At the moment, Lake Rodó offers good chances for restoring through biomanipulation, taking into account the nutrient reduction (due to the water re-circulation system), and the low abundance of filamentous cyanobacteria. Clear water phases observed during springs were concomitant with a greater mesozooplankton herbivores abundance (above all cladocerans). Before the restoration activities, high transparency levels (more than 1m of Secchi depth) were not found. The higher percentage of edible microalgae could be the crucial factor that explain it. Probably, the presence of abundant populations of small omnivorous fish maintains a high predation pressure on the zooplankton and leads to an absence of large-bodied herbivores.

Obviously, there is a great difference between getting a short period of clear water and not achieving clear water at all. In this sense, an efficient nutrient removal programme associated with mechanical harvest of floating plants, and more effective removal of small omnivorous fish could be the key factors to achieve a stable clear water phase. However, if the Microcystis bloom occurs in summer and spring, additional measures will be needed to improve water transparency.

INTRODUCTION

Lake Rodó (34°55´S, 56°10´W) is a small (1.3ha), shallow (maximum depth 2m) and turbid system, a characteristic mostly attributed to algal biomass. This man-made lake, constructed in 1917, is located in a park in Montevideo (1.5 million inhabitants), Uruguay. The lake is used for recreational purposes (walking, jogging, boating, fishing) and cultural activities. The eutrophication process is conditioned by a high nutrient inputs from the urban area surrounding the lake, and the main consequences of this process are the loss of aesthetic and recreational value. Since spring 1996, this urban lake has been under restoration to improve water quality and establish a clear water phase. On that purpose, it was completely drained and sediments removed, and nowadays there is no stream inputs. Groundwater was used to re-fill the lake, and as the water source (480 m³ d^-1) between January 1997 to July 1998, but after July 1998 the groundwater pumping was eliminated. The lake has been in connection with two pools covered with free-floating plants and under a water re-circulation regime. The biomanipulation technique applied was the removal of the small omnivorous fishes with the seine net principally in winter.

Our objectives were to analyse the physicochemical and biological conditions in a shallow hypertrophic lake under restoration, to understand the main factor affecting water transparency, to analyse the main responses related to the measures applied, and to propose the more suitable measures to establish a stable clear water phase.

MATERIALS AND METHODS

Water samples were weekly collected since 01/97 to 04/97 and biweekly up to 01/99 from four stations (groundwater, lake, pool1 and pool2). Temperature, dissolved oxygen, conductivity, pH and transparency were registered in situ. Chemical analysis included alkalinity, suspended solids, soluble reactive phosphorus (SRP), nitrate, ammonium, total nitrogen (TN), total phosphorous (TP), silicate (SiO₂^-2) and chlorophyll a. Sediment samples were taken every two months and the content of organic matter, total nitrogen and total phosphorous were determined.

RESULTS AND DISCUSSION

A strong decrease in water transparency was observed during the first three months after the re-filling and it was closely related with an increase of phytoplankton biomass. The highest water transparency (more than 1m Secchi Depth) were registered during both springs. These clear water phases lasted two or three weeks and in these periods the lowest phytoplankton biomass and the highest ammonium and phosphorous concentrations were observed.

The only available data of the previous conditions of the lake was published by Sommaruga and corresponds to samples taken in 1992. The principal differences between the lake condition in 1992 and 1997-1998, are the increment of water transparency from an average of 0.2 to 0.5 m SD, together with a reduction of phytoplankton biomass and changes of phytoplankton composition. Other relevant changes were a decrease of total phosphorus (60%) and a significant increase of total nitrogen (256%). The mean TN:TP ratio (by weight) varied from 8.5 to 67.6, turning the lake from N-limiting into P-limiting conditions.

After the re-filling, the lake rapidly reached a hypertrophic state mostly determined by the high nutrient input, due to groundwater characteristics. In this sense, conductivity, SRP and nitrate showed clear differences between sampling sites. Groundwater supply had higher conductivity and nutrient concentrations compared with the lake and pools. The analysis of nutrient concentrations at four selected sampling stations allowed us to identify a clear gradient from groundwater to pool 2. Considering this information, we designed a re-circulation system (covered with free-floating plants) pumping water from pool 2 towards the lake. Groundwater has been used only to maintain the water level in periods of high evaporation. The main changes observed after this re-circulation system implementation were nitrate and chlorophyll a reduction, and a decrease of TN/TP and Chla/TP ratios. At present, we are evaluating the capacity of nitrogen and phosphorous removal associated with the aquatic plant management.

The high algal biomass and the sedimentation process caused an increase of nitrogen, phosphorous and organic matter content of the sediment (mainly at surface). The organic matter decomposition promoted anoxic conditions in some periods during summer. The benthos genera registered during the study were Coelotanypus, Procaldius, Larsia, Polypedium, Goeldichironomus (Chironomidae), Hellobdella (Hirudinea), and Limnodrilus hoffmeisteri (Oligochaeta). Benthos density varied strongly, including some periods when no organism was found. The total abundance was negatively correlated with the proportion of total phophorous and organic matter of the sediments. The algal sedimentation and decomposition process associated might have influenced the temporal benthic abundance decrease and prevented the colonisation of sediments.

The phytoplankton community showed a very complex sequences of replacement and more than 150 species were identified. The main differences between 1997 and 1998 were the predominance of diatoms during the first year and of cyanobacteria during the second. It's remarkable that Microcystis genera was present in a much longer period in 1998 than the year before. The principal factors that influenced the temporal replacement sequences are the self-shading effect, TN/TP ratio changes and grazing pressure by mesozooplanton herbivores.

The zooplankton community was dominated by small herbivores represented mainly by rotifers (principally Keratella tropica). Copepods showed the higher abundance at the beginning of the spring (Notodiaptomus incompositus, Metacyclops mendocinus and Tropocyclops prasinus meridionalis). Cladocerans were the less abundant group, presenting their higher abundance in spring and summer 1997-1998 (caused by the presence of Moina micrura in spring and Diaphanosoma birgei in summer).

Before the beginning of the restoration programme, phytoplankton was dominated along the whole year by Planktothrix agardhii (between 82 and 99% ot the total abundance), which was the only Cyanobacteria species registered. At the moment, the community is more diverse and with lower biomass. The increment of TN/TP and Si/TP ratios after re-filling, could have influenced the green and diatom algae predominance, especially during 1997. On the contrary, species composition and relative abundance of zooplankton before and after re-filling were fairly similar. However, the abundance of rotifers was lower than in the past, while mesozooplankton herbivores showed higher abundance.

Some ecologist groups without connection with restoration programme introduced some fishes species to the lake during the re-filling process. The species composition and abundance of fish stocked are unknow, but the community became rapidly characterised by a high fish biomass, with dominance of small omnivorous fishes and the absence of piscivores. Cnesterodon decemmaculatus (Poeciliidae), was the dominant species and together with Jenynsia lineata (Jenynsiidae), both small omnivorous, represented between 58 and 99% of the total fish biomass. The average values estimated were 307 kg/ha in the first year and 160 kg/ha in the second. The successful reproductive strategy of C. decemmaculatus and J. lineata coupled with short generation times and low predation pressures, have favoured the fast re-colonisation and the high biomass reached in a short period. The longer generation times of the Cichlids contributed to their low abundance. In the future, Cichlasoma might exercise a greater predation pressure on the young omnivorous individuals.

At the moment, Lake Rodó offers good chances for restoring through biomanipulation, taking into account the low abundance of filamentous cyanobacteria. In this sense, the clear water phase was concomitant with a greater mesozooplankton herbivores abundance (above all cladocerans). Before the restoration activities, a cladocerans increment was not correspondent with high transparency levels. The higher percentage of edible microalgae could be the crucial factor that explain it. Probably, the presence of abundant populations of small omnivorous fishes maintains a high predation pressure on the zooplankton and leads to an absence of large-bodied herbivores. Obviously, there is a great difference between getting a short period of clear water and not achieving clear water at all. In this sense, an efficient nutrient removal programme associated with mechanical harvest of floating plants, and more effective removal of small omnivorous fish could be the key factors to achieve a stable clear water phase. However, if the Microcystis bloom occurs in summer and spring, additional measures will be needed to improve water transparency.