Epipelon biomass responses to different restoration techniques in a eutrophic environment

Eutrophication is a worldwide problem. In eutrophic lakes, phosphorus release from stored sediment hinders restoration processes. The epipelon is a community that grows attached to the sediment surface and has the potential to help phosphorus retention by autotrophic organisms. This study evaluated epipelon responses to four lake restoration techniques. The responses of abiotic variables and phytoplankton biomass were also evaluated. Four simultaneous mesocosm experiments were performed in a shallow eutrophic lake. The applied techniques were aeration, flocculant, floating macrophytes, and periphyton bioreactor. Water and epipelon samples were taken on days 3, 10, 17, 27, and 60. The aeration treatment and macrophytes decreased light availability in the epipelon, which had a predominance of heterotrophic components. Flocculant and periphyton bioreactor treatments favored epipelon growth with a higher contribution of autotrophic components. Therefore, some techniques may favor the epipelon growth, while others may harm the community, resulting in less efficient restoration processes. For the complete restoration of a lacustrine ecosystem, the choice of techniques to be applied must consider the restoration and maintenance of the benthic environment.


Introduction
Many studies show that nutrient overload triggers eutrophication in aquatic environments. Intense phytoplankton blooms, increased turbidity, the disappearance of submerged macrophytes, changes in the food chain, and biodiversity loss are consequences of eutrophication (Søndergaard et al., 2003;Schindler et al., 2016;Moal et al., 2019). In lakes, eutrophication commonly leads to increased phytoplankton biomass, which affects ecosystem functioning (Scheffer et al., 2001).
Most of the world´s lakes and reservoirs are small and shallow (Downing et al., 2006), where the interaction between sediment and water column is more intense than in deep lakes (Søndergaard et al., 2003). For example, in eutrophic shallow lakes, the phosphorus (P) released from the sediment into the water column under anoxic conditions maintains eutrophication (Bicudo et al., 2007). Internal P loading is one of the main problems in the restoration of shallow eutrophic lakes (Dittrich et al., 2011). Studies indicate that the sediment P release can last for about 10, 20, or more years after the elimination of the P input (Jeppesen et al., 2005). Although restoration successes have been reported, some shallow temperate lakes have returned to the eutrophic condition after approximately 10 years or less due to self-fertilization . Considering that the P stored in the sediment, usually for years, can interfere with the restoration process, benthic communities can assume a crucial role in P immobilization.
Among the benthic communities, the epipelon is composed of autotrophic and heterotrophic organisms and detritus that grows attached to the surface sediment (Wetzel, 1983). The epipelon has morphological and physiological features that can help recover eutrophic systems through the immobilization of P in the sediment (Dodds, 2003;Genkai-Kato et al., 2012). Epipelic algae assimilate and adsorb P acting on P cycling in the system and contribute to the oxygenation of the sediment/water column interface via photosynthesis (Dodds, 2003;Liboriussen and Jeppesen, 2006). The photosynthesis of epipelic algae can cause the supersaturation of oxygen in the sediment, contributing to P immobilization (Dodds, 2003). In this sense, the growth of the phototrophic epipelon can contribute to sediment oxygenation, minimizing self-fertilization during and after the restoration processes of shallow eutrophic lacustrine systems. However, the responses of the epipelon in the processes to reverse eutrophication are still poorly investigated worldwide (Amaral et al., 2020).
In general, the first phase of the restoration process is the reduction and/or elimination of nutrient input from point and diffuse allochthonous sources of anthropic origin (Janssen et al., 2019). Then, some traditional restoration techniques are commonly applied, such as: (i) sediment dredging, removal of nutrient-rich soil layer (Zhang et al., 2010); (ii) use of aluminum, which acts in phytoplankton flocculation ); (iii) hypolimnion aeration/oxygenation, which increases oxygen in the sediment and acts on P retention (Gerling et al., 2014); (iv) flushing, which decreases the water residence time and replaces it with nutrient-poor water, with the aim of increasing water flow, carrying the phytoplankton biomass out of the lake (Janssen et al., 2019). In contrast, green or eco-friendly technologies are technologies that have a less adverse impact on the environment, for example, the use of clean energy, organisms from the system itself, and biological preparations (Simon and Joshi, 2021). Among the techniques considered green technologies, there are: (i) biomanipulation, which alters the trophic chain usually with the removal or introduction of fish ; (ii) flocculation and sediment cover, which uses clays with or without submerged macrophyte seeds (Pan et al., 2011); (iii) submerged macrophytes, which can eliminate phytoplankton dominance, reestablish the clearwater phase and enhance sediment stabilization (Moss, 1990); (iv) floating macrophytes, which can compete for nutrients with phytoplankton (Yan et al., 2016); (v) periphyton bioreactors that use the periphytic community to remove P from water (Ko et al., 2019) or the use of artificial rivers Algal Turf Scrubbers -ATS (Mayr et al., 2014).
Studies demonstrate that the epipelon can immobilize phosphorus in the sediment and therefore minimize the return of P to the water column, contributing to the success of restoration processes (Dodds, 2003;Liboriussen and Jeppesen 2006;Genkai-Kato et al., 2012). This study aimed to evaluate the epipelon responses to different techniques for restoring eutrophic environments. Considering that the eutrophic lake restoration techniques can reduce phytoplankton biomass and increase light availability in the benthic environment (Scheffer et al., 2001;Grochowska et al., 2017), we expect that applying techniques based on green technologies would favor epipelon growth. Specifically, we hypothesized that phytoplankton bloom would decrease and the increased light resulting from the applied restoration techniques would favor epipelon growth, particularly in the algal community.

Experimental Design
The mesocosm experiment was performed in the littoral zone of a eutrophic reservoir. The treatments were performed simultaneously in triplicates: control (C); aeration (A); natural flocculant (F); macrophytes (M); and periphyton bioreactor (PB). The water and epipelon samplings were carried out on days 3, 10, 17, 27, and 60. In the experimental period (December/2019 to January/2020), we performed five samplings and collected 75 abiotic and biotic samples.
Based on the bathymetric map, 15 mesocosms were positioned at an average distance of 10 meters from the shore to avoid the entry of allochthonous matter. The depth was standardized at 1 m and the distance between the mesocosms was 40 cm to minimize contamination. The positioning of the mesocosms was decided to maintain similarity in-depth, light availability, and minimize contamination between them. The open-bottom mesocosms consisted of polypropylene cylinders of 120 cm in diameter, 150 cm in height, and 08 mm in thickness. The mesocosms were buried in the sediment up to the clayey part, retaining water inside (Fig. 1B). After installation (11/21/2019), we standardized the water column depth of the mesocosms at 1 meter and the volume at approximately 1130 l. After seven days of acclimatization, treatments with different restoration techniques were randomly distributed and installed (T0; 29/11/2019). The experiments were carried out in summer (rainy season) when there is an intense phytoplankton bloom (Crossetti et al., 2018). Considering the short life cycle and the rapid response of algae to environmental changes, sampling was carried out weekly in the first month. To cover the two hottest summer months (December and January), another sampling was carried out over 60 days.
We selected restoration techniques classified as green technologies, which interfere less with ecosystem functioning than other technologies, such as inorganic chemicals (Zhang et al., 2015). In the control treatment, no technique was applied. In the aeration treatment (A), compressed air was transported through compressors to the mesocosms and released as small bubbles through a ring-shaped porous hose placed 40 cm from the sediment to minimize resuspension. The electrical energy used to operate the compressors (BOYU air compressor S-2000A 127 V pressure 0.012 MPa) came from an off-grid solar energy system (455 W h/day), which ensured the constant and continuous operation of the compressors with clean energy (Fig. 1C) and without interruptions during the trial period. For the flocculant treatment (F), we used a biodegradable product (Tanfloc SG). This product is an organic salt of the extract (powder) of the Black Acacia tree -Acacia mearnsii de Wild., which is a tannin-based coagulant (Barrado-Moreno et al., 2016), provided by TANAC S.A. (Montenegro, Brazil). After previous tests in the laboratory using water from the Garças reservoir, we applied the flocculant at a concentration of 50 mg L-¹. To minimize the effect on the epipelon, the flocculant was added in two steps on days 0 and 13 of the experimental period. In treatment M, 15 individuals of Eichhornia crassipes (Mart.) Solms, a dominant species in the study area, were introduced. For standardization, non-senescent adult macrophytes with four leaves of similar size and roots, covering 33% of the mesocosm surface, were used. During management, senescent leaves were removed on experimental days 13, 27, and 47 (12/12, 12/26, and 1/15). The epiphyton was not excluded. In the periphyton bioreactor (PB) treatment, 60 glass slides (25 cm×10 cm x 0.3 cm) were inserted vertically into the mesocosms. The slides were positioned submerged at 20 cm and 80 cm from the surface to homogenize the removal of nutrients in the water column. The slides were removed, washed, and replaced on days 20 and 40 of the experimental period (12/19 and 01/08) to avoid the periphyton detachment phase. The PB management interval was based on the lake´s periphyton biomass peak (Borduqui and Ferragut, 2012). Images and other details about the experiment are available on the Experimental Project Garças page (garcasproject.blogspot.com). Subsurface water was collected manually using polyethylene bottles (1 L) and transported to the laboratory under refrigeration and in the dark. The surface sediment sampling (1 cm) was collected using a manual corer sampler (5 cm in diameter PVC tube). Each treatment used a different sampling tube to avoid contamination. The sampling location was randomly distributed, avoiding sampling near the mesocosm walls. After each sediment collection, the sampled location was marked to avoid resampling.

Environmental Variables Analyzed
The underwater radiation was measured in the field (photosynthetically active radiation -Li-Cor LI-250A lux meter). The light attenuation coefficient (%) was calculated using an equation: 100 (L 0 -L 2 )/L 0 , where L 0 is the light at the upper layer and L 2 is the light in the lower layer (Wetzel and Likens, 1991). The light was measured as close as possible to the center of the mesocosm, without the influence of the wall and standardized place.
In the lab, total nitrogen (TN) and total phosphorus (TP) (alkali persulfate method) were determined according to APHA (2012). For particulate matter (PM) measurement, water samples were filtered through a pre-calcined glass fiber filter (Whatman GF/C), stored at 100°C until use, and weighed (APHA 2012). Macrophyte cover was estimated with the square method using the relative abundance of macrophytes (Thomaz and Bini, 2003). Macrophyte cover was measured in the experiment´s beginning (0d) and end (60d).

Phytoplankton
Water samples were preserved with 4% formalin for phytoplankton species identification. For quantification, the samples were fixed with acetic Lugol and analyzed under an inverted microscope (Zeiss Axio Observer D1; 400x) according to the Utermöhl method. The criterium used for counting was a rarefaction curve with a counting efficiency > 95% and reaching 100 individuals of the most abundant species. The mean algal biovolume was obtained from the local literature (Fonseca et al., 2014) and for species not mentioned, we calculated based on Hillebrand et al. (1999). The biovolume was expressed as biomass (mg L -¹).
The phytoplankton chlorophyll-a concentration (corrected for pheophytin) was obtained from samples filtered through a glass fiber filter (Whatman GF/F) and extracted using ethanol (90%) for 24 h in the dark (Sartory and Grobbelaar, 1984).

Epipelon
The sediment sample was diluted in a known volume of distilled water to determine the ash-free dry mass (AFDM) and chlorophyll-a concentration. For AFDM measurement, epipelon samples were filtered through a pre-calcined glass fiber filter (Whatman GF/C) and stored at 100°C until constant weight. Subsequently, the filters were calcined (500°C, 1 h) and weighed (APHA, 2012). The chlorophyll-a concentration in epipelon was measured by the same method used for phytoplankton chlorophyll-a (Sartory and Grobbelaar, 1984). AFDM: Chlorophyll-a ratio was used to assess changes in autotrophic and heterotrophic periphyton components.

Trophic State Index (TSI)
Trophic State Index (TSI) was calculated according to Lamparelli (2004) for tropical reservoirs, using chlorophyll-a (Chloa) and total phosphorus (

Statistical Treatment
To identify significant differences in epipelon biomass and abiotic variables between the control and each treatment with different techniques, we used repeated measures analysis of variance (two-way RM-ANOVA -treatment and time). A significance level of <0.05 was adopted using Tukey's a posteriori test. Data were logarithmized to meet the assumptions of normality and homogeneity when necessary. The software used for analysis was SigmaPlot 12.0. For water quality assessment, we conduct the two-factor repeated measures ANOVA (two-way RM-ANOVA; treatment and time factors) with axis 1 scores of each treatment resulting from the principal component analysis (PCA) of the abiotic variables (Light attenuation, TP, TN, and PM). We adopted a significance level of <0.05 and Tukey's a posterior test. To evaluate the effects of the techniques on the epipelon biomass, the permutational multivariate analysis of variance (two-way PERMANOVA) was used. In this analysis, the data were logarithmized using the Bray-Curtis similarity measure with 9999 permutations in the Past 4.01 statistical software.

Aeration Experiment
Compared to the control, A treatment showed a mean increase in light attenuation and TN concentration of 7.4 and 6%, respectively, during the experimental period ( Fig. 2A  and C). The TP and PM concentrations had a mean decrease of 11.5 and 25.7%, respectively ( Fig. 2B and D). PM and TP concentrations decrease from the 10th day in A treatment. Regarding the two-way RM-ANOVA results (Table 1), the experimental days differed significantly in light attenuation coefficient and TN. The TP concentration differed between treatments and days. The PM concentration differed between days, and the interaction effect on days 30 and 60 (Tukey: p = <0.002 and <0.006, respectively).
Considering the average in the experimental period, phytoplankton biomass decreased 10.7% when compared to the control (Fig. 2E). The chlorophyll-a concentration differed only between experimental days (Table 1).
On average, the epipelon chlorophyll-a decreased by 7.8% when compared to the control, while the AFDM and Chlorophyll-a ratio increased by 10.6 and 57.9%, respectively (Fig. 2F, G, and H). The epipelon chlorophyll-a differed between days (Table 1). The effect of time on the epipelon AFDM was significant, and the interaction with the treatment effect was significant only on day 60 (Tukey: p = 0.006). No difference in epipelon biomass estimates was detected between treatments and days (two-way PERMANOVA).

Flocculant Experiment
On average in the experimental period, light attenuation, TP, TN, and PM increased 22.6, 55.5, 302.4, and 33.9%, respectively, compared to the control (Fig. 3A-D). The light attenuation coefficient differed between treatments and days (Table 1). In contrast, TN concentration differed between treatments and days with significant interaction between factors on all days (Tukey: p = <0.001). Similarly, TP concentration differed between treatments and days, and the interaction between factors was significant (Table 1). There were notable differences between TP concentration in the control and F treatment on days 3, 17, 27, and 60 (Tukey: p = <0.001; 0.034; 0.019 and <0.001; respectively). The highest decrease in PM was detected on day 3, increasing later during the experimental period. The peak of PM concentration occurred on day 60 when the biomass was four times higher than the control (Fig. 3D). PM differed between days and the interaction between factors was significant. Treatments on days 3, 17, and 60 differed significantly (Tukey: p = 0.036; 0.013 and 0.044, respectively).
Regarding the experimental period, the phytoplankton biomass had an average decrease of 28.1% in relation to the control (Fig. 3E). The interaction between treatment and days was significant for phytoplankton biomass (Table 1), and the difference between control and treatment was only significant on day 3 (Tukey: p = <0.001). We observed a 78.5% reduction in phytoplankton biomass until day 27, but there was a loss of flocculant effect due to a 159.7% increase on day 60.
The epipelon chlorophyll-a and the AFDM: Chlorophyll-a ratio increased by 26.8 and 6.3% on average, respectively (Fig. 3F, H), while AFDM decreased by an average of 24.8% (Fig. 3G). According to RM-ANOVA results, epipelon chlorophyll-a and AFDM differed between days. The AFDM: Chlorophyll-a ratio differed between days, and the interaction was significant; differences between control and F treatment occurred on days 3 and 27 (Tukey: p = 0.039 and 0.009, respectively). The epipelon had a predominance of heterotrophic components in the initial phase (day 3), and subsequently, the autotrophic components increased and reached a maximum on day 27. Two-way PERMANOVA evidenced differences in epipelon biomass estimates between days (F = 48.947; p = 0.002).

Floating Macrophyte Experiment
Compared to the control, the attenuation of light, TP, TN, and PM had an average increase of 34.7, 105.5, 14.9, and 279.9% in M treatment, respectively ( Fig. 4A-D). Considering the RM-ANOVA results, light attenuation differed between treatments and significant interaction (Table 1). Light attenuation between the control and M treatment differed on days 17, 27, and 60 (Tukey: p = <0.001, <0.001, and 0.031, respectively). In contrast, TP concentrations differed between days, and interaction between factors was significant, but differences between control and M treatment occurred on day 60 (Tukey: p = < 0.001). TP concentration increased on average 416.7% on day 60 when compared to the control. TN concentrations differed between treatments and days with significant interaction (Table 1) and differences between the control and M treatment occurring on days 17, 27, and 60 (Tukey: p =< 0.001). TN concentration decreased on days 17 and 27 and increased strongly on 60. PM concentrations differed between treatments, days and interaction between factors was significant on day 60 (Tukey: p = <0.001) when increased the concentration. At the end of the experiment, the average macrophyte coverage reached 96% of the water surface. Phytoplankton biomass increased by 169.8% on average in the M treatment, compared to the control (Fig. 4E). The phytoplankton biovolume did not differ between treatments.
Regarding the experimental period, epipelon chlorophyll-a decreased by 24.6% on average (Fig. 4F), and the AFDM and the AFDM: Chlorophyll-a ratio increased by 16.7 and 57.5% (Fig. 4G and H) in the M treatment compared to the control. The epipelon chlorophyll-a concentration did not differ significantly between treatments and days. The epipelon AFDM and the AFDM: Chlorophyll-a ratio differed between days. Two-way PERMANOVA evidenced differences in epipelon biomass estimates between treatments (F = 70.236; p = 0.0054) and days (F = 33.424; p = 0.0099).

Periphyton Bioreactor Experiment
Considering the experimental period, the light attenuation coefficient, TN, and PM increased by 17.2, 7.8, and 11% on average in PB treatment, respectively, compared to the control (Fig. 5A, C, and D). Differently, the TP concentration decreased by 2.8%, on average, in the PB treatment in the experimental period (Fig. 5B). Considering the RM-ANOVA results, light attenuation differed between days and significant interaction (Table 1). Differences in light attenuation between the control and PB treatment occurred on day 27 (Tukey: p = 0.003). The TP concentration differed between days. The TN concentration differed between days with a significant interaction. However, treatment was effective only on day 3 (Tukey: p = 0.010). Differences were detected in PM concentration between days, and the interaction was significant. Differences in PM concentration between PB treatment and control were detected only on day 60 (Tukey: p = 0.002).
The phytoplankton biomass increased by 42.5%, on average, in the experimental period (Fig. 5E), and the interaction between treatment and days was significant ( Table 1). The control and PB treatment differed at day 60 (Tukey: p = 0.002).
Despite the temporal variation, the highest epipelon chlorophyll-a was found on day 60 (Fig. 5F). On average, epipelon chlorophyll-a increased by 20.6% compared to the control. However, the average of the AFDM and AFDM: Chlorophyll-a ratio showed a 5.1 and 22.1% decrease, respectively (Fig. 5G, H). The epipelon chlorophyll-a differed between days, and the interaction between factors was significant with treatment effect on days 17 and 60 (Tukey: p = 0.044 and 0.001, respectively). There was a time effect on the epipelon AFDM. The AFDM: Chlorophyll-a ratio differed between days, and the interaction between factors was significant with treatment effect on days 17 and 60 (Tukey: p = 0.009 and 0.016, respectively). Based on the biomass estimates, two-way PERMANOVA showed differences in the epipelon biomass between days (F = 3.587; p = 0.0093), and the interaction between factors was significant (F = 34.785; p = 0.0097).

Treatments Comparisons
The TSI results for the experimental period showed that only the aeration treatment was able to improve the TSI to mesotrophic (TSI = 58), all other treatments remained in the eutrophic category: Control (TSI = 59), Flocculant (TSI = 69), Macrophyte (TSI = 61) and Periphyton Bioreactor (TSI = 60).
Considering the mean of environmental (light attenuation, TP, TN, PM) and biotic (phytoplankton biomass, epipelon chlorophyll-a, AFDM, AFDM: Chlorophyll-a ratio) variables in the trial period (Fig. 6A-G), we found differences between treatments. The light attenuation increased in all treatments. In contrast, we detected increased attenuation, especially in the macrophyte treatment (Fig. 6A). TN concentration had no difference between treatments, and the flocculant treatment had the highest concentrations (Fig. 6C). TP and PM concentrations decreased in the A treatment. RM-ANOVA results of PCA axis 1 score for abiotic variables showed the difference between treatments (F = 67.89; p = <0.001) and days (F = 11.89; p = <0.001), and the interaction between factors was significant (F = 2.96; p = 0.017). Differences were detected between treatments on all sampling days (Tukey: p = <0.05).

Discussion
Our findings demonstrate the effect of four restoration techniques used during an experimental period on abiotic conditions. Significant changes occurred in the light and nutrient availability in all treatments. Despite significant abiotic changes in the treatments, only the aeration treatment improved the TSI, changing from eutrophic to mesotrophic. In the A treatment, abiotic and biotic responses show a significant decrease in eutrophication, which was evidenced by a decrease in TP and PM concentrations and phytoplankton biomass. The applied techniques also promoted a decrease in the phytoplankton biomass, mainly in the flocculant, aeration, and periphyton bioreactor treatments. Within this experimental scenario, the epipelon responses evidenced that the periphyton bioreactor and the flocculant techniques favor the increase in algal biomass while the others could negatively affect the community.
Our results show that the bottom aeration promoted the sedimentation of nutrients and particulate matter. The decrease in TP concentration in the A treatment showed the beneficial effect of aeration on phosphorus loss. Studies showed that oxygen availability acts to trap P in the sediment, and anoxic conditions favor P release, determining the relationship between P in water and sediments (e.g., Nygrén et al., 2017). However, in our experiment, the decrease in water TP concentration can be more associated with the sedimentation of particulate matter in general. Grochowska et al. (2017) reported that aeration improves water transparency. Despite the reduced particulate matter, the light entry did not improve, as the light attenuation coefficient increased in the A treatment. Thus, our results suggest that air bubbles could have acted in light dispersion, decreasing its availability. Under the environmental conditions of A treatment, the epipelon biomass and the AFDM:Chlorophyll-a ratio responses no indicated increase the epipelic algal biomass. Sedimentation of particulate matter may have contributed to the Chlorophyll-a ratio, due to the contribution of organic matter, as phytoplankton. In addition, the sediment oxygenation can favor the growth of heterotrophic organisms (Craig et al., 2015) and increase grazing. Therefore, bottom aeration promoted several positive responses for lake restoration, such as reduced phytoplankton blooms and TP availability. However, the phototrophic epipelon biomass did not respond positively to the aeration treatment on the experimental days.
We observed that adding a natural flocculant (F treatment) reduced the average phytoplankton biomass during the 27 experimental days. However, there was a pronounced increase in biomass after 60 days. Furthermore, light attenuation increased throughout the experimental period. Thus, our results demonstrate that the flocculant dragged particulate matter, including phytoplankton, to the bottom. Consequently, the water transparency decreased, indicating that the flocculant caused the shading. We also found an increase in N concentration from the experiment´s beginning and TP from day 17. In a jar-test experiment, Hou et al. (2018) reported the degradation of the Tanfloc flocculant and Microcystis aeruginosa cell lysis, generating secondary pollution. Considering the decrease in phytoplankton biovolume and the increase in P and N concentrations in the F treatment, it is likely that flocculant degradation and cell lysis have occurred in this treatment, promoting an increased nutrient concentration. Despite the decreased light availability, the AFDM: Chlorophyll-a ratio indicated an increase in autotrophic components in the epipelon at 27 days. However, planktonic algae and cyanobacteria could have contributed to the chlorophyll-a concentration in the epipelon. In the studied reservoir, Microcystis aeruginosa is often dominant (Crossetti et al., 2018). This cyanobacteria species has a benthic phase within its temporal dynamics in eutrophic systems (e.g., Latour et al., 2004). Thus, it has adaptive strategies to grow in the sediment after flocculant drag. Therefore, the effect of the flocculant on the epipelon was inconclusive based on biomass, but the taxonomic analysis may help clarify the community response. Similar to what happens with chemical flocculants (Zhang et al., 2015), our results suggested that adding the natural flocculant can cause secondary pollution over time, favoring the return of phytoplankton blooms.
The experiment introducing the floating macrophyte Eichhornia crassipes increased the number of individuals throughout the experimental period, despite the established population control. Numerous studies report the high growth of water hyacinth under eutrophic conditions (e.g., Thamaga and Dube, 2018) and, still, it can colonize the entire lake surface (Bicudo et al., 2007). In restoration processes, submerged macrophytes are frequently used to maintain the clear water phase in shallow lakes (Scheffer et al., 2001). However, we used the dominant floating macrophyte Eichhornia crassipes, commonly used in sewage treatment and lake restoration (Wang et al., 2013) since there were no records of submerged macrophytes in the lake analyzed. Light attenuation, TN, TP, and PM concentrations increased in treatment M during the experimental period. The increase in nutrient concentration at 60 experimental days suggests that the management adopted may not have been sufficient to control the increase in phytoplankton biomass. In the case of the M treatment, the removal control of the individuals must have a time interval of fewer than 15 days due to the high rate of growth and senescence of Eichhornia crassipes, particularly in a eutrophic environment. Also, there was an increase in phytoplankton biomass and particulate matter on day 60. According to Gamage and Asaeda (2005), the senescence process of Eichhornia roots can increase the amount of particulate matter in the water. Several studies have reported that floating macrophytes can cause strong shading and compete for resources with algal communities (e.g., Yan et al., 2016). These phenomena can explain the phytoplankton biomass decrease during the initial phase of the experimental period (3-17 days) and the dominance of heterotrophic components in the epipelon. Therefore, our findings showed that the introduction of Eichhornia crassipes improved the water quality only up to day 17. Thus, E. crassipes removal as a management strategy should be carried out in less than a 15-day interval to favor the phototrophic epipelon growth and improve water quality. The use of macrophytes in the restoration of eutrophic ecosystems can be a problem in the absence of adequate management due to their high reproduction rate, dispersion, and resistance to different environments (Yan et al., 2016).
Our results showed that periphyton bioreactor (PB treatment) changed the environmental conditions during the experimental period. In this treatment, there was a decrease in PM and phytoplankton biomass. Despite the promising environmental responses in the first two weeks, we found a return of phytoplankton bloom. Studies reported that nutrient removal from the water, especially P, is most efficient early in the periphyton colonization process (He et al., 2017;Carneiro and Ferragut, 2022). In addition, in the first management, we observed the visible growth of metaphyton in the PB treatment from day 20, demonstrating the need for management to be carried out in a shorter period. Metaphyton can replace submerged macrophytes in the competition for resources with phytoplankton and maintain the clear phase in lakes (Hao et al., 2018). The environmental conditions of the PB treatment favored the increase in epipelic chlorophyll-a and the decrease in the AFDM: Chlorophyll-a ratio on most sampling days. Our results indicated an increase in the contribution of autotrophic components to the epipelon structure during the experimental days. Therefore, our results showed that proper management of PB, apparently with less than a 15-day interval, can improve the environmental conditions with a decrease in phytoplankton biomass and growth of phototrophic epipelon.
Based on green technologies, we confirm the hypothesis of epipelon growth in treatments with the eutrophic lake restoration technique. However, epipelon biomass increased only in the periphyton bioreactor and flocculant treatments. In addition, favorable conditions for the epipelon growth seems to depend on frequent management of the applied technique, mainly in the treatment with macrophytes and periphyton bioreactor, even in the case of "environmentally friendly" practices. We emphasize that the epipelon responses in the restoration processes of eutrophic lacustrine ecosystems need to be better investigated.

Data availability
The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.