The Fernando Costa Park, known as Parque da Água Branca, is na urban park located in the central region of the Municipality of São Paulo, and has more than 79 thousand m² of green area, in a total of 137 thousand m². The park houses different educational and research institutions and presented various cultural activities and educational partners, as well as serving as a space for exhibitions [11] . The two lakes were built in cascade, the first lake presents a source of aeration (fountain) that is fed by the water of the second lake, forming a process of partial feedback (return) of the waters.

The study included two stages: 1) the environmental diagnosis through the limnological characterization of the lakes; 2) the elaboration, implantation and monitoring of mitigating measures to control pollution.

In the first stage of this research, 12 water samples were collected on the sub-surface of 2 lakes populated with carp named: Tank 1 with fountain and Tank 2 without fountain, in November 2017. The collections occurred at 3 different times in 2 consecutive days to evaluate water quality and verify the dynamics of phytoplankton metabolism.

From the results obtained in the first stage it was understood that the bioremediation methodology would be adequate to reach the proposed objectives. Therefore, Eichhornia crassipes confinement network structures were implanted at this stage. Tank 2 was used for the implantation of the structure named Artificial Floating Island (AFI). Tank 2 had an average depth of 0.80 m (approximately), a surface area of 437 m² and a volume of about 350 m³, and was chosen to receive water from the Tank 1 and present an area for AFI implantation. The implantation of the AFI occurred in April 2018, being colonized with aquatic macrophytes in the proportion of 2.5 kg per m², covering approximately 10% of the area of lake 2. To follow the development of the macrophytes, biometrics were done to estimate the total biomass of the macrophytes. To estimate the biomass, plastic bags were used, digital portable scale of up to 50 kg and a quadrant (0.5 × 0.5 m) made with PVC pipes. The quadrant was randomly played where the macrophytes were confined, collecting the plants within the known area for weighing and biometrics. After the measurements, the macrophytes were returned to the confinement site, this same procedure was repeated for another 2 times.

In step 2, the water samples were collected weekly in the subsurface of the water column, the monitoring lasted 247 days, starting on 03/28/18, the IFAs were implanted after 28 days of the beginning of the monitoring.

The pH, water temperature (˚C), electrical conductivity (mS∙cm⁻¹), turbidity (NTU) and dissolved oxygen (mg∙L⁻¹) were measured “in locu” with Horiba W-53 multisonda. The water transparency was estimated “in locu” according to the visual disappearance of the Secchi disk (m). The concentrations of the nutrients: ammonium (mg∙L⁻¹), total nitrogen (μg∙L⁻¹), orthophosphate (μg∙L⁻¹) and total phosphorus (μg∙L⁻¹) [12] were determined in the Reference Laboratory in Limnology of the Fisheries Institute. The estimation of phytoplankton biomass was quantified by the concentrations of Chlorophyll a (μg∙L⁻¹). For the taxonomic analysis of the phytoplankton community, water sample was filtered through a plankton network with a 20 μm mesh opening. The samples were stored in a glass vial and fixed in 4% - 5% form. The taxonomic analysis was performed using Zeiss-Axioplan 2 optical microscope, in magnifications of 400 and 1000 times.

The descriptive analysis of the data was used to characterize the water quality during the first stage.

The principal component analysis (PCA) was used to evaluate the influence of the implantation of an artificial floating island system (AFI) on the water quality of an urban lake, using program PC-ORD 6.0 for Windows [13] . Data were converted applying [log(x + 1)], except for pH.

The mean values of the limnological variables were similar among the lakes (Table 1).

The high concentrations of total phosphorus recorded in lakes Tank 1 and Tank 2, with average values of 92.37 and 99.89 μg∙L⁻¹, respectively, evidenced the existence of a source of this nutrient and indicate an environment favorable to the growth of planktonic algae. However, the low values of chlorophyll concentrations observed, were contrary to this assertion and also to the aspect of dense and

uniform blue-green coloration observed during the collection. These values may be related to the type of water collection carried out on the subsurface of the water column, since the algae during the blooms are on the surface of the water column, an adaptive process to capture a greater amount of light.

During the taxonomic survey of the phytoplankton community, some cyanobacteria considered to be blooming and potentially toxic were identified in the lake samples (Microcystis aeruginosa (Kützing) Kützing, Microcystis brasiliensis (Azevedo & Santana) Rigonato, Microcystis wesenbergii (KomÁrek), Radiocystis fernandoi KomÁrek & KomÁrkovÁ-LegnerovÁ). However, it is important to note that the presence of these toxins in the fish ponds can be considered a health risk for organisms inhabiting these sites (ex. fish and birds) due to their toxicity [14] , mainly by the bioaccumulation of these toxins by the fishes [15] . MAGALHãES et al., [16] verified the bioaccumulation of microcystins in tilapia (Tilapia rendalli) that fed on toxic cyanobacteria in Lake JacarepaguÁ and confirmed the accumulation and persistence of microcystins in muscle tissues even after the decline of cyanobacteria blooms.

The low renewal of lake water, where part of water exiting the lake 2 spillway is used by the lake 1 fountain, making a water recirculation system, may have provided a stable system, providing competitive advantage for slow growing species, such as flowering cyanobacteria (ex. Cylindrospermopsis, Dolichospermum, Microcystis) [17] .

In view of the data presented, we concluded that there was a need for intervention in the lakes with a view to improving the quality of water, seeking not only the scenic beauty of the lakes, but also the quality of life of the organisms existing in these places. Different mitigation measures were proposed: increase of water in intake volume (more important action that would bring fast and permanent results; mechanical cleaning of the surface (it contributes with the scenic beauty, but it is an action that must be done daily); removal of algae by lowering the water column from 4 to 6 cm (indicated action if it is not possible to increase the volume of incoming water); use of aquatic plants (ex. Eichhornia crassipes) to control the amount of nutrients (biomanipulation technique); use of bioflocos (probiotic) to reduce algae.

In this sense, after verification of all possibilities, it was understood that for the conditions in which the lakes presented technique of biomanipulation would be the most adequate. The biomanipulation technique with the use of constructed wetland (CWs) is alternative to conventional treatments. Constructed Wetlands are designed and constructed systems for the treatment of effluents in order to use natural processes in the removal of pollutants [18] . A variation of CWs has been tested more recently and is called Artificial Floating Islands (AFIs). The AFIs are a variation of the treatments systems by CWs aiming to improve the water quality of different pollutant sources, besides being attractive landscape. Therefore, among the mitigating measures described above, biomanipulation using phytoremediation techniques by AFIs seemed to be the most suitable for nutrient control and phytoplankton overgrowth, controlling the eutrophication process and promoting the improvement of the scenic beauty of the landscape park lakes.

Principal Components Analysis (PCA) was used to evaluate the contribution of the implantation of an Artificial Floating Islands system on the water quality of an urban lake (Figure 2; Table 2). The PCA using seven environmental variables,

summarized 74.3% of the total variability of the data in the first two ordering axes, and was statistically significant for the first axis (p < 0.01). The first axis summarized 62.5% of the total variability of the data and was the most important to explain the temporal distribution. On the negative side of axis 1, collections were ordered before or shortly after the implantation of the AFIs, these were associated with the highest values of chlorophyll a, total phosphorus, water temperature and turbidity. The lower values of these variables and the higher values of pH and Secchi disk were ordered on the positive of axis 1.

After 60 days of aquatic plant implantation, it was recorded decrease in the concentrations of total phosphorus and chlorophyll a (Figure 3), the chlorophyll a represents the estimate of the phytoplankton biomass. During this period was also observed improvement in both the visual disappearance of the Secchi disk (Figure 3) and the appearance of the surface water (Figure 4). NABOUT and NOGUEIRA [3] studying four urban lakes, found that local factors such as limnological, geomorphological and landscape characteristics of lakes are strongly associated with structure of the phytoplankton community, often more than seasonal climatic variations.

Different species of macrophytes present different capacity of removal of nutrients from the water. This ability depends on some factors, which vary from species to species, such as biomass production, capacity of store nutrients in biomass, adaptation to local climatic conditions and the physical and chemical characteristics of water [19] . Studies with aquatic macrophytes for treatment of aquaculture effluents, in general, show that these systems efficiently remove nutrients and may still allow their reuse of water in aquaculture activities, such as the work of LIN et al. [20] ; SCHULZ et al. [21] ; LIN et al. [22] ; HENRY-SILVA & CAMARGO [19] ; HENRY-Silva & CAMARGO [23] and SU et al. [24] . The removal of nutrients was evident in the present research indicating that the Artificial Floating Islands technology was efficient to control the pollution of the studied lakes by improving their water quality. It can also be observed that the

macrophyte used in this study Eichhornia crassipes (free floating plant) appeared to be effective in the removal of nutrients from the water as well as the emergent macrophytes used in the studies of HUBBARD et al. [8] ; HEADLEY & TANNER [9] and LYNCH et al. [10] .