Geographic coordinates and elevations of the sampling sites along the reservoir cascade.
The study identified limnological patterns in the Paranapanema River reservoir cascade, one of the main tributaries of the high Paraná River, La Plata Basin, southeast Brazil. Samplings were carried out in eight reservoirs from a total of 37 sites. We analyzed the water transparency, depth, and vertical profiles of temperature, pH, conductivity, dissolved oxygen, biochemical oxygen demand, total solids, suspended solids, total nitrogen, total phosphorus, chlorophyll a, and thermotolerant coliforms. Additionally, the trophic state index for tropical/subtropical reservoirs, the water retention time and morphometric characteristics of each reservoir were calculated. Longitudinal compartmentalization is conspicuous in storage reservoirs, whereas the magnitude of temporal changes is higher in run-of-river systems. The lateral component of spatial heterogeneity was also very important for some reservoirs, determined basically by the entrance of tributary rivers. On the vertical dimension, summer thermal stratification, followed by oxygen decrease in bottom layers, in the central channel and lacustrine zones of deeper and larger reservoirs was observed. The ultraoligotrophic condition prevailed, despite signals of intensive land use for agriculture—recurrent high phosphorus values. The acquired experience provided a baseline for a permanent limnological and water quality program, which subsidizes management actions in the basin.
- water quality
- trophic state
- spatial variability
- seasonal variability
River damming for hydropower production is presently a major human interference on fluvial systems, all over the world. In these first decades of the twenty-first century, the construction of 3700 dams is expected, increasing the global hydropower production by 73%. This would correspond to an intensification in the exploitation of the technically feasible hydropower potential from 22 to 39% .
Hydropower is crucial in Brazil, representing the main source of electric generation, 107,601 MW of a total of 166.872 MW, according to the database of the Brazilian National Agency of Electric Energy (
Hydroelectric dams are an integrated and complex generation system that provides relatively clean and renewable energy. However, independent of the reservoir design and operation, they promote substantial interference on the ecological structure and functioning of river basins  and permanent loss of the regional geodiversity [3, 4].
Classical studies on reservoirs [5, 6] distinguish at least three limnological zones in large reservoirs: the region with strong influence of the incoming river—lotic region; an intermediate transition region; and finally a region similar to lakes, near the dam. However, it is known that in large tropical reservoirs, this multidimensional zonation is greatly influenced by the input of secondary tributaries and by the differential water retention time of each reservoir arm [7, 8, 9, 10]. These distinctive areas or compartments in reservoir ecosystems may also change seasonally, in a complex dynamic [5, 7, 11, 12].
The large-scale variability of reservoirs along the main axis is determined by longitudinal gradients of flow velocity, depth, width, particle sedimentation, transparency and light penetration, and thermal stratification [11, 12, 13]. Compared to lakes, reservoirs exhibit high watershed area/water body area, shorter but varying retentions times, a rapid aging process related to watershed uses, and high capability to retain organic and inorganic matter [14, 15].
An integrated approach, including continuous monitoring and innovative researches, is required for the understanding of the reservoir ecosystems, giving their inherent complexity, with many components and subsystems that interact intensively in space and time .
The present study was planned to discriminate the main limnological patterns of a tropical/subtropical reservoir cascade aiming to provide scientific support for management purposes. We analyzed the morphometric characteristics; operation type; water retention time; physical, chemical, and biological variables as well as the trophic status. The acquired experience served as the base for the establishing of methodological protocols that have been employed in the long-term (permanent) limnological and water quality monitoring program (8 years now) in the Paranapanema River reservoir cascade (southeast Brazil). The two most important premises are: (1) the reservoir operation (run-of-river or storage systems) is determinant for the establishing of distinct limnological patterns and (2) the Paranapanema River reservoir cascade is influenced by the intensification of the agriculture in the watershed.
2. Study area
The Paranapanema River is one of the main tributaries of the Paraná River (La Plata Basin), located between the coordinates 22–26° S and 47–54° W, on the tropical/subtropical boundary. The river, with a length of 929 km, is under federal jurisdiction, because it is the natural border between the states of São Paulo and Paraná.
Since the 1950s, 11 hydropower plants were constructed in the main river course. In last two decades, these reservoirs (eight larger ones) have been intensively studied by researchers of the São Paulo State University, Campus of Botucatu, São Paulo [17, 18].
The reservoirs selected for the study are Jurumirim (JR), Chavantes (CH), Salto Grande (SG), Canoas II (CII), Canoas I (CI), Capivara (CP), Taquaruçu (TQ), and Rosana (RS), arranged in a cascade (upstream → downstream) system. Three of the them, Jurumirim, Chavantes, and Capivara, are storage systems (i.e., with high water retention times), whereas the others are run-of-river systems. The total installed potential is 2241 MW.
For the development of the study, information was obtained at 37 sampling sites (Figure 1), whose distribution intended to cover the entire river’s continuous (inter-reservoir) variability as well as the internal (intra-reservoir) longitudinal gradient established between the lotic (Paranapanema River entrance) and lentic (dam) areas of each reservoir. Additionally, the influence of important secondary tributaries (river mouths) was also considered. At least three to six sampling sites were selected in each reservoir. Table 1 presents the geographic coordinates as well as the altitude data of the sampling sites, obtained with a Garmin Etrex Vista H GPS.
|Reservoir||Sampling site||Coordinates||Altitude (m) (a.s.l.)|
3. Material and methods
The sampling campaigns for physical, chemical, and biological measurements were carried out in two periods of the year: March—wet season (late summer)—and October—dry season (spring) of 2011. All sites of each reservoir/period were sampled in the same or in two consecutive days.
The limnological variables measured in situ, and respective methodologies, are shown in Table 2.
|Water transparency (Transp)||Secchi disk depth|
|Water temperature (Temp)||Eureka multi-parameter probe—vertical profile|
|Dissolved oxygen (DO)||Eureka multi-parameter probe—vertical profile|
|pH||Eureka multi-parameter probe—vertical profile|
|Conductivity (Cond)||Eureka multi-parameter probe—vertical profile|
|Turbidity (Turb)||Eureka multi-parameter probe—vertical profile|
Water samples were collected with a Van Dorn bottle at three depths along the water column, corresponding to the surface, middle, and bottom (about 0.5–1.0 m above sediments), for the analyses of nutrient (total nitrogen and phosphorus), total solids, suspended solids, chlorophyll a, biochemical oxygen, and thermotolerant coliforms. When filtration was required, Millipore AP40 membranes and vacuum pump were used. For weight determination, a Denver analytical scale (0.00001 g) was used. Laboratory determinations followed methodological principles presented in Table 3.
|Total nitrogen (TN)||Spectrophotometry [19, 20]|
|Total phosphorus (TP)||Spectrophotometry [19, 21]|
|Total solids (TS)||Evaporation/gravimetry (100°C) |
|Suspended solids (SS)||Gravimetry |
|Biochemical oxygen demand (BOD)||Incubation 20°C/5 days |
|Chlorophyll a (Chl)||Spectrophotometry [24, 25]|
|Thermotolerant coliforms (Coli)||Multiple tube MPN |
The shore line development index was calculated according to . The theoretical residence time (days) was defined as the ratio of reservoir volume and the flow calculated using the formula: TRT = V/(Q × 86,400), where V = reservoir volume (m3); Q = mean flow (m3 s−1); and 86,400 = number of seconds contained in a day. Input data for each reservoir are available at the National Water Agency (Agência Nacional de Águas, in Portuguese) (
The trophic state index was determined in accordance with  for tropical/subtropical reservoirs (TSItsr), which includes six categories: (U) ultraoligotrophic (≤51.1), (O) oligotrophic (51.2–53.1), (M) mesotrophic (53.2–55.7), (E) eutrophic (55.8–58.1), (S) supereutrophic (58.2–59), and (H) hypereutrophic (≥59.1), respectively.
The studied parameters (when applicable) were compared with the standard references (Table 4) established by the federal framework directive for Class 1 waters—the best possible condition (after the Special Class), with no human use restriction and appropriated for communities’ protection (
|Dissolved oxygen||>6 mg L−1 O2|
|Biochemical oxygen demand||>3 mg L−1 O2|
|Total solids||>500 mg L−1|
|Total nitrogen||>1.27 mg L−1 lentic systems|
|>2.18 mg L−1 lotic systems|
|Total phosphorus||>0.020 mg L−1 lentic systems|
|>0.025 mg L−1 intermediate systems*|
|Chlorophyll a||>10 μg L−1|
|Thermo tolerant coliforms||>200 NMP 100 mL−1|
A principal component analysis (PCA) was performed to summarize variation tendencies or patterns for limnological variables during both sampling periods, using the PRIMER v6 statistics package for Windows (Plymouth Routines in Multivariate Ecological Research,
The general characteristics of the studied reservoirs (morphometry, operation, trophy, etc.) are compiled in Table 5 and the wide spectrum of conditions certainly influence not only the structure but also the limnological functioning of these ecosystems, as pointed out earlier.
|Type (modus operandi)*||S||S||R||R||R||S||R||R|
|Shore line development index||17.1||15.3||6.6||6.1||6.1||18.2||9.5||8.2|
|Water retention time (days)||400||335||2||4||6||115||7||17|
|Maximum measured depth (m)||32||79||10||15||25||40||26||27|
|Altitude (m) (a.s.l.)||563||473||389||367||354||325||271||241|
The obtained results, in situ and laboratory measurements, are presented following the reservoir’s sequence, from upstream toward the mouth. Graphical representations (Figures 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33) are standardized for all reservoirs, including monthly variation of the water level versus flow, vertical profiles of temperature and dissolved oxygen, and intra-reservoir variability (upstream/tail → dam) of transparency, electric conductivity, pH, turbidity, total solids, suspended solids, biochemical oxygen demand, total nitrogen, total phosphorus, chlorophyll a, and thermotolerant coliforms.
5. Inter-reservoir variability
The Paranapanema River reservoir cascade exhibits a wide variability of conditions, when compared the distinct physical dimensions. Salto Grande, the oldest reservoir (60 years), is smaller, with area of 12 km2; perimeter of 81 km; volume of 63 hm3; and a very low water retention time, only 2 days. Conversely, the Capivara reservoir (40 years) has the highest area, perimeter, and volume—576 km2, 1550 km, and 11,743 hm3, respectively. Capivara is also the most dendritic reservoir, with a shore line development index of 18.2. The two newer (19 years) reservoirs, Canoas I and Canoas II, have the lowest shore line development index, 6.1. The first reservoir in the series, Jurumirim, has the highest water retention time, annual mean of 400 days, followed by the second, Chavantes, with 335 days. The water accumulated in these two upstream reservoirs is fundamental for the flow control along the entire cascade. For both, there was an increase in water release at the end of the dry season (September and October) to supply bellow hydropower plants and, consequently, a decrease in water retention time. In these two storage reservoirs, as well as in Capivara (retention time of 115 days), there is an accentuated fluctuation in the water level, between 2 and 4 m along the studied year. This annual amplitude can be even higher, depending on the year, up to 9 m in Capivara, but not necessarily coupled to the rain/dry regime . In the other reservoirs, operated as run-of-river systems (retention time between 2 and 17 days), the level fluctuation is minimum (<0.5 m). Data show that flow is intensively manipulated in order to keep the best condition for hydroelectric generation. The absolute flow values tend to increase along the river, as expected, from 181 m3 s−1registered in Jurumirim (August—dry season) to 1679 m3 s−1 in Rosana (February—rainy season). Ref.  analyzed bellow dam releases of Capivara and Taquaruçu reservoirs (ninth and tenth in the cascade) in short periods of time (24 h cycles), showing that differences between flow peaks reached impressive 1500 m3 s−1. The authors demonstrated that the effects of these hourly discharge variations are significant on the limnological variables’ dynamics, especially when intake to the turbines comes from hypolimnion (bottom layers) such as in Capivara dam. In this case, there is a transference of low oxygenated waters, disturbing the limnological and biota rehabilitation of bellow dam river stretches, whose importance has already been evidenced [29, 30, 31, 32]. However, this recovery process may not be effective in reservoir cascades, where the distance between consecutive reservoirs is too short. In case of Paranapanema, the longest inter-dam stretch is about 110 km, between Canoas I and Capivara.
The heat accumulation in the reservoir’s water mass contributes to an increase of temperature along the cascade. Mean values for the water column varied from 21.4°C (sampling site 12) to 26.9°C (sampling site 33) in wet-summer and from 19.7°C (sampling site 12) to 24.6°C (sampling site 36) in dry-spring. The first corresponds to Chavantes dam, the deepest site in the cascade, and the other value to the two last reservoirs, Taquaruçu and Rosana, respectively.
Principal component analysis, performed to provide an ordination and synthesis of the data set, evidenced the main spatial and temporal trends (Table 6; Figure 34). The organization during the dry period was mainly represented by Component 1 (explaining 30.7% of data variability) and the wet season by Component 2 (17.6% of data variability). The first component was negatively correlated with transparency and depth, grouping sampling sites (dam zones and central channel sites) of the storage reservoirs: Jurumirim and Chavantes. Sampling stations of Salto Grande, Capivara and Rosana, were associated to the positive side of Component 1, determined by suspended solids, nutrients, and chlorophyll (more eutrophic conditions). During the wet season, most sampling stations were associated to the negative side of Component 2, correlated with higher values of temperature, pH, nutrients, and lower concentrations of dissolved oxygen.
|Limnological variables||PC1 (30.7%)||PC2 (17.6%)|
|Biochemical oxygen demand||0.143||0.124|
|Inorganic suspended solids||0.415||0.191|
|Organic suspended solids||0.390||0.165|
The predominant trophic state classification for the studied reservoirs was ultraoligotrophic, except for Canoas I, which was oligotrophic. A few results, from a total of 74 individual samples or mean values in case of vertical profiles, were not in conformity with the standard references of the federal legislation for Class 1 waters : 2 values of pH, 6 of turbidity, 1 of chlorophyll, 6 of thermotolerant coliforms, and 19 of total phosphorus. This indicates punctual and isolated sources of degradation. The only exception is phosphorus, above the limit in 25.6% of the measurements. A detailed comparison of the Paranapanema River reservoirs based on distinct water quality and trophic indices is presented by . The study corroborates the good environmental condition of the river but also evidences some negative effects (e.g., phosphorus) associated to the land use intensification—boom of the agrobusiness, especially for the reservoirs in the middle river basin during the wet season .
Another potential eutrophication problem is the expansion of cage-aquaculture . This activity has been intensified in the recent years, mainly in Canoas I and Canoas I reservoirs.
6. Intra-reservoir spatial organization—longitudinal dimension
Conspicuous longitudinal gradients (upstream lotic zone → lacustrine dam zone) were observed in the studied reservoirs, especially for the larger storage ones. Classical studies pointed out that compartmentalization enhances the spatial and temporal complexity of reservoirs [5, 13, 34]. This structural characteristic has been verified for several Brazilian reservoirs [7, 10, 11, 12, 35, 36].
The Secchi disk transparency, a very simple (easy and cheap) measurement, but limnologically not trivial, is a robust indicator of the presence of distinct water masses along the main axis of a reservoir. In Jurumirim, we registered a clear increasing of transparency toward the dam, about five times in wet season and nine times in dry season. The opposite trend, a longitudinal decrease, occurred for turbidity and suspended solids.
Other variables that indicate well-defined longitudinal gradients were nitrogen, phosphorus, and chlorophyll, whose concentrations reduce toward the dam.
The Chavantes reservoir exhibits an additional spatial complexity due to the existence of “two longitudinal axes,” instead of one. An axis corresponds to the former Itararé River Valley (stations CH1, CH2, and CH3), laterally inserted into the Paranapanema River, the other axis (stations CHUp, CH4, and CHDam). In general, variables such as turbidity, suspended solids, total nitrogen, and total phosphorus (in this case only for wet season) exhibited lower values in the Paranapanema axis, compared to the Itararé axis. In both spatial axes, there was a transparency increase toward the dam, with higher values in the Paranapanema, 6.9 and 8.7 times in wet and dry seasons, respectively.
Different from the previously mentioned reservoirs, in Canoas I, Canoas II, Taquaruçu, and Rosana, all run-of-river systems (retention time between 4 and 17 days), the longitudinal gradient was moderate or even a relative homogeneous condition was found. For instance, in Taquaruçu during both campaigns and in Canoas II and Rosana during the wet campaign, the maximum difference along the reservoirs’ main axes was only 0.4 m, despite the distance of tens of kilometers between sampling points. For these reservoirs, the magnitude of changes seems to be more relevant on the temporal scale—alternating between dry (winter/early spring) and rainy (late spring/summer) periods. Nutrient concentrations (nitrogen and total phosphorus), for instance, are clearly higher in the wet season.
Very low, and probably transient, transparency in some sampling sites of Rosana and Capivara during the dry season can be attributed to atypical rain events in the upstream areas of both reservoirs immediately (2 days) before sampling. The sediment loads introduced by tributaries (Cinzas River in Capivara and Pirapó River in Rosana) resulted in a remarkable local increase in mineral turbidity (~155 NTU).
The results of Capivara, whose water retention time of 115 days is higher than in run-of-river and lower than in upstream storage reservoirs, demonstrate the importance of both spatial compartmentalization and seasonal variation on the physical and chemical characteristics. The spatial variability is determined by longitudinal gradient and tributary rivers’ (more eutrophic waters’) contributions. The concentrations of nutrients (nitrogen and phosphorus) varied widely among the different compartments of the reservoir. The main tributary of the Capivara reservoir is the Tibagi River (its mouth corresponds to the sampling site CP3), which drains agricultural areas intensively cultivated and is the second largest river in the whole basin, after the Paranapanema.
In the Salto Grande reservoir, despite its small size, the spatial complexity is high, due to the influence of a large reservoir (Chavantes) located upstream and the entrance of secondary tributary river (Pardo River), which introduces high loads of nutrients and suspended solids [1, 17, 37]. The transparency of water, for example, is higher in the upstream region (SGUp and SG2) and lower at the mouth of the Pardo River (SG1), with a difference of 3.3 m and 2.9 in wet and dry season, respectively. The opposite pattern of variation, as expected, was verified for turbidity and suspended solids, with values lower than 5 NTU at SGUp and SG2 (wet and dry seasons) and higher than 80 NTU in SG1 (wet season).
In addition to the river’s prevalent longitudinal pattern of physical, chemical, and biotic organization , the results of Capivara and Salto Grande are very important to show that the lateral dimension, mostly the effects of tributaries’ entrance, should not be neglected if one aims to understand the spatial structure and functional processes of large rivers and reservoir ecosystems.
7. Intra-reservoir spatial organization: vertical dimension
The development of thermal stratification, commonly followed by chemical differences, is a profound modification in riverine ecosystems after damming. Interactions with the atmosphere are enhanced due to the increase of the exposed surface area, higher water retention time, and reduction in advective transport of mass and energy. In the Paranapanema reservoir cascade, we observed a strong thermal stratification in Chavantes, the deepest reservoir, during summer. Maximum difference between epilimnion (25°C) and hypolimnion (18°C) layers, of 7°C, was verified at the sampling point near the dam (CH12). Other two sampling sites, CH3 and CH4, in the central channel also exhibited a well-defined thermal stratification. The oxygen along the water column decreased approximately from 3 to 4 mg L−1.
During summer, we also observed stratification at the dam zone of the other two storage reservoirs, Jurumirim and Capivara, with differences between layers of 3 and 2°C, respectively. In Jurumirim, particularly, there was a remarkable drop (about 5 mg L−1) of dissolved oxygen in deeper layers, certainly the effect of an extended period (late spring/summer) of stratification. The upstream compartments of these larger reservoirs (especially CHUp, CPUp) were characterized by homogeneous physical and chemical profiles, indicating a continuous mixture regime.
Surficial stratification, probably ephemeral, due to intense solar heating in summer was observed in some sampling sites (e.g., SGDam, SG3, CP3, CP4).
For the other reservoirs, run-of-river systems, isothermal conditions or a slightly gradual decrease of temperature with depth was observed, resulting in a homogeneous or relatively homogeneous distribution of the other variables measured along the water column.
8. Concluding remarks
As previously mentioned, this study has a “historical” importance, because since then it was established a permanent limnological and water quality monitoring program in the Paranapanema River reservoir cascade. Eight consecutive years of continuous and standardized evaluation produced valuable information, which has subsidized important management actions in this large water basin of southeast Brazil.
The larger (and dendritic) storage reservoirs exhibit a well-marked longitudinal compartmentalization, considerable water level fluctuation as well as summer thermal stratification in the central channel and lacustrine zones toward the dam, followed by oxygen decrease in bottom layers. Conversely, run-of-river reservoirs are morphometrically simpler, the spatial complexity is moderate with minor variation in the water level and a continuous mixing regime. Nevertheless, the second kind of reservoirs is less resilient to seasonal changes (dry-wet periods).
In addition to the expected upstream-dam gradients, the lateral component of the spatial heterogeneity was very important for some reservoirs, determined basically by the entrance of tributary rivers which transport considerable loads of nutrients and sediments from intensively cultivated lands.
Finally, it is important to highlight that most measurements were in conformity with the federal standard references for Class 1 waters (good water quality). Exceptions are locally restricted. This fact is corroborated by the predominant low trophic state (77% of ultraoligotrophic determinations). The Paranapanema River reservoir cascade is a strategic regional hydric resource. This is particularly important for the state of São Paulo, the most populous and industrialized state of the country, where important fluvial systems are hypereutrophic and heavily polluted  and the metropolitan area of São Paulo city is already facing recurrent water crises, in quantity and quality.
The authors are grateful to Dr. Norberto Castro Vianna, from Duke Energy, Geração Paranapanema, the company in charge of hydropower generation in the Paranapanema reservoir cascade (presently China Three Gorges, Brazil) that provides financial support and technical partnership. We also thank the graduate and undergraduate students that kindly helped during field work expeditions and laboratory analyses and the Foundation of the Biosciences Institute—State University of São Paulo (FUNDIBIO) for administrative support (contract 0100015003).
Conflict of interest
|Wet season||Secchi (m)||Deph (m)||Temp. (°C)||pH||Cond. (μS cm−1)||Turb. (NTU)||DO (mg L−1)||BOD (mg L−1 O2)||TS (mg L−1)||Inorg. SS. (mg L−1)||Org. SS (mg L−1)||TN (μg L−1)||TP (μg L−1)||Chl. a (μg L−1)||Thermo Coli (NMP/100 mL)|
|Dry season||Secchi (m)||Deph (m)||Temp. (°C)||pH||Cond. (μS cm−1)||Turb. (NTU)||DO (mg L−1)||BOD (mg L−1 O2)||TS. (mg L−1)||Inorg. SS (mg L−1)||Org. SS (mg L−1)||TN (μg L−1)||TP (μg L−1)||Chl. a (μg L−1)||Thermo Coli (NMP/100 mL)|
- See also Appendixes 1 and 2.