Water quality in a shallow eutrophic lake is unaffected by extensive thinning of planktivorous and benthivorous fish species

improvement in water clarity or reductions of nutrients, organic particles, chlorophyll concentrations, or watercolor, despite a 6-fold thinning of total estimated fish biomass, from 112 to 19 kg ha (cid:0) 1 . Over the period, the macrophyte cover increased from 0.8 to 13.5 %, but no recruitment of large piscivorous fish (perch and pike ( Esox lucius) > 10 cm) was detected. We found higher correlations of particle concentration and water clarity to water temperature than to wind speed, which indicates sediment particle resuspension by the remaining fish community (mostly carp Cyprinus carpio ) that forage on benthos in shallow lakes. Further system-ecological research in Lake Bromme should evaluate whether thinning the stock of carp and increasing plant cover may improve water quality and test which optical properties sustain high water turbidity and prevent shallow, eutrophic lakes like Lake Bromme from responding to intense fish thinning.


Introduction
Re-establishing good water quality in eutrophic freshwater lakes is often hindered by internal phosphorus (P) cycling deriving from the historical nutrient accumulation in the sediments (Jeppesen et al., 2000;Sand-Jensen et al., 2017).Thus, many lakes continue to suffer from dense summer phytoplankton blooms and turbid water, which constrain submerged macrophytes and maintain unbalanced food webs (Jeppesen et al., 2000).The European Water Framework Directive aims to establish habitability for aquatic biodiversity, ensure access to safe freshwater resources, and facilitate recreational interest by obtaining good water quality, thus, much effort is dedicated to lake restoration.High external P loading is often identified as a primary reason for sustained poor water quality, but wastewater management and restricted P application on nearby farmland have reduced external P loading.However, the problem of internal P loading in European lakes remains, despite broad lake restoration efforts (Van De Bund and Van Donk, 2002).
Some have proposed that an imbalance in a lake's food webs might contribute to an unfavorable eutrophic state (Jeppesen et al., 2000;Søndergaard et al., 2008).Meanwhile, it has also been suggested that biomanipulation of the functional structure of biological communities might promote a more balanced trophic structure (Gulati et al., 2008).
Many European lakes have been subjected to a variety of biomanipulation attempts having involved planting submerged macrophytes to inhibit sediment resuspension and nutrient retention, stocking mussels to enhance grazing pressure on phytoplankton, or altering the fish composition (Van De Bund and Van Donk, 2002;Gulati et al., 2008).The latter has frequently been part of attempts to reduce sediment bioturbation and increase herbivorous zooplankton to enhance their grazing pressure on phytoplankton (Drenner and Hambright, 1999;Olin et al., 2006;Søndergaard et al., 2008).
A balanced, well-functioning fish community is achieved when planktivorous and benthic-feeding fish co-exist with a sustainable stock of piscivores that can regulate the recruitment of fish fry and maintain the stock of mature fish at an eco-appropriate level (Hambright et al., 1991;Gulati et al., 2008).However, facilitating such a balance is fraught with difficulties.High densities of small cyprinids like roach (Rutilus rutilus L.) often harm water quality as they consume large quantities of herbivorous zooplankton and increase water turbidity while scavenging the littoral sediments for plants, residue, and invertebrates (Brabrand et al., 1990;Horppila, 1994).Meanwhile, a high abundance of large, opportunistic benthivorous cyprinids like bream (Abramis brama L.) and carp (Cyprinus carpio L.) may degrade the water quality by preventing sediment consolidation (Scheffer et al., 2003) and suspending large quantities of loose sediments and particle-bound nutrients (Meijer et al., 1990;Breukelaar et al., 1994;Huser et al., 2022).Moreover, they may deplete benthic prey populations (Zambrano et al., 2001) and dislodge macrophytes while searching for benthic invertebrates (Lammens et al., 2002).In Northern European lakes, the main piscivores are European perch (Perca fluviatilis L.) and pike (Esox lucius L.).Even young perch control cyprinid yearlings effectively (Beeck et al., 2002), and, while young pike may also predate on fish fry, and adult pike can engulf larger prey fish than perch; pike is generally unable to effectively constrain the density of 0+ cyprinid stocks (Skov et al., 2003).
Ideally, fish thinning should serve as a 'shock therapy' that induces a regime shift from an algae-dominated lake ecosystem favoring cyprinids to a stable, top-down food web regulated by piscivorous fish (Hansson et al., 1998).Such an environment can be stimulated by removing >75 % (Perrow et al., 1997) of fish biomass quickly, within one to three years (Hansson et al., 1998), either alone or combined with the stocking of predatory fish (Jurajda et al., 2016).Manipulation of fish populations may be directed towards thinning adult and juvenile fish, as well as removing fish larvae and egg strips manually (Jurajda et al., 2016).While thinning planktivorous fish should facilitate the survival of more herbivorous zooplankton to graze on the phytoplankton, the thinning of benthivorous fish should reduce sediment resuspension, decrease predation pressure on benthic macroinvertebrates, and in combination, improve the underwater light climate and facilitate the presence of submerged macrophytes (Meijer et al., 1990;Zambrano et al., 2001;Huser et al., 2022).The ambient light intensity is also a fundamental factor controlling physiological processes in fish (e.g., growth rate, feeding, and spawning activities) according to the species' adaptations in morphology and behavior (Ruchin, 2021).Thus, less turbid water may also contribute to restructuring the functionality of the lakes' fish communities.As good recruitment of large, adult piscivorous fish is highly dependent on sufficient refuge and food availability for juveniles (Beeck et al., 2002), thus, thinning large juvenile cohorts may be necessary to prevent intra-as well as interspecific competition and promote fast growth of large adult stages (Jurajda et al., 2016).
Previous studies on biomanipulation relying on fish thinning have been based on the reasonable expectation that removal of fish biomass leads to rapid declines in water P concentration, lower phytoplankton, and cyanobacterial biomass, increased water transparency, recolonization of benthic invertebrates and expansion of submerged macrophytes (Van De Bund and Van Donk, 2002;Søndergaard et al., 2017;Huser et al., 2022).However, the effects of lake restorations that rely exclusively on fish thinning are rarely long-lasting, as most biomanipulated lakes return to their initial turbid state after a few years (e.g., 6-10 years) even when the initial effect was substantial (Van De Bund and Van Donk, 2002;Søndergaard et al., 2008).Similarly, despite efficient fish thinning, water quality sometimes does not improve, as the fundamental mechanisms that govern the water quality in some lakes are indifferent to the effects of changes in fish populations (Bernes et al., 2015;Jurajda et al., 2016).Though external P loading (Peretyatko et al., 2012) and high internal P release from the sediments are fundamental to inhibiting long-lasting effects from biomanipulation (Van De Bund and Van Donk, 2002;Jurajda et al., 2016;Søndergaard et al., 2017), relapse of poor water quality has also been attributed to a recurring imbalance within the fish community, as it has often been proven difficult to prevent extensive yearly recruitment of cyprinid fry, along with their deleterious effects on zooplankton densities and promotion of enhanced nutrient turnover, in the long term (Hansson et al., 1998;Olin et al., 2006).Moreover, attempts to manually increase piscivorous stocks to suppress cyprinid recruitment, mainly by stocking pike, have commonly failed (Skov and Nilsson, 2007;Gulati et al., 2008).
We were aware of the general outlines of the theory, but sought to test the impact of a novel amendment to the theory.Usually, pelagic and benthic-feeding fish are thinned simultaneously, using traditional fishing gear (Jurajda et al., 2016), perhaps complemented with such drastic interventions as lake drainage or the addition of toxins to eradicate total fish communities (Van De Bund and Van Donk, 2002).However, simultaneous removal prevents examination of the respective feeding groups' influence on water quality and macrophyte coverage, thereby complicating the formulation of optimal thinning plans.By contrast, we were intrigued by the idea that fish belonging to different feeding groups have relatively distinct ecosystem influences and can be regarded as engineers of aquatic ecosystems (Jeppesen et al., 2000;Scheffer et al., 2003).Thus, substantial interventions that focus on specific influences are expected to yield significant yet bounded results (Breukelaar et al., 1994;Zambrano et al., 2001).Moreover, seasonal weather changes are rarely addressed quantitatively in biomanipulation studies, although seasonal temperature fluctuations are known to control fish spawning and feeding activity (Volkoff and Rønnestad, 2020).Additionally, wind influences sediment consolidation and resuspension, and may trigger the release of nutrients and elevated light attenuation in the water column of shallow lakes (Kragh et al., 2017;Martinsen et al., 2022).Likewise, episodic precipitation may deliver large quantities of nutrients and organic matter from the catchment, potentially inducing algal blooms, oxygen depletion, and fish kills (Kragh et al., 2020).
We aimed to evaluate the environmental impacts of successive, extensive thinning of benthivorous and planktivorous fish in shallow, eutrophic Lake Bromme, Denmark.
We hypothesized that: 1) reducing the stocks of large benthic feeding cyprinids (bream and tench) would reduce sediment resuspension; 2) thinning planktivorous fish (roach, small bream, and small perch) would suppress the phytoplankton biomass; and 3) the two effects in combination would improve water quality and underwater light conditions for submerged macrophytes.Simultaneously, we examined the influence of temporal variation of precipitation, wind speed, and water temperature (as a proxy for fish feeding activity) on water chemistry.

Study site
Lake Bromme is a small (12.5 ha), shallow (mean depth 1.48 m, maximum depth 2.4 m), alkaline, eutrophic lake located near the city of Munke Bjergby, in Sorø Municipality, Denmark (55 • 28′52"N, 11 • 30′54"E) (Fig. 1).The catchment (1.1 m 2 (Müller and Rostgaard, 2018)) is mainly forested and includes a nearby coniferous plantation and a few houses.The lake is classified as a calcareous, non-brown water lake.The lake was once used to store raw timber that fell during a storm in 1967.Today, only a few timbers remain in the lake, which despite having a distinct cloudy yellowish-brown color (Høj and Dahl, 1993), does not contain high levels of colored dissolved organic materials, thus degradation of the logs is not affecting the water coloration.The state of Lake Bromme's water chemistry and biota have for decades been monitored within the Danish lake monitoring program.Mean summer Secchi transparency is between 0.72 m and 1.30 m (May-September averages, 2001-2017 (The-Danish-Environmental-Protection-Agency, 2023),).The lake has no surface inflow; incoming water derives from precipitation and diffusive seepage, and the estimated water retention time is 1.5 years (Müller and Rostgaard, 2018).At high water levels in winter and after heavy rainfall, water flows from the lake through a shallow ditch to a downstream lake (Fig. 1), though with a negligible possibility of fish migration.Lake Bromme is eutrophic (28-60 μg total phosphorus L − 1 , 1066-1559 μg total nitrogen L − 1 May-September averages 2005-2017, ODA, 2023)).The external P loading, equivalent to 21 μg P L − 1 , was modeled based on external P inputs from complementary Danish catchment compositions Kragh (2022), with no major input from ducks, swans, geese, or piscivorous birds (e.g., cormorants), and the mean summer level of chlorophyll a was relatively modest (19-32  Historical data of the submerged macrophyte community indicate a downward trend in the relative plant-covered area from 5.18 % in 2005 to 3.3 % in 2013, presumably caused by the high water turbidity (Müller and Rostgaard, 2018).
National routine monitoring of the fish diversity, conducted as part of the NOVANA program, found nine fish species, including the piscivorous perch and pike, the omnivorous roach, sunbleak (Leucaspius delineatus), rudd (Scardinius erythrophthalmus L.) primarily feeding in the pelagic, and the opportunistic, benthic feeding bream, ruffe (Gymnocephalus cernua L.), carp, and crucian carp (Carassius carassius L.) searching for food in the upper sediments (The-Danish-Environmental-Protection-Agency, 2023).Like many other Danish lakes (Carl and Rask Møller, 2012), Lake Bromme contains a small stock of old, adult carp kept for private recreational fishing.According to the fishing rights owner, no recreational fishing for carp took place during the biomanipulation, thus no use of pre-bait, adding nutrients to the lake, and no recent stocking of new carp was performed.The presence of opportunistic and piscivorous European eel (Anguilla anguilla L.) and benthivorous tench (Tinca tinca L.) was known from recent recordings during recreational fishing submitted to the Danish fish database 'Fishatlas' (DK: Fiskeatlas).
During the latest national standardized fish survey in 2013, the lake's fish biomass was estimated at six tons (437 kg ha − 1 ), heavily dominated by bream and roach, with a negligible presence of piscivorous perch and pike (>10 cm; Müller and Rostgaard, 2018).Following the Danish lake restoration recommendations (Søndergaard et al., 2020), 80 %, equivalent to 4800 kg, of the non-predatory fish biomass was to be removed.

Fieldwork
The field study was conducted between March 2019 and September 2021.Standardized fish surveys were conducted yearly to assess selected species' density and population structure.The water's physicochemical parameters were measured throughout the study.Data on weather conditions (precipitation, air temperature, and wind speed and direction) were retrieved from the Danish Meteorological Institute.Lake bathymetry, sediment density (hardness), and submerged macrophyte cover were measured by sonar and side-scan analyses following Kragh et al. (2017) and Baastrup-Spohr et al. (2016).Data on sediment chemical properties was retrieved from the Danish national lake survey database (The-Danish-Environmental-Protection-Agency, 2023).The relationships between wind speed and direction on water clarity and suspended particulate organic matter in the lake water were examined with a linear model to evaluate the effect of the seasonal weather conditions on water turbidity.
Fish removal -Fish thinning was conducted at weekly intervals, initially targeting large bream and, later, planktivorous roach, young bream, and small perch (<10 cm).With the intention of sparing piscivorous fish (perch >10 cm and all pike) while culling large bream, we used a combination of three-dimensional fish traps, cod, eel, and shrimp fykes, and umbrella fish traps baited with a mix of ground-bait, corn, and maggots.The opening apparatus of the umbrella fish traps was modified to allow for more convenient and fast handling of the entrapped fish.The traps and fykes were deployed from a boat and checked every 1-3 days; entrapped live piscivore specimens were released.Up to 52 umbrella fish traps were operated at a time.Cast nets were thrown a few times from a boat to thin schools of small cyprinids.Umbrella fish traps and fykes were replaced by gillnets in mid-June 2019.To selectively target roach, small bream, and small perch, finemeshed gillnets (10-60 mm) were deployed from April 2020 to September 2021.All removed fish were sorted into species and weighed.An intermediate plastic bulk container (1000 L) equipped with a pound net was installed in late March 2020 to entrap large cyprinids missed in year one.
Fish surveys -Fish surveys were conducted twice annually to assess the population structure and biomass of species in response to the thinning.Three of the fish surveys were conducted within the standard official period between August 15 and September 15 (herein the "standardized fish surveys") as described in the following.Each year, we also conducted a similar fish survey earlier in the summer, between late June and mid-July, to allow for examination of the adult fish growth and survival rate over the summer.Six pelagic gillnets (multi-mesh, with EUstandardized dimensions, 30 × 1.5 m with 2.5 m sections of different mesh sizes (CEN)) were equipped with floaters and anchors and placed from a boat overnight.The gillnets were repeatedly deployed at the same sites, located with GPS, and retrieved the following morning after 16-18 h of exposure, transported to an indoor facility, and immediately sorted into species and measured for length (nearest mm total length) and fresh mass (nearest 0.1 g) of every individual.Species-specific catchper-unit data was calculated for two size groups: longer and shorter than 10 cm, as number per unit effort (NPUE) and biomass per unit effort (BPUE) relative to the number of gillnets (n = 6).For monitoring the progression of fish fitness in the course of the thinning, assuming that fewer fish would decrease food competition, the condition of all roach and perch caught during the fish surveys was examined by calculating Fulton's condition factor K, K = 100 • W/L 3 , with L being the total length of the fish (cm), W being the fresh mass of the fish (g) (Schreck and Moyle, 1990).The progressions of roach and perch fitness in response to whether the thinning improved food availability were tested by a paired, two-tailed t-test on fish caught during the first and final standardized fish surveys.
Fish biomass estimation -In order to acquire the most reliable estimate of the total fish community, we applied several estimation methods to assess the densities of the respective fish species.For the species caught during the fish surveys (perch, roach, ruffe, sunbleak, and rudd), we utilized the species' BPUEs obtained during the fish surveys conducted in the first two years together with the continually removed quantity of fish biomass to generate linear regression models that described the decline in the species' biomasses over time and identify upper and lower confidence intervals.The R packages 'Tidyverse' and 'FSA' (Fish Stock Assessment) (Ogle, 2016) were used.
Large bream and tench cannot be estimated from potential catches in the gillnets used in standardized fish surveys, as they are rarely entangled in relatively small mesh sizes.Thus, we estimated the reduction of tench (>30 cm) and large bream (>20 cm) via the data of the tench and bream that were caught in baited fish traps during the spring of 2019, where the catch rates of both species were high.
Though some small pike were caught during the fish surveys, pike is not ideally estimated with gillnets.Pike densities should instead be assessed by angling (capture-recapture), which was not conducted in this study; thus, we acknowledge, that our estimation of the pike is underestimated.
Knowing that carp roam the lake, we developed estimates of their stock composition and density in 2019 based on information on the carp stock provided by the fishery manager and long-term owner of the fishing right, who shared knowledge of the historical stocking numbers.
No new carp were added during the biomanipulation period, nor was the recruitment of carp fry produced by the present stock presumed, as though carp may produce offspring in Danish lakes, successful recruitment of fry has only been confirmed within a few shallow ponds due to insufficiently high water temperatures during the spawning period and predation by other fish species (Carl and Rask Møller, 2012).As the carp stock mainly contained mirror carp (which have unique scales in mosaic patterns and thus are recognizable in photos), reporting on carp recaptures done by skilled carp fishers who have been fishing in Lake Bromme for years was also taken into consideration for the carp estimate.This estimate was weighted alongside minimum estimates obtained by visual assessments of sun-basking carp in the sun-exposed regions with shallow water observed during several fishing trips in year one.We thereby estimated the presence of a minimum of 60 fish of 6.5 kg (personal comment H. Carl).Removal of carp was avoided in 2019 and 2020, to preserve the lakes' recreational sport-fishing value, but some were removed in April 2021 by angling.
After we discovered European eels during our fishing in the spring of 2019, we conducted a capture-recapture study to estimate the population size and biomass, thereby gaining a better understanding of the total fish community structure.The eels were marked using Northwest Marine Technology, Inc. Visible Implant Elastomer system (NMT, Inc. (Northwest Marine Technology, 2017)) between September 5, 2019, and June 16, 2020.In this period (226 days), we captured 24 eels in fykes, baited fish traps, and the intermediate bulk container, which were all marked and weighted.During the period, we recaptured four marked eels (Supplementary Table 1).By relying on visual recognition of injected subdermal colored elastomer tags, the eel population was estimated according to Schnabels' mark-recapture model (Schnabel, 1938).
Catches of crucian carp only constituted two fish, therefore, no estimates of the population's density could be calculated.
Water chemistry -Water chemistry was measured almost weekly.Chlorophyll a (μg L − 1 ) was measured as a proxy for the pelagic phytoplankton biomass and was measured immediately on duplicate water samples filtered through GF/F filters (Whatman, UK) and subsequently extracted in ethanol and measured spectrophotometrically according to Jespersen et al. (1987).Alkalinity and initial pH were measured on unfiltered water by acidimetric Gran titration (Gran, 1952).Samples for later nutrient analyses were stored at − 18 • C. Duplicate water samples for total nitrogen (TN, μg N L − 1 ) and total P (TP, μg P L − 1 ) were acid-boiled in an autoclave and measured on a SKALAR autosampler.
Soluble Reactive Phosphate (SRP, μg P L − 1 ) was measured on GF/F filtered water samples using the molybdenum-blue method and spectrophotometric measurements (Grasshoff et al., 1999).
Suspended particulate organic carbon (POC, mg C L − 1 ) was measured using the procedure introduced by Kragh and Søndergaard (2004), with small modifications.Duplicate water samples were filtered onto combusted GF/F filters.The filters were dried using compressed air and stored in separate wells in a closed plastic container at − 18 • C until analysis.For analysis, the organic particles on the filters were combusted at 650 • C, and the developed CO 2 was measured on an IRGA (ADC225-MK3, ADC-gas).Dissolved glucose added to filters served as standards for reference.
We calculated the relative organic phytoplankton carbon biomass, assuming a carbon/chlorophyll a ratio of 50 % (Riemann et al., 1989) and a maximum value of 100 % to assess the contribution (percentage) of algae biomass to the POC (mg C L − 1 ) level.Water clarity was measured as Secchi depth, and the water light attenuation coefficients, Kd, were estimated as Kd = 2.303/Secchi depth (m − 1 ), following Kirk (1994).Light absorbance of colored dissolved organic matter (CDOM) was measured in a scanning spectrophotometer (Thermo Scientific Gen10S UV-Vis P) across the 400-700 nm spectrum and converted to the mean light absorption coefficient (m − 1 ) (Madsen-Østerbye et al., 2018).Continuous data loggers, installed on a pole in the northern part of the lake, measured water temperature (HOBO UA-002-64) and photosynthetic active radiation light (PAR; Odyssey Submersible Photosynthetic Active Radiation Logger, New Zealand) once and four times per hour, respectively.Light sensors were cleaned every hour with automatic hydro-wipers (Zebra-Tech Dataflow Odyssey PAR Sensor Hydro-wiper, New Zealand).The loggers were removed during the cold period between 2020 and 2021 to avoid the risk of damaging them from ice potentially covering the water's surface.As Secchi depths were measured for a longer part of the year than the loggers remained in the lake, we calculated the light attenuation as formerly described according to Kirk (1994), while using the PAR data to validate the Secchi measurements.Previous levels of water physicochemical variables estimated before the initiation of this biomanipulation project (2019)(2020)(2021), and data on the upper sediments' (0-10 cm) content of total P, and iron (Fe) were retrieved from the National Danish database (The-Danish-Environmental-Protection-Agency, 2023).
In-vitro lake water clearance rate assessment -To assess Lake Bromme's water clearance rate/progression during controlled conditions without the influence of wind-induced particle resuspension, the addition of allochthonous materials led from the catchment to the lake during precipitation and particle resuspension maintained by benthic feeding fish, we sampled lake water and surface sediment (approximately the upper 0-10 cm) from the northern littoral zone on a day during summer with a Secchi depth of 47.5 cm.In the lab, the raw sediment (including present benthos) and the lake water were added to an aquarium with approximately 6 cm of sediment depth.The aquarium was placed in a temperature and circadian rhythm light-controlled room at 20 C o and was monitored daily during the following weeks.
Sonar analysis -Sonar analysis was conducted each year during summer by boat using Lowrance HDS-12 Gen3 and HDS-16 Live equipped with a Lowrance Hybrid Dual Imaging (HDI) Skimmer, Lowrance Active Imaging 3-in-1, and an AIRMAR TM150M Transducer as described in Kragh et al. (2017).The analyses provided data on lake bathymetry, sediment properties, and macrophyte coverage (Baastrup-Spohr et al., 2016).Historical data records of the macrophyte cover were retrieved from the National Danish database (The-Danish-Environmental-Protection-Agency, 2023).
Water quality drivers -In order to examine the relative importance of local climatic conditions upon water turbidity (POC mg C L − 1 ), precipitation (mm) and wind direction and speed (m s − 1 ) were retrieved from the Danish Meteorological Institute.The effect of wave disturbance on sediment resuspension was assessed with a MIKE 21 hydrodynamic model (Danish Hydraulic Institute, Denmark, (Warren and Bach, 1992).The model uses water depth, wind speed, and direction to calculate the sediment shear stress, expressed as the wave disturbance (Newton, N m − 2 ) per square meter of sediment surface (m − 2 ).The lake was split into an ungritted mesh of 1151 triangles ranging in size from 0.9 to 100 m 2 , with each triangle assigned the average depth within that triangle.Water depths accorded with the bathymetry map (Fig. 1).Daily average wind speed and direction used in the model were collected from a nearby weather station in Flakkebjerg, Denmark, located 20 km away from Lake Bromme (Institute, 2023).The effect of the sediment shear stress on water POC concentrations was tested in two linear models, using sediment shear stress on the day of the POC measurement, and the day before, as independent variables, respectively.A conceptual animation sequence visualizing the spatiotemporal patterns of the sediment shear stress across the whole lakebed throughout the study period (884 days) was produced using MIKE 21.
We analyzed the relationship between water temperature and POC concentrations and interpreted the data in accordance with the contemporary composition of the fish community to assess the potential regulatory effect of fish-mediated sediment suspension as a water temperature-driven causal factor on turbidity.
Among benthivores, we estimated the stock of large bream before the thinning to be 165 individuals (between 147 and 182 95 % CI) and kg biomass in total, 16.0 kg ha − 1 (between 185 and 215 kg in total 95 % CI) (Supplementary Table 3).We removed 18 kg ha − 1 (225 kg in total) and 160 large bream.The start estimate of tench was 38.9 kg in total, 3.1 kg ha − 1 (between 38.5 and 44.9 kg in total 95 % CI), and 32 individuals (between 31 and 40 95 % CI).We removed 38 tench with a compiled biomass of 48 kg.We removed 14.8 kg ha − 1 large bream and 3.1 kg ha − 1 kg tench in April and May 2019, but the catches drastically decreased after mid-June.In the first half of 2020, the umbrella fish traps and fykes entrapped an additional 2.3 kg ha − 1 bream and 0.8 kg ha − 1 tench.A minor quantity of small bream and a few young tench were removed by gillnets when these replaced umbrella fish traps in mid-June 2020, and during 2021, 0.8 kg ha − 1 small bream was caught, but no tench.Carp avoided traps and gillnets, but seven (3.8 kg ha − 1 ) were removed by angling in March 2021.Thus, compared to the initial stock estimate (60 carp of 6.5 kg) being 31 kg ha − 1 in 2019, this declined to 27 kg ha − 1 due to angling.Only two crucian carp (0.2 kg ha − 1 ) were caught during all three years, thus no population estimates exist.
The BPUEs of the species caught during the six fish surveys are presented in Fig. 3, and the species' NPUEs and BPUEs with 95 % CI are shown in supplementary Table 4.We estimated the initial total roach biomass to be 636 kg, 50 kg ha − 1 , which dominated the relative catch composition in the first years (Fig. 3).We removed 9.2 kg ha − 1 roach in April 2019, but by the end of the month, the catch efficiency markedly decreased.In 2020, gillnets caught large quantities (40.4 kg ha − 1 ).The roach stock was further reduced in 2021 by 4.8 kg ha − 1 .In total, we removed 681 kg of roach (54.4 kg ha − 1 ).Our estimates indicated an initial total stock of perch of 96 kg, 7.7 kg perch ha − 1 (average 9.5 cm) was removed in 2019.Additional 0.5 kg ha − 1 and 0.8 kg ha − 1 were later removed in 2020 and 2021.The perch population was reduced by kg in total, 8.6 kg ha − 1 , thus exceeding the initial estimate.While being relatively sparse in the fish surveys the first two years, perch biomass dominated the catch composition by the end of the biomanipulation.We did not observe any increased recruitment of large perch >10 cm over the period.Rudd, ruffe, and sunbleak were generally scarce in both the fish surveys and regular fishing all years and constituted a compiled removed biomass of 41 kg ('Others' in Fig. 3).
A few pike (1.2 kg ha − 1 ) were caught during the regular fishing each year but were absent in most fish surveys, thus no population estimate was calculated.
Eel was highly abundant in the three-dimensional traps in spring 2019 and frequently detected in 2020 and 2021, displaying their characteristic 'knotting' of the gillnet mesh and biting of entangled fish.Between September 5, 2019, and June 16, 2020, we marked a total of eels.Four of these eels were recaptured (Supplementary Table 1), providing a stock estimate of 72 eel (between 36 and 291 95 % CI).The average weight was 0.420 kg (between 0.960 and 0.140 kg).The estimated eel biomass was small, 30 kg in total (between 15 and 121 kg % CI), 2.4 kg ha − 1 , according to capture, tagging, and recaptures.
All years, we observed higher BPUEs in the early summer compared to later in the summer (Fig. 3).We found BPUEs of 2.8 kg in August 2019, decreasing to 0.22 kg by September 2021.In summary, the estimated total fish biomass before the thinning was initiated was 1.404 kg (between 1163 and 1645 kg) (112 kg ha − 1 ).By removal of 1.169 kg (93.3 kg ha − 1 ), the fish density was reduced from an estimated 112 kg ha − 1 to 19 kg ha − 1 by the end of the biomanipulation, equivalent to a 6fold decrease.
Fish fitness -No significant changes in the fitness of perch (t-test, p = E. Polauke et al. 0.31) and roach (p = 0.18) were observed between the first and last of the fish surveys (see bar plots in supplementary Figure 1).
Lake water clearance rate assessment -During the first days of inspecting the water turbidity in the aquarium, the water was highly unclear and clouded.The following week, the water in the aquarium got gradually clearer, and on day 14, it was clear, resembling the in-lake conditions during the cold period where the Secchi depths were >2 m (Fig. 4, left) and light attenuation (Kd) was low at 0.96-1.04m -1 (Fig. 4, right).
Water chemistry -The TP concentrations remained stable during the three years of fish thinning, ranging between 45 and 47 μg P L − 1 , which resembled concentrations recorded in previous monitoring years (Fig. 5A, Table 2) (The-Danish-Environmental-Protection-Agency, 2023).SRP concentrations were low during all three years (4-9 μg P L − 1 ) (Fig. 5B-Table 2), with frequent peaks during the summer of 2021.Mean summer TN concentrations ranged from 1038 to 1328 μg N L − 1 , with a

Table 1
The yearly removed biomasses (kg ha − 1 ) for the eleven fish species thinned in 2019, 2020, and 2021, listed from highest to lowest.Umbrella fish traps, eel-fykes, and gillnets regularly entrapped eel, but all were returned alive.decreasing trend in the course of the fish removal (Fig. 5C-Table 2).Alkalinity and pH showed little seasonal variability and no significant changes over the three years (Fig. 5D-Table 2).Chlorophyll a concentrations ranged between 13 and 30 μg L − 1 and had short peaks in late summer and late winter (Fig. 5E.Table 2).POC concentrations were higher during summer (22-26 mg C L − 1 ) than in winter (16-21 mg C L − 1 ) (Fig. 5F-Table 2).CDOM correlated with POC variations (May-Sep: R 2 = 0.20, Oct-Apr: R 2 = 0.25, entire period: 0.03), whereas chlorophyll correlated with the POC variations only during the winter (May-Sep: R 2 = 0.002, Oct-Apr: R 2 = 0.40, entire period: 0.003) (Supplementary Figure 3).Though Secchi depths correlated better with chlorophyll than POC, the algae biomass constituted very little of the POC, between 4 and 6 % during the winter and summer periods, respectively, when assuming an algae-carbon ratio of 50 (Supplementary Table 5).
Sediment properties -The sediment density was mapped in 2019 (Fig. 1, middle).The sonar maps showed high sediment density areas being located in the littoral and western regions, and soft sediment density sites being located in the central and northern regions.Examination of data on the upper sediment showed a mean total P of 1.5 mg g DW − 1, and Fe content of 11 mg g DW − 1 , equivalent to a Fe:P of 7.6 (2016 (The-Danish-Environmental-Protection-Agency, 2023)).
Macrophyte cover -The macrophyte cover over the sediment was examined in July between 2019 and 2021 via sonar side scan.The interannual spatial development cover of submerged macrophytes is visualized in Fig. 1 (right).The cover of fully submerged plants increased over the three years from 0.8 % in 2019, 3.9 % in 2020, and 13.5 % in 2021.In 2019, the plants were mostly found in the northern region.Over the years, higher plant densities were observed in areas where plants had been present in the former years, and by 2021, the cover had also expanded noticeably in the southern region.The larger species, shining pondweed Potamogeton lucens, and sago pondweed Stuckenia pectinata, expanded most prominently.The cover of submerged and floating parts of yellow waterlily Nuphar lutea remained constant at 8 %.

Drivers of water quality
We assessed the effects of the thinning on the progression of the water quality by analyzing the three years' monthly mean summer (May-Sep) values of water turbidity and nutrients in relation to the monthly estimated standing fish biomass (kg ha − 1 ) for months where fishing removal and sampling of water was conducted contemporarily (Fig. 6).The Secchi depths declined as the standing fish biomass was reduced (Fig. 6A).Chlorophyll a and TP levels increased with decreasing fish biomass (Fig. 6B and C), while TN and POC levels slightly decreased (Fig. 6D and E).
To assess the influence of weather conditions on water turbidity, we analyzed precipitation, wind speed, and water temperature in relation to POC.Precipitation may transfer particles from the catchment to the lake, wind speed is a model proxy for wind-induced shear stress and particle resuspension from surface sediment, and water temperature may serve as a proxy for the feeding intensity of benthivore fish.Precipitation did not correlate with water turbidity (POC or CDOM) or nutrient levels (TN or TP) over the three years (Supplementary Table 7).
Two models were developed to assess the influence of wind-driven shear stress on suspended POC concentration, by incorporating average wind speed data retrieved for the day of POC sampling as its independent variable.Scatter plots of the two models can be found in Supplementary Figure 4. Model 1 showed a non-significant negative correlation of POC (mg C L − 1 ) with mean weighted shear stress (p = 0.49, t-value = − 0.7, estimate = − 243, R 2 = − 0.008).Model 2, which uses wind speeds for the day before sampling, indicated a significant negative relationship between increasing POC levels and weighted mean  shear stress (p = 0.01, t-value = − 2.7, estimate = − 708, R 2 = − 0.09).Thus, we regard the result as an output of other effects.The whole-lake spatiotemporal analysis of wind-induced lakebed shear stress showed that the effect of shear stress was highest during periods of high wind speeds and had the highest impact on sediment resuspension in the shallow, coastal areas, whereas the open central region was seemingly unaffected.An animation showing the bed shear stress in Lake Bromme can be found in supplementary materials for Fig. 5. Summer water temperatures were high (>20 • C) and winter temperatures low (2.1 • C Dec-Feb average, Fig. 7).The lake was ice-covered during the early months of 2021.Average daily water temperatures correlated positively with POC levels, mainly during the cold period (May-Sep: R 2 = 0.014, Oct-Apr: 0.43, Entire period: 0.30) (Supplementary Figure 6).

Discussion
Lake Bromme was continually biomanipulated over three years through successive thinning of various fish species and feeding groups to promote good water quality, more piscivorous fish, and increased cover of macrophytes.Water chemical parameters, fish quality, and macrophytes were systematically monitored throughout the biomanipulation to assess the effects of thinning the respective feeding groups on the lake ecosystem.However, the water quality in the shallow eutrophic lake was unaffected by extensive thinning of planktivorous and benthivorous fish species.Common suggestions for effectless fish-manipulation projects most often relate to complex and synergistic issues of insufficient reduction of planktivorous fish, lack of piscivorous fish, and sustained P pollution issues promoting dense algae blooms while macrophytes and densities of herbivorous zooplankton groups remain sparse (Gulati et al., 2008;George, 2021).Ambitious to explore the underlying system-ecological causalities for the unresponsive Lake Bromme, the post-biomanipulation state of the lake is discussed.
Attempting to calculate reliable estimates of fish population densities, especially in fish communities consisting of numerous species, is highly challenging due to the respective species' characteristics in morphology, behavior, and home range, as well as the lake ecosystems' size, bathymetry, and structural features (e.g., macrophytes).To mitigate these challenges to evaluate the thinning efforts, we applied several methods for estimating the respective species' initial biomasses and repeatedly monitored the populations' densities in the course of the fish thinning.
For perch and roach, we moved approximately 113 % and 107 % of the biomasses that were initially estimated.This decline was reflected in the very low BPUEs of both species, which were caught at the final fish survey.Only a few sunbleak, rudd, and ruffe, were observed and removed during the biomanipulation.We did not estimate the density of small bream, as this size group, in contrast to the indented outcome of the thinning, exhibited a stock recruitment rather than a decline during the fish surveys, likely as a result of the decimation of fish within the same feeding niche.Therefore, it was not possible to calculate an estimate based on the thinning efforts of this group.We calculated estimates of large bream and tench from the catch rates in the baited traps during year 1.Compared to our total catch of bream and tench, the removed quantities nicely matched what we initially estimated for both species (Supplementary Table 3).Though we shifted from using baited traps to gillnets, entrapment of bream and tench in the intermediate bulk container was continuously rare, thus indicating, that these stocks were efficiently reduced in year 1.Pike and crucian carp were not estimated due to low catch efficiencies.Given that crucian carp was present in large densities, based on former experiences (unpublish data), we would assume to have caught more in the baited fish traps and the intermediate bulk container.As pike was not a target species for removal, assessment of the pike population was not prioritized in this study, but population assessment should be attempted through an individual recognition study (Kristensen et al., 2020).Similarly, estimation of the carp stock should ideally be performed through a mark-recapture study to obtain reliable stock estimates.Still, as the carp were initially preserved to ensure Lake Bromme's recreational value, we did not remove carp until year three, to assess whether the removal of a few carp would promote any improvement in the water quality.
In summary, we estimated an initial fish biomass of 1.404 kg (between 1163 and 1645 kg) and removed 1.169 kg equivalent to a reduction of 83 % (between 71 and 100 %).Considering the challenges of estimating true fish stock densities, the quantities removed during the biomanipulation closely reflected what was initially estimated, thereby implying a drastic reduction in the total fish density.While upper confidence intervals for perch and roach implied, that we could have removed an additional 15 kg perch ha − 1 and 5.5 kg roach ha − 1 compared to the removed biomass, we removed above or very close to what was estimated for the remaining targeted species.This was supported by the very low catch rates of large cyprinids early in the thinning, and by the BPUEs of the smaller fish species observed in the final fish survey.We, therefore, advocate, that to acquire reliable estimates for different fish species for biomanipulation purposes, efforts should be put into both applying the ideal approach for encountering the species (baited traps, angling, etc.) and in detail noting the species' removed quantities over time, to continually assess the decline in the estimated biomass.
We hypothesized that reducing the abundance of benthivores fish would reduce the suspended particle concentration due to less sediment disturbance from fish.However, POC levels and water clarity did not decrease following the removal of bream and tench.We also hypothesized that thinning zooplanktivorous fish (roach, small bream, and small perch) would reduce predation on herbivorous zooplankton and thereby lower phytoplankton chlorophyll.However, the summer-average chlorophyll a concentration was slightly higher in 2021 than in previous years.Also, POC levels remained the same throughout the period, as did the vertical light attenuation and Secchi depth.Extensive fish thinning did not improve the lake's water quality.TN declined over the period, perhaps as a combined result of fewer fish contributing to the nutrient turnover by defecating and enhanced N uptake from the increase in macrophytes (Meijer et al. 1990(Meijer et al. , 1994)).While the TP levels fluctuated more in the first years, the summer averages remained unchanged.SRP levels were generally very low and stable in 2019-2020 but fluctuated and had higher peaks in 2021.In addition to P limitation, the relatively low chlorophyll a concentrations may have been due to light limitation by high particle levels, which thereby limited the phytoplankton production (Kragh et al., 2017).
No recreational carp fishing or stocking was performed in Lake Bromme during the biomanipulation, which otherwise, due to the addition of pre-baiting rich in nutrients like P may add to eutrophication, thus complicating the restoration efforts (Arlinghaus and Mehner, 2003;Skeate et al., 2022).Thus, no significant nutrient loadings from ground baits reached the lake during the study period.The bait added to lure the cyprinids into the baited fish traps during the biomanipulation, did not add significantly to the trophic state (approximately 148 g P added in total).No inlets continually lead water to Lake Bromme; thus, inputs of new water arrive as seepage from the catchment during periods of precipitation.We did, however, not find any relationship between water nutrients and local precipitation.While intrusion of P-rich groundwater, sewage, and wastewater has been found to corrupt lake restoration programs relying on repeated fish removal in eutrophic lakes (Peretyatko et al., 2012;Søndergaard et al., 2017), none of such pollution was assumed to affect Lake Bromme.Though the external P load was low, it may still contribute to stabilizing the eutrophic conditions.Assessment of internal P release from the surface sediments was not conducted during this study, but former examinations found low contents of P (<2 mg g DW − 1 ) in the upper 0-10 cm sediment profile.Still, as the Fe:P in the sediment was also low, the release of redox-sensitive P species bound in oxidized iron complexes adding to the internal P loading could likely be expected during prolonged periods with low oxygen saturation in the bottom water (Jensen et al., 1992;Reitzel et al., 2005).However, as we detected similar mean water P levels during the summer and winter periods, we do not assume any profound internal P loading during 2019-2021.Moreover, as the water P levels were stable during all seasons, including the warm periods where the feeding activity of the benthic-feeding fish should be highest, thus their increased disturbing of the sediments should transfer more settled particles into the water body, the large cyprinids did assumably not significantly regulate the water P dynamics within the lake.
Fish biomanipulation guidelines state that 75-80 % of the fish biomass must be removed before improved water quality will be noticeable (Perrow et al., 1997;Hansson et al., 1998).Presuming we meet this criterion, by reducing the estimated biomass between 71 and 100 %, the water quality did still not improve.Thus, in an attempt to identify reasons, we assessed possible alternative drivers of water quality.
We analyzed the relationships of POC, chlorophyll a and nutrient concentrations, and Secchi depth with precipitation, wind-induced shear stress, and water temperature (the latter as a proxy for benthivore fish activity).No correlations were found between precipitation and POC, Secchi depth, and nutrients, indicating that negligible loads of particles and particle-bound nutrients reached the lake via run-off from the catchment.The chlorophyll levels correlated better with the Secchi depths than the POC levels.However, regardless of season, the algae carbon only constituted a minor fraction of the POC, thus implying, that the algae biomass had little influence on the water's particle pressure and that the remaining carbon fractions should derive from other C-rich compounds.
In shallow water bodies, wind-induced shear stress might reduce water quality by triggering wind-wave erosion of the surface sediment and sustaining the suspension of particulate matter in the water (Martinsen et al., 2022).Fish may enhance wind-induced sediment resuspension as they feed in surface sediment: their constant transfer of settled particles to the water body could make the lakebed more sensitive to wind-wave erosion (Scheffer et al., 2003).Despite Lake Bromme's shallow water, which exposes its surface sediment to wind-generated shear stress (Martinsen et al., 2022), we found no relationship between POC and shear stress.With prevailing wind from the southwest, the fetch of the wind across the surface is likely too short to induce sufficient turbulence in the bottom water and cause sediment erosion (Supplementary Figure 5).
So why does Lake Bromme fall into the category of lakes that are unresponsive to fish biomanipulation?In a comprehensive systematic review regarding the effects of reducing planktivorous and benthivorous fish on the water quality in eutrophic lakes, Bernes et al. (2015) argue that most unresponsive lakes exhibit high intervention strength from a combination of lake physicochemical properties and the intensity of the applied fishing effort.Common characteristics of these lakes are large surface areas, high water retention, low TP values before fish reduction, and low fishing intensity.Bernes also argues that small lakes with initial high chlorophyll a and TP levels are the most responsive to biomanipulation.Meanwhile, a prerequisite for successful restoration programs has been suggested at a summer water TP of <80-150 μg L − 1 (Jeppesen et al., 1990), which Lake Bromme is far below.Lake Bromme is small and eutrophic, has experienced intense fishing pressure through a 6-fold thinning of total estimated fish biomass to just 19 kg ha − 1 , and with mean summer water P values < 50 μg L − 1 , thus was expected to fall intoor close tothe category of expectedly responsive lakes.Nevertheless, Lake Bromme remained unresponsive.
Alternative causes for Lake Bromme's unresponsiveness could involve factors and mechanisms that are unaffected by the species of fish we manipulated, and the fish species populations which remained relatively untouched.We identified extensive temporal variation in both Secchi depth and POC concentrations.There was a substantial decrease in Secchi depth when POC concentrations were higher, presumably independent of chlorophyll a concentration, which only constituted a very small fraction of the suspended particulate organic carbon (supplementary Table 5).As the highest peaks in chlorophyll a were observed in the late summer, the algae biomass was likely dominated by cyanobacteria, which are often inedible to the herbivorous zooplankton.However, no examinations of the relative densities or species compositions of the phytoplankton or zooplankton were conducted in this study.
Secchi depths in late winter were 3-4 times higher than during summer, where the shallower Secchi depths coincided with higher summer water temperatures, while wind impact was not significant.We propose that these temporal changes in water clarity are driven primarily by an increase in the biological activity of fish feeding on bottom invertebrates, leading to the resuspension of sediment particles (Meijer et al., 1990).It is known that large, bottom-feeding cyprinids can cause intensive sediment resuspension (Breukelaar et al., 1994;Huser et al., 2022).The stocks of large bream and tench were presumably drastically thinned, but the carp were not, thus the carp stock would likely continue to transport large loads of settled particles into the water column, thus keeping the water quality low.
Prolonged periods during summer with low winds and high water temperatures did not yield clear water.Conversely, the highest Secchi depths >2 m occurred during the windy and cold periods.Our in-vitro water clearance assessment without fish present showed that the water cleared after two weeks.While finding no correlations between POC levels and precipitation, and the wind models implied that wind did E. Polauke et al. not influence POC, we found some seasonal correlations to water temperature.Given that the thermal requirements for carp to initiate and intensify their foraging activities are water temperatures above 12 C o (Elliott, 1981), this reflects our findings by around this temperature measured lower Secchi depths during the summer period, when the carp are hyperactive and feeding.Likewise, we measured higher Secchi depths during late winter, when the overwintering carp were likely to be dormant, thus supporting our assumption that the turbidity in Lake Bromme is primarily driven by the biological activities of carp causing sediment resuspension when searching for benthos (Su et al., 2023).We estimated that carp density declined only from 31 to 27 kg ha − 1 .Though we thinned bream and tench efficiently, the remaining carp feeding near the lakebed in the warm period might continue to resuspend sediment particles, thus maintaining the turbid water.The ecological impacts of benthivore fish are highly dependent on a lake basin's morphology and its sediment's biogeochemical properties, such as substrate types, prey abundance, and plant cover (Meijer et al., 1990;Lougheed et al., 1998).Assuming our initial estimate of 60 carp was realistic, the removal of the seven carp left the stock 88 % intact (27 kg ha − 1 or 4.2 medium-sized carp ha − 1 ).Given that the stocks of other large benthivorous fish were removed, this would imply that even a very low density of carp inhabiting food-depleted lakes with loose surface sediments is enough to prevent improving water quality.However, as many lakes are valued for their recreational and socio-economical aspects related to sports fishing for carp, maintaining carp conflicts with the ambitions of establishing good water quality, which therefore should be considered when selecting the lakes for fish manipulation (Arlinghaus and Mehner, 2003;Skeate et al., 2022).While stocking planktonic filter-feeding carp sp.into eutrophic Chinese freshwater lakes has shown potential as a nontraditional biomanipulation tool for combatting algae blooms, releasing such non-native cyprinids to enhance bottom-up biocontrol is prohibited in Danish lake management (Chen et al., 2023).Identifying and assessing the effects of benthivorous carp and other sediment feeders in regard to water turbidity and nutrient turnover dynamics in eutrophic lakes would be a valuable future study (Su et al., 2023).Uncovering a possible impact of carp, as well as other large benthivorous cyprinids, on turbidity requires continued monitoring of physicochemical variables after removing the carp stock.
We hypothesized that increased macrophyte cover should follow the fish reduction as a result of improved underwater light conditions due to decreased sediment resuspension from fewer fish disturbing the sediment and uprooting the macrophytes (Meijer et al., 1990).Sonar side-scans revealed that the plant cover had increased from 0.8 % in 2019 to 13.5 % in 2021.Larger submerged species reaching from the lakebed to the water surface exhibited the highest increase in cover, as the poor light conditions would not be a critically limiting factor near the surface water.Many environmental features regulate the macrophyte communities in lakes, including water depth, slope, and sediment characteristics and textures, which restrict the areas being colonizable, and the degree of physical disturbance and herbivores grazing (Lauridsen et al., 2003).The increase in plant cover was highest in areas with high sediment densities, where plants were already present (Fig. 1), thus indicating new shoots were more likely derived from existing plants than from seeds.Lake Bromme is generally shallow, with the deepest site <2.5 m, meaning, that in a clear-water state with good light conditions, the total lakebed could likely be densely colonized.
As water clarity did not improve, and could therefore not be assumed to be the key promotor of the macrophyte's growth, we propose that less feeding and disturbance induced by the reduced stocks of bream and tench on the submerged macrophytes may be responsible for the increased plant cover (Lammens et al., 2002).Though rudd is known for consuming macrophytes, the abundance of rudd was probably too low to have an influence.Meanwhile, due to insufficient animal prey, roach might have been grazing on the plants (Horppila, 1994).We thus submit that roach reduction in 2020 and 2021, in combination with the continued vacancy of large bream and tench during 2019-2020, may have stimulated the increase in plant cover from a decreased grazing pressure.
Both the planktivorous and benthivorous fish stocks were drastically reduced.However, no increase in the abundance of piscivorous fish species was detected post-biomanipulation. Thus, a natural bottleneck limiting recruitment of cyprinid fry was likely not established, therefore, the prospects of reaching better water quality or maintaining the positive development in the macrophytes by ensuring high predation pressure on future cohorts of cyprinids, are likely limited.Continued measurements in Lake Bromme will enable us to assess how the macrophytes and the fish community respond to the ongoing fish manipulations and whether further expansion of the plant cover may eventually significantly enhance particle settling and turn the turbid lake into a clear-water lake.This should be assisted by detailed characterization of the optical properties (e.g., complexes of resuspended biological components, mineral and humic colloids, and dissolved organic material) that collectively influence the lake water's light attenuation and Secchi depth.

Funding
We thank Aage V. Jensen's Nature Foundation and Poul Due Jensen's Foundation for grants supporting the Ph.D. research program of EP and facilitating the research and monitoring activities during the biomanipulation at Lake Bromme.The fish thinning program and parts of the lake survey program at Lake Bromme were financed by the Danish Environmental Protection Agency under the River Basin Management Plans 2015-2021.

Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Fig. 1 .
Fig. 1.Map of Denmark with the location of Lake Bromme shown with a red dot (55 • 28′52"N, 11 • 30′54"E).Map A illustrates water depth (0.5 m intervals) produced from sonar mapping (2019).A sediment density map (B) was produced from sonar analysis (2019) showing areas with soft-density sediments (brown and orange colors) and hard-density sediments (turquoise and light blue colors).The outer blue contour illustrates areas covered in reeds or too shallow (0.5 m) to be mapped by the transducer.The inter-annual progression in the cover of submerged macrophytes between the summer of 2019 (light green) and 2021 (dark green) mapped from sonar analysis are shown in map C. PLEASE USE COLOR FOR PRINT.(For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 2 .Fig. 3 .
Fig. 2. Chronological scheme illustrating monthly removed species-specific fish biomasses (kg ha − 1 ) that were caught during the successive fishing program conducted between April 2019 to September 2021.'Others' include the biomass of rudd, ruffe, crucian carp, and sunbleak.PLEASE USE COLOR FOR PRINT.(Forinterpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 4 .
Fig. 4. Secchi depth (cm) during 2019-2021.The white area indicates summer measurements (May-September) and the grey areas indicate winter measurements (October-April) (A).Secchi depth (m) and vertical light attenuation coefficients, Kd, throughout the biomanipulation period (B).PLEASE USE COLOR FOR PRINT.(For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 5 .
Fig. 5. Water chemical variables: Total P (A), SRP (B), Total N and monthly removed and estimated standing fish biomass (kg ha − 1 ) (C), pH and alkalinity (D), chlorophyll a (E), POC and CDOM (F) assessed throughout the biomanipulation period.PLEASE USE COLOR FOR PRINT.(For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)

Fig. 6 .Fig. 7 .
Fig. 6.Plots of monthly mean summer values on water chemical parameters measured in the course of the biomanipulation and the decline in the estimated standing fish biomass: Secchi depth (A), Chlorophyll a (B), TP (C), TN (D), and POC (E).PLEASE USE COLOR FOR PRINT.(For interpretation of the references to color in this figure legend, the reader is referred to the Web version of this article.)