant effects within a short time.Another example comes from Ke et al. 159 By using three large fish pens (each 0.36 km 2 ) in Lake Taihu (China), the authors evaluated the influence of silver and bighead carp (approximately 40 g/m 3 ) on plankton and found that total phytoplankton biomass decreased, but changes in Microcystis biomass and MCs concentrations, when compared to surrounding lake water, were statistically insignificant.However, as the pens were made of net with a mesh size of 2 cm×2 cm, the plankton inside and outside the fences could be exchanged freely.Shen et al. 154 conducted a pond experiment to assess the impact of silver carp (at biomass levels of 0, 25, 50, and 100 g/m 3 ) on phytoplankton, and claimed that silver carp stimulated phytoplankton growth.Two concerns are listed here.First, the experiment only lasted one month.Second, final Chl-a amounts in all ponds (with fish or without fish) declined to nearly below 6 µg/L that is too low to test the effectiveness of NTB that targets cyanobacterial blooms with high biomass.
Evidence from a whole-lake manipulation So far, few studies have experimentally tested the applicability of NTB theory at a whole lake level.Heavy cyanobacterial blooms occurred in Lake Guozheng (12.8 km 2 ), a main part of the subtropical shallow Lake Donghu in Wuhan City, in the summer of each year since the 1970s, and the phytoplankton community was dominated by colony-forming Microcystis and filamentous Anabaena and Oscillatoria. 21,160Since 1971, the stateowned Donghu Lake Fish Farm has stocked the lake every year with mainly filter-feeding silver and bighead carp (the carp are not provided with fertilizer or other food sources, but rather, they feed on plankton in the lake).Yet, the blooms disappeared suddenly in 1985 when the fish standing stock exceeded a threshold of 50 g/m 3 , 21 and remained absent for as long as 36 years, demonstrating the effectiveness and stability of NTB.On the other hand, since 2000, tens of billions of CNY has been spent to reduce nutrient loadings to the lake, which has greatly reduced nitrogen and phosphorus levels in the lake water.However, in the summer of 2021, Microcystis blooms suddenly reoccurred again in the lake (Figure 1, Video S1), with a coverage of 0.06 km 2 on July 6, and increased to 3.46 km 2 on August 16, and further to 11.2 km 2 (ca.87% of Lake Guozheng) on September 11.The blooms lasted for nearly half a year even though the government spent over 30 million CNY to combat the blooms, mainly through algicide spray applied manually by boats or by unmanned aerial vehicles and via pumping the cyanobacterial scum into drainage pipes (Video S2).This indicates that nutrients, especially phosphorus, in the lake water were still abundant enough for the outbreak of cyanobacterial blooms.
The re-occurrence of cyanobacterial blooms in Lake Donghu can be attributed to several management changes.The first is an inadequate fish abundance.The standing stock of silver and bighead carp declined gradually after 2003, falling below a threshold of 50 g/m 3 during 2017 and 2021 (Figure 2).The second is an improper species composition.Share of bighead carp in total fish stock increased to as high as 80%, weakening grazing pressure on algae per fish stock.The third is an inappropriate size at stocking.Carp size at stocking increased from 100-250g to 750-1000g, decreasing grazing pressure on algae per fish yield.The fourth is an incorrect harvest timing.Fishing time was extended from winter to a period from summer to winter, resulting in the release of grazing pressure on algae in summer.Zooplankton could not be a cause for the outbreak of cyanobacterial blooms in 2021, because from 2017 to 2020, there was an increasing trend in zooplankton (especially the crustacean) biomass, and the large-sized Daphnia hyalina that is used for traditional biomanipulation showed an obvious increase, despite being a small number (Figure 3).
The long-term monitoring data (Figure 2) in Lake Donghu also suggest a substantial grazing pressure of filter-feeding fish on phytoplankton.In terms of annual mean, fish standing stock was negatively related with total algal biomass (r=-0.572,P=0.004), filamentous cyanobacterial biomass (r=-0.508,P=0.013), and Microcystis biomass (r=-0.320,P=0.136).The lower correlation with Microcystis was likely due to rather low Microcystis abundance before 2021.
To prevent the re-occurrence of Microcystis blooms, management adjustments were made in 2022 as follows.First, the amount of fish released in 2022 was increased to twice that in 2021, with a ratio of silver carp to bighead being 2:1, and with 1/3 being 100-250g fingerlings (Video S3).Second, duration of fishing was limited to winter (fishing was immediately stopped after the occurrence of the blooms in 2021).As a result, fish stock in the lake increased to 82 g/m 3 in March 2022, and further to 192 g/m 3 in September 2022, according to estimates by echo sounder.These measures were particularly successful: Microcystis blooms were completely suppressed in 2022 (i.e., no bloom occurred) despite the occurrence of an extremely hot and dry summer in Wuhan for over half a century.In 2022, not only did the government not spend any money to spray algicide, but Donghu Lake Fish Farm also made a profit of millions of CNY by commercial fishing in winter.It is estimated that in 2022, stocking brought 9.9 t N and 2.1 t P to the lake, while harvesting removed 17.4 t N and 3.7 t P from the lake.As there were 20.9 t N and 2.8 t P in the water column, fishing eventually removed an amount equal to 36% N and 57% P in the water column.These demonstrate not only effectiveness but also environmental and economic sustainability of NTB.
It has long been a concern how algae respond to planktivorous fish recycled nutrients.Interestingly, although fish standing stock was over 4 times higher in 2022 than in 2021, annual mean TN, TP and NH 4 + concentrations in  Heavy Microcystis blooms suddenly occurred in 2021 after a gradual decline of fish standing stock since 2004, but were completely suppressed in 2022 through greatly increasing stocks of filter-feeding carp, validating the effectiveness of nontraditional biomanipulation (NTB).
the lake water declined respectively from 1.04 mg/L, 0.13 mg/L and 0.15 mg/L in 2021 to 0.75 mg/L, 0.10 mg/L and 0.12 mg/L in 2022, and algal biomass also showed a significant decline (Figure 3).This indicate that concerns on the so-called "ichthyoeutrophication" in natural lakes by fish excrement are unnecessary 127,[161][162][163][164][165][166] despite such phenomenon can be frequently observed in small experimental systems. 127,167,168

FACTORS AFFECTING THE EFFECTIVENESS OF NTB Niche divergence
Gill raker spacing is considered the most important determinant of filtering capability in planktivorous fish. 59The niches of silver and bighead carp diverged somewhat due to different gill raker spacings.The gill rakers of bighead carp are sparser than those of silver carp; therefore, under the same conditions, silver carp collects more algae than bighead carp, whereas bighead carp collects more zooplankton than silver carp.In other words, silver carp is primarily phytoplanktivorous, whereas bighead is predominantly zooplanktivorous.However, silver and bighead carp are also believed to feed opportunistically, lacking stable ecological niches, for example, phytoplankton usually constitute a small percentage of the diet of bighead carp, but "bloom" conditions could result in increased consumption of algae amounting to 70% of the food bolus. 169ood selection of filter-feeding carp is also dependent on their own density, i.e., if fish density is low, the proportion of zooplankton in their food will increase, reducing feeding on cyanobacteria. 159At very low stocking densities, silver and bighead carp may experience less competition and shift their feeding behavior toward zooplankton, thereby decreasing the overall grazing efficiency of cyanobacteria. 170

Fish density dependency
Successful NTB relies on sufficient grazing pressure on cyanobacteria, which requires sufficient stocking densities.As algae consistently reproduce, it is necessary that the grazing pressure from fish exceeds the rate of increase in algae or the particular algal species that need to be controlled (e.g., bloom-forming cyanobacteria).Both grazers and algae essentially have similar responses to temperature changes, i.e., with an increase in temperature (within certain ranges), the rates of both grazing and algal growth increase.However, algal growth is also affected by physicochemical factors (nutrients, light, rainfall, hydrological conditions, etc.).For example, effects of filter-feeding animals on phytoplankton could be lowered by high nutrient loading, 147 due to nutrients-driven increase in growth rates of phytoplankton.
It is crucial to establish a fish density threshold for the application of NTB.For example, in eutrophic Lake Donghu, silver and bighead carp should be kept at a stocking threshold of > 50 g/m 3 , because a decline in grazing of these carp would make the water body vulnerable to cyanobacterial blooms again (Figure 2).Under ideal conditions, both silver and bighead carp can grow rapidly.Selecting a threshold can be a challenge, i.e., is the threshold the initial stocking density (generally at the beginning of the year), the average fish density in the growing season, or the total fish yield?Because the feeding efficiency of fish varies widely across different ages, the size of fish at the time of stocking will also have an impact.In addition, the fishing period will affect the dynamics of the algal community, thereby altering the effects of NTB.

Shape preference
Filamentous algae cannot be efficiently retained by the filtering apparatus of silver and bighead carp.In Florida (US) ponds, bighead carp were found to feed selectively (Ivelv's index > ±0.9) on Botryococcus braunii, a spherical phytoplankton with a size of 80±51 µm, but the filamentous Lyngbya lagerheimii and L. limnetica (125-130 µm in length and 1-2 µm in diameter) were not efficiently consumed (Ivelv's index -0.1 to +0.1). 114In enclosures in the Paranoa Reservoir (Brazil), suppression of filamentous C. raciborskii by silver carp at a density as high as 125 g/m 3 was not as strong as expected. 121In a pond experiment by Januszko, 117 bighead carp even resulted in a large increase in cyanobacteria (mainly A. flos-aquae, A. solitaria var.planctomica and O. agardhii).In Lake Donghu (China), filamentous cyanobacteria are more resistant to grazing by filter-feeding carp than the spheric Microcystis colony (Figure 2): mean biomass of Microcystis and filamentous cyanobacteria were respectively 0.05 mg/L and 0.35 mg/L during 2000-2020, and were respectively 1.97 mg/L and 0.81 mg/L in 2021 when heavy cyanobacterial blooms occurred, but declined respectively to 0.01 mg/L and 0.10 mg/L when fish standing stock was increased by more than 4-times.It appears that silver and bighead carp prefer spherical shapes to filamentous shapes (Table 3), therefore being more efficient at feeding on Microcystis than on filamentous cyanobacteria.While, the long-term data of Lake Donghu indicate that stocking of silver and bighead carp did greatly decrease the abundance of filamentous cyanobacteria (Figure 2).

ADVANTAGES AND DISADVANTAGES OF NTB
NTB measures has advantage in controlling cyanobacterial blooms over TB measures because colony-forming (e.g., Microcystis) or filamentous cyanobacteria (e.g., Anabaena, Aphanizomenon) are inedible for Daphnia 51 but edible for silver and bighead carp. 106,107The approach is quite different for the two types of biomanipulation, due mainly to the size of food the animals filter.Differences in the lifespans of the species also affect the stability of the Microbial infections or Daphnia grazing may cause sudden or short-term declines in algal biomass 3 but can rarely achieve a persistent decline in cyanobacterial blooms, whereas silver and bighead carp can achieve prolonged suppression of cyanobacterial blooms for decades or cause a rapid elimination of cyanobacterial blooms within a short period of time (Figure 2).Because silver and bighead carp can only reproduce naturally in large rivers, 110,[177][178][179] their populations in isolated lakes are controllable, i.e., if the stocking is stopped, their populations will decline gradually.TB typically works most effectively in lakes with low levels of eutrophication (TP< 50 µg/L in shallow lakes, <20 µg/L in deep lakes), where small algae usually dominate. 180Nevertheless, NTB can be applied to eutrophic or hypertrophic waters with dense blooms of cyanobacteria (especially the colonial Microcystis), but it is not recommended to use it in oligotrophic or less productive waters to control phytoplankton.][183][184] Silver and bighead carp may favor the development of nanophytoplankton, as they remove not only its competitors (netphytoplankton) but also its consumers (filter-feeding zooplankton), 67,115 however, the opposite results were observed in fertilized aquaculture ponds. 117Algal control is not necessarily the same thing as cyanobacterial control because cyanobacteria represent only a fraction of all algae.Cyanobacteria that form surface blooms have gas vesicles, by which they can float on the water.They consist mainly of Microcystis, as well as many filamentous species.Constrained by the pore size of the gill rakers, silver and bighead carp feed mainly on algae larger than 10-20 μm (Table 1) and thus provide more niches for the development of small algae, sometimes even leading to much higher total algal abundance, and thus declines in water clarity. 185The colony size of Microcystis is also changed from large to small by grazing of filter-feeding carp. 186Stocks of these carp do not necessarily cause co-declines of algal biomass, Chl-a and nutrient concentrations (TN, TP), but can effectively control the size structure of the algal community.Thus, NTB is used to combat colony-forming cyanobacterial blooms but not necessarily total algal abundance.Although filter-feeding fish are "passive size-selective grazers", their direct grazing impact on phytoplankton is a consequence of the interaction between fish feeding rates and algal growth rates, and the latter usually decreases with increasing algal size, 59 also favoring increases in small-sized algae.
In conclusion, NTB has been proven to be a powerful in-lake ecotechnology to combat cyanobacterial blooms, especially when nutrients cannot be reduced sufficiently within a short period of time.It has advantages of quick effect, good stability, environmental friendliness and economical sustainability.Although challenging, it is now a time to strengthen the use of NTB, as a part of systematic control of eutrophication, to overcome cyanobacterial bloom problems in large eutrophic lakes.This review is expected to provide decision-makers with a good scientific evidence base for the use of NTB.

Figure 1 .
Figure 1.Outbreak of cyanobacterial blooms (A-C) and emergent treatment via algicide spray by manual application from boats (F) and by unmanned aerial vehicles (E) and via pumping scum into drainage pipes (D) in the summer of 2021 in Lake Donghu.

3 )
Silver carp and bighead carp

Figure 2 .
Figure 2. Seasonal changes of phytoplankton, nutrients and standing stock of silver and bighead carp during 2000 and 2022 in Lake Donghu (mid-lake station), Wuhan, China Filamentous cyanobacteria include Dolichospermum, Lyngbya, Leptolyngbya, Oscillatoria, Phormidium, Planktolyngbya, Planktothrix, Pseudanabaena and Raphidiopsis.Heavy Microcystis blooms suddenly occurred in 2021 after a gradual decline of fish standing stock since 2004, but were completely suppressed in 2022 through greatly increasing stocks of filter-feeding carp, validating the effectiveness of nontraditional biomanipulation (NTB).

Figure 3 .
Figure 3. Changes in the annual means of NH 4 + , total phosphorus, total nitrogen, and biomass of phytoplankton, Cladocera and zooplankton during 2000 and 2022 in Lake Donghu.

Table 1 .
Size of food items and the distance between gill rakers of the silver and bighead carp

Table 2 .
Microcystis bloomed in other E, but adding of fish quickly eliminated blooms in E1, E2, E5.Blooms in no fish E6 were persistent.No bloom in the lake with dense SC+BC.Sep., 1992 Microcystis bloomed in all E without fish.Blooms were decreased quickly by SC or BC, but not by grass carp.Blooms were persistent in fish-free E1 and E2.Blooms were absent in the lake with SC+BC.

Table 3 .
174d selection and preference of silver and bighead carpDong and Li 1994174Removal rate increased with increasing food size between 3.2-493 µm.Silver carp feed on algae more efficiently than bighead carp Removal rate increased with increasing food size between 3.2-493 µm.Bighead feed on zooplankton more efficiently than silver carp Mu et al. 2012 175 Ivlve's selective index was much higher for cyanobacteria (especially Merismopedia punctata 17.0, Microcsystis incerta 5.9, Coelosphaerium dubium 4.89) than for other algae Li et al. 2013 176 Cyanobacteria comprised 74.5% of the plankton food.Ivlve's selective index was 1.53 for M. aeruginosa, and 1.42 for total cyanobacteria Cyanobacteria comprised 56.0% of the plankton food.Ivlve's selective index was 1.30 for M. aeruginosa, and 1.07 for total cyanobacteria biomanipulation: Daphnia can live only for months, whereas silver and bighead carp can live over 7-8 years, reaching a maximum body weight of 50-80 kg.