We developed and validated a Planktonic Index of Biotic Integrity for subtropical reservoirs to assess their ecosystem health. For this purpose, we analyzed the phytoplankton and zooplankton communities and determined reservoir trophic status in the Paranapanema River system (Southeast Brazil). Eleven dams were constructed in the main course of this river to supply hydropower plants. Three of the reservoirs are accumulation systems (i.e. with high water retention times), whereas the others are run-of-the-river systems. For the study the three larger reservoirs (Jurumirim, Chavantes and Capivara) were selected. Physical, chemical, and biological (phytoplankton and zooplankton) data were obtained in two sampling campaigns carried out in March (wet season) and October (dry season) of 2011. For each reservoir we sampled six stations, arranged on a gradient established between the lotic (Paranapanema River entrance) and lentic (dam) areas. According to the Trophic State Index for tropical/subtropical reservoirs, the sampling stations were categorized between ultraoligotrophic and mesotrophic. Four metrics achieved significant discrimination (out of a set of 20 candidate metrics). The individual metric scores were summed to provide a Planktonic Index of Biotic Integrity score, which ranged as Mesotrophic (4-9), Oligotrophic (10-14) and Ultraoligotrophic (15-20), corresponding to the classification of fair, good and excellent, respectively. Following the longitudinal sequence, Jurumirim was classified as Oligotophic (Good) and both Chavantes and Capivara as Ultraoligotrophic (Excellent). This study demonstrated that the Planktonic Index of Biotic Integrity is a potential tool for monitoring large subtropical reservoirs, as planktonic organisms are sensitive to environmental changes and this index integrates distinct temporal and spatial scales.
Brazil is a reservoir-orientated country, where most electricity production comes from dammed rivers. According to The World Bank (2014), 32 countries, including Brazil, use hydropower to produce more than 80% of their electricity requirements. Further, at least 3,700 major dams, each one with a capacity of more than 1 MW, are either planned or under construction in countries with emerging economies (Zarfl et al., 2015).
Construction of large reservoirs for hydropower generation during the last decades occurred throughout the country (about 200 dams in operation and 200 under construction). These human-designed environments are particularly common in the Southeast region, affecting most rivers and deeply changing the surrounding landscapes. Besides the relatively clean and renewable energy production, additional positive aspects related to reservoirs are their usage for water storage, flood control, recreation for residents, tourism, and other economic opportunities, such as fisheries and aquaculture (Tundisi and Matsumura-Tundisi, 2003). Among the negative impacts are the considerable changes in the rivers’ biota (Agostinho et al., 2008; Nogueira et al., 2008; Nogueira et al. 2010).
Studies on reservoir limnology indicate that they constitute a particular class of aquatic environment due to the dynamic interaction between riverine and lacustrine compartments (Thornton et al., 1990; Armengol et al., 1999; Kennedy et al., 2003). This distinctive pattern of spatial organization has been evidenced for several large Brazilian reservoirs (Nogueira et al., 1999; Pinto-Coelho et al., 2006; Soares et al., 2008; Perbiche-Neves et al., 2011). The temporal and spatial complexity is even higher in the case of reservoir cascades (i.e. distributed in series and arranged in a cascade system). Despite some accumulated ecological information (Barbosa et al., 1999; Jorcin and Nogueira, 2005a, b; Nogueira et al., 2008; Naliato et al., 2009; Nogueira et al., 2010; Nogueira et al., 2012; Perbiche-Neves et al., 2011; Matsuura et al., 2015), research efforts are still necessary to understand the limnological changes along the river continuum (Vannote et al., 1980), including the structure and functioning after construction of series of dams - upstream and downstream transference effects (e.g. downstream exportation of low oxygenated water (deep located turbines), export of algae biomass, upstream nutrient and solids retention, etc. (Matsuura et al., 2015; Portinho et al., 2016).
Integrity of a given ecosystem can be assessed through the diagnosis of biological attributes or indicators, which ideally are sensitive to a range of stresses, able to distinguish stress-induced variation from natural variation, relevant to society concerns, and easy to measure and interpret. The complexity of biotic systems dictates that integrity assessments should incorporate a variety of indicators (including elements and processes) from multiple organizational levels and spatiotemporal scales (Angermeier and Karr, 1994).
The Index of Biotic Integrity (IBI) (Karr, 1981) is an ecologically based multi-metric index for assessing the biological integrity of surface waters. It considers distinct biotic attributes, ranging from individual to ecosystem-level properties. The IBI was originally developed in the 1980s and used fish assemblages as an indicator of aquatic ecosystem health (Karr, 1981; Karr et al., 1986). This tool has been adapted and modified for evaluation of aquatic ecosystem health worldwide. A variety of organisms have been used like littoral zone plants (Rothrock et al., 2008), benthic macroinvertebrates (Fore et al., 1996, Li et al., 2010), aquatic insects (Silva et al., 2010), benthic diatom communities (Wu et al., 2012a), phytoplankton (Gómez et al., 2012; Wu et al., 2012b; Li et al., 2013), zooplankton (Carpenter et al., 2006) and combined phyto- and zooplankton (Kane et al., 2009; Kane et al., 2015). IBIs have been applied to different aquatic environments such as rivers (Karr, 1981; Wu et al., 2012a; Casatti et al., 2009; Esteves and Alexandre, 2011), estuaries (Carpenter et al., 2006; Gómez et al., 2012), lakes (Kane et al., 2009), reservoirs (Wu et al., 2012a), and reservoir cascades (Li et al., 2013) as well.
Although IBIs have been used for many purposes, there is a unique study focusing on the impact of cascading dam construction, which includes metrics related to phytoplankton assemblages (Li et al., 2013). Values from this index agreed with the pattern of increased abundance and biomass of phytoplankton assemblages in reservoir areas and provided evidence of aquatic ecosystem degradation (when compared with natural riverine stretches).
All components of reservoir functioning can be influenced in major ways by the dynamics of the phytoplankton and zooplankton, and the plankton communities (e.g. chlorophyll a concentrations) are directly related to nutrient/trophic status (i.e. total phosphorus and eutrophic-oligotrophic gradients). Therefore, phytoplankton and zooplankton dynamics have a considerable impacts on aquatic ecosystem and, consequently, on humans who interact with these environments (Kane et al., 2009).
With the purpose of effectively assessing the ecosystem health of subtropical reservoirs, we developed and validated a Planktonic Index of Biotic Integrity (P-IBI) for these systems. The study is based on a reservoir cascade system located in Southeast Brazil and includes metrics of the entire plankton community – phyto and zooplankton. In contrast to traditional water quality approaches, the P-IBI is an aggregative indicator that can not only capture aquatic trophic status but also identify variations in the aquatic ecosystems associated with the biota. The development and application of a viable Planktonic Index of Biotic Integrity (P-IBI) for subtropical reservoirs can be useful for management purposes – stakeholder’s decision processes, improvement of monitoring protocols, and expansion of scientific knowledge.
Following Kane at al. (2009), our study included five goals: 1) to develop and implement sampling protocols; 2) to develop a multimetric Planktonic Index of Biotic Integrity (P-IBI) for subtropical reservoirs; 3) to validate the P-IBI for subtropical reservoirs statistically; 4) to apply the P-IBI for subtropical reservoirs using plankton selected data sets (2011 – March and October) from a Water Quality Monitoring Program carried out in the Paranapanema River reservoirs cascade and 5) to disseminate the results of the analyses to citizens and policy makers.
Study area and site locations
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 (Southeast/South Brazil). The river is the natural border between the states of Paraná and São Paulo (Fig. 1), with a total length of 929 km. Since the 1950’s, eleven hydropower plants have been constructed in the main river course. Three of the reservoirs are accumulation systems (i.e. with high water retention times), whereas the others are run-of-the-river systems. For this study the three larger storage (accumulation) reservoirs (Jurumirim, Chavantes and Capivara) were selected, based on the criteria that they are more lake-like which allows for seasonal succession of plankton assemblages. These three reservoirs have high shoreline development (>15), high retention time (≥150 days), and are relatively deep (>30 m near to the dam) (Table 1).
For each reservoir we determined six sampling stations including the main spatial compartments identified from previous studies (Nogueira et al., 1999; 2012), which are arranged on a gradient established between the lotic (Paranapanema River entrance) and lentic (dam) areas (Fig. 1).
Sampling campaigns for physical, chemical and biological (phytoplankton and zooplankton) measurements were carried out during two periods of the year, March 2011, corresponding to the end of the wet season, and October 2011 corresponding to the end of the dry season – the most contrasting seasonal periods.
Limnological analyses and Trophic State Index calculation
Water samples for total phosphorus and chlorophyll a analysis were collected with a Van Dorn bottle in three depths: surface, middle and bottom of the water column. Samples for total phosphorus were acid digested (Valderrama, 1981) and then analyzed spectrophotometrically (Strickland and Parsons, 1960). Total chlorophyll a concentration was determined in replicates after vacuum filtration (Millipore AP40 membranes) of 1 L of water from each considered depth. For pigments extraction it was used cold acetone (90%) after manual maceration (Talling and Driver, 1963; Golterman et al., 1978).
For both, TP and Chl a, average values among depths of each sampling station was used to calculate the Trophic State Index to distinguish among trophic status classes.
The TSItsr takes into consideration 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).
Phytoplankton and zooplankton samples
For phytoplankton, at each sampling station, an integrated sample was collected for qualitative analysis (entire water column) through vertical net hauls (20-μm mesh size) and immediately preserved in 4% formalin. The net samples were observed in an optical microscope (maximum magnification of 1000×) for taxonomical identification and determination of assemblage total richness. For phytoplankton quantitative analysis, three unfiltered samples were collected (van Dorn bottle) at the subsurface (ca. 0.2 m), middle of the water column and near to the bottom (ca. 1 m above the sediment). The samples were fixed and preserved with Lugol’s solution. After sedimentation, the organisms (cell, colony, and filament) were counted using inverted microscopy (sensu Utermöhl, 1958) at a magnification of 400×. At least 120 optical fields distributed in parallel transects were examined, and at least 150 organisms were counted per sample. The quantitative data were expressed as mean values for the water column.
The zooplankton samples were collected using a conical net (30 cm mouth diameter and 50 μm mesh size) and vertical hauls from near bottom (ca. 1 m) to the surface. In each site/campaign an additional sample for qualitative analysis was collected. Samples were fixed and preserved in 4% formalin. For the quantitative analyses, most organisms were counted at species level using sub-samples. Rotifera, and nauplii of Copepoda were counted in Sedgwick–Rafter chambers, in optic microscope Zeiss Standard 25 (at a magnification of × 200); and Cladocera, copepodites and adult stages of Copepoda were counted using a stereo microscope Zeiss Stemi SV 6 (maximum magnification of × 120). At least 150 specimens were counted per sub-sample. Additional sub-samples, or even the entire sample, were analyzed when the density of organisms was low (generally less than 100 organisms per 5 ml of sample, in case of Cladocera and Copepoda, and less than 100 organisms per 1 ml of sample, in case of Rotifera). The specimens of the main zooplankton groups (Rotifer, Copepod, and Cladocera) were analyzed at the lowest taxonomic level. The densities of the organisms, expressed as numbers of individuals per cubic meter, were calculated using the volumes of filtered water and of the concentrated samples.
Developing a Planktonic Index of Biotic Integrity (P-IBI) for subtropical reservoirs
Metric selection and statistical validation
A number of plankton characteristics are directly related to the assemblage’s structure and composition and reflect the status of a range of environment attributes. It is well known from scientific studies carried on along more than a century worldwide that plankton features are susceptible to anthropogenic influences and can indicate distinct levels of eutrophication. Based on previous experience on developing the P-IBI (Kane et al., 2009) as well as on the accumulated regional knowledge (e.g. Nogueira et al., 2008; Nogueira et al., 2009; Perbiche et al., 2011; Perbiche et al., 2016) 20 candidate metrics were considered and included in the discriminant analysis to be used for the multimetric index for tropical and subtropical reservoirs: total calanoid density; total cyclopoid density; total cladocera density; total rotifer density; total crustacean density; zooplankton ratio calanoid/cladocera + cyclopoid; zooplankton ratio calanoid/cyclopoid; zooplankton richness; zooplankton diversity; zooplankton evenness (J'); total phytoplankton density; % cyanobacteria; abundance of the most common Cyanobacteria genera Anabaena (Dolichospermum), Aphanizomenon, and Microcystis; % of Anabaena (Dolichospermum), Aphanizomenon and Microcystis; edible phytoplankton density; inedible phytoplankton density; abundance of the invasive species Ceratium cf. furcoides; phytoplankton richness; phytoplankton diversity, and phytoplankton evenness (J').
Total phosphorus concentration (µg l−1), and total chlorophyll a concentration (µg l−1) were used to classify the sites TSItsr (from ultraoligotrophic to hypereutrophyc as proposed by Cunha et al., 2013). We then judge the ability of plankton metrics to distinguish among trophic state classes, reflecting levels of degradation as proposed by Kane et al. (2009).
Discriminant analysis (DA) was used to evaluate the plankton metrics responses. DA discriminates among pre-specified groups of samples based on a set of variables to find gradients among groups of samples, then variation among groups is maximized, while within group variation is minimized (McGarigal et al., 2000). Discriminant analysis has been identified as an acceptable statistical method for the development of Indices of Biotic Integrity and can be used to identify variables that discriminate between levels of degradation (USEPA, 1998). We performed discriminant analyses (Statistica, version 7.0, 2006) to give a broad range of conditions in both trophic status and in the candidate metrics. All 20 metrics and sites were included in the discriminant analysis.
Four out 20 metrics were selected through the discriminant analysis. In order to calculate individual metric scores we constructed ‘‘boxplots’’ of the significant individual plankton metric’s frequency distributions. We used the 95th percentile as the upper boundary and zero as the lower boundary (Karr et al., 1996) and trisected each of the final individual metrics included in the multimetric P-IBI for subtropical reservoirs (based on significance in the discriminant analyses) into ranges that were assigned a score of 1, 3, or 5 to match those values assigned to the TSItsr condition. A 5, represented the better environmental condition (ultraoligotrophic) range of trisection, while a 1, was the most degraded (mesotrophic) range of the trisection. The statistically significant metrics for each site were summed to provide a P-IBI score and a classification applicable for subtropical reservoirs, in this case, ranging from Fair to Excellent, or Mesotrophic (4-9), Oligotrophic (10-14) and Ultraoligotrophic (15-20), respectively.
We performed two steps to calculate the P-IBI: 1) Use cutoff scores for each variable to calculate individual metric values and 2) Estimate a subtropical reservoir mean metric score.
Where: CAijk= Total Calanoid density metric score; Cyijk= Total Cyclopoid density metric score; PRijk= Phytoplankton richness metric score; PDijk= Phytoplankton diversity metric score; M = number of metrics; S = number of sites and R = number of reservoirs.
Finally, a weighted Cohen’s Kappa statistic (k) (Cohen, 1960) was calculated in order to judge the accuracy of classification. Cohen’s Kappa is a statistical measure of the agreement of two raters or two rating methods. In our case, we compared the trophic status determined by TSItsr and the developed P-IBI for subtropical reservoirs. Significance for the weighted Cohen’s Kappa was judged α = 0.05 for hypothesis testing. A Cohen’s Kappa of 1 indicates perfect agreement between the raters and 0 indicates that any agreement is totally due to chance. There is no clear-cut agreement on what constitutes good or poor levels of agreement, although a common set of criteria is: <0.00 = poor, 0.00-0.20 = slight, 0.21-0.40 = fair, 0.41-0.60 = moderate, 0.61-0.80 = substantial, 0.81-1.00 almost perfect (Landis and Koch, 1977).
Case studies in Brazil (Tundisi and Matsumura-Tundisi, 2008) demonstrated how the services provided by reservoirs diversified from a single use (hydroelectricity) to a multiple complex matrix of activities such as fish and fisheries, recreation, navigation, tourism, irrigation and industrial, besides hydropower production.
The demands of the regional/local societies for ecological services are also relevant for the diversification of the services and an easy and efficient communication between managers and the residents is necessary, so an important goal of IBI development is to communicate the biotic integrity results to a variety of different groups of stakeholders (Karr and Chu, 1997). This group ranges from scientists and managers to citizens and policy makers, interested on the management of the reservoirs water quality.
Further, we graphed the data as simple bar plots of P-IBI scores and also determined qualitative categories of the P-IBI for subtropical reservoirs that reflect the level of degradation scoring system (i.e. poor, fair, good, excellent) and are effective for summarizing reservoir biological integrity.
Phytoplankton and zooplankton ecological attributes of the community
The phytoplankton community was composed of 198 taxa found in Jurumirim, 179 in Chavantes and 191 in Capivara, distributed respectively into the classes: Bacillariophyceae (≈30%); Chlorophyceae (≈24.8%); Cyanophyceae (≈19%); Euglenophyceae (≈18%), Zygnemaphyceae (≈5.3%); Chrysophyceae (≈2.3%); Cryptophyceae (≈2.1%); Dinophyceae (≈1.4%); Oedogoniophyceae (≈0.5%); Rhodophyceae (≈0.4%) and Ulotrichophyceae (≈0.2%).
The highest density occurred in JR2, 2,014,969 ind. l−1, CH1, 1,262,734 ind. l−1 and CP2, 1,467,902 ind. l−1 in March, while the lowest occurred at the JRUp, 688,225 ind. l−1, CHDam, 586,626 ind. l−1 and CPUp, 425,342 ind. l−1. In October the highest density occurred at station JRUp, 988,051 ind. l−1, CH1, 1,262,734 ind. l−1 and CPDam, 1,641,923 ind. l−1 while the lowest occurred at the JR4, 504,086 ind. l−1, CHUp, 505,203 ind. l−1 and CP2, 316,744 ind. l−1. In general, the highest density corresponds to the class Cryptophyceae, but in the JRDam was observed 128,850 ind. l−1 corresponding to the class Cyanophyceae. In this sampled station we can assume that there was a "bloom" of cyanophytes in March. We also had a relatively high density of cyanophytes at CPDam, CP4, CP3 and CP1in March.
For diversity we observed the higher values in the JRDam, 2.69 bits ind.−1, 0.633, CHUp, 2.46 bits ind.−1 and CPDam, 2.28 bits ind.−1 and lower in the JR4 station, 1.97 bits ind.−1, CHDam, 1.74 bits ind.−1and CPUp, 1.28 bits ind.−1 in March while in October higher diversity occurred at the stations JR1, 2.90 bits ind.−1, CH2, 2.11 bits ind.−1 and CP4, 2.58 bits ind.−1 and lower at stations JR2, 1.97 bits ind.−1, CHDam 1.87 bits ind.−1 and CP1, 1.37 bits ind.−1. The phytoplankton evenness ranged between 0.533 JR4 and 0.633 JRDam; 0.519 CHDam and 0.707 CH2, and 0.486 CP1 and 0,805 CPUp in March while in October between 0.517 JR2 and 0.660 JR3, 0.562 CHDam and 0.666 CH4 and 0.536 CPUp and 0.671 CPDam. The edible and inedible phytoplankton density was quite variable ranging from 303,170 ind. l−1 (CP2) to 1,777,911 ind. l−1 (JR2) for edible and from 0 (JRUp, CHDam, CHUp and CP2) to 237,055 ind. l−1 (JR2) for inedible. For the abundance of the invasive species Ceratium cf. furcoides, we were able to quantify in only a few sampling stations and with insignificant values, only one with a considerable quantity, 103,067 ind. l−1 (JR 2).
The zooplankton community was composed of 89 taxa represented by Rotifera with 52 taxa (58%), Copepoda with 13 taxa (15%), of which 5 taxa belonging to the Order Calanoida and 8 of the Order Cyclopoida and Cladocera with 24 taxa (27%). The most frequent species were Collotheca spp.; Conochilus coenobasis; Conochilus unicornis; Kellicottia bostoniensis; Ploesoma truncatum; Polyarthra spp.; Synchaeta spp. and Synchaeta stylata; among the rotifers; Notodiaptomus cf. spinuliferus; Notodiaptomus henseni and among Copepoda (Calanoida), Thermocyclops decipiens (75%) among Copepoda (Cyclopoida) and the Cladocerans Bosmina freyi, Bosmina hagmanni, Bosminopsis deitersi, Ceriodaphnia cornuta f. cornuta, C. cornuta f. rigaudi, Daphnia gessneri, Diaphanosoma birgei and Diaphanosoma spinulosum, all presenting 100% frequency of occurrence.
The total density highest values were observed in CP4 with densities >25,000 ind. m3 for total Calanoid and >15,000 ind. m3 for total Cyclopoid; CH1 with >15,000 ind. m3 for total Cladocera and JR1 with > 112,000 ind. m3 for total Rotifera. The total number of crustaceans was higher in the JR sampling stations ranging from >17,000 at JR1 in October up to >73,000 in March. All densities were higher during the wet season. The zooplankton ratio calanoid/(cladocera + cyclopoid) and zooplankton ratio calanoid/cyclopoid were highly variable with values between 0.002 in CH4 and CHDam and 8.23 in CPUp, both also in March.
The species richness ranged from 12 in CH in October and 18 in CP, March and JR and CP in October for Rotifers; between 7 in JR March and CP October and 9 in JR October and CH March for Copepoda and between 14 in CH October and 32 in JR in October for Cladocera. The diversity ranged from 2.39 bits ind.−1 at CHDam in October to 4.23 bits ind.−1 at CP1 in March and the evenness between 0.520 at JR2 and 0.870 at CP1, both in March.
Reservoirs trophic state condition and Planktonic Index of Biotic Integrity
The selected Paranapanema River reservoirs exhibited low or relatively low concentrations of TP (varying from 3.1 µg l−1 in Jurumirim to 67.2 µg l−1 in Capivara) and Chl a (varying from 0.4 µg l−1 in Chavantes to 7.0 µg l−1 in Capivara) (Fig. 2). The observed concentrations maintained the range used on developing P-IBI from ultraoligotrophic to mesotrophic status (Table 2), which was supported with the final classification from fair to excellent.
The plankton community metrics seem to correspond to the trophic conditions of the selected reservoirs. The results of the P-IBI Score highlight the good water quality in the Paranapanema River for the selected reservoirs (Fig. 3-A). All of them could be classified in Ultraoligotrophic condition except for Jurumirim in March, which was classified as Oligotrophic. These results fairly correspond with the values obtained through the Trophic State Index for tropical/subtropical reservoirs as presented in Table 2. There is also a good correspondence with the classification proposed by the P-IBI for subtropical reservoirs, with scores ranging from Good in Jurumirim reservoir and Excellent in the Chavantes and Capivara reservoirs.
Figure 3 (B to D) shows the P-IBI for subtropical reservoirs for each one of the selected reservoirs and sampling sites. Variations between seasons and also among sampling sites support the spatial and temporal complexity of reservoirs. The score values ranged between Good and Excellent in the distinct reservoirs and were higher and more homogeneous in Chavantes (Fig. 3-C), followed by Capivara (Fig. 3-D), and then Jurumirim (Fig. 3-B), with the highest spatial and temporal variation.
Large scale reservoir construction has economic and social consequences, in addition to environmental impacts (Redman et al., 2004), which justifies studies focusing on reservoir water quality and ecosystem health assessments. Kay and Schneider, (1991) use the term ‘‘ecosystem integrity’’ to refer to the ability of an ecosystem to maintain its organization.
The development and application of a viable Planktonic Index of Biotic Integrity (P-IBI) can be useful for monitoring protocols and management purposes of the Paranapanema River reservoirs. The Paranapanena River has been considered as a fluvial system that preserves a relatively good “water quality” condition, with frequent classification of its reservoirs as oligotrophic or oligo-mesotrophic (Jorcin and Nogueira 2005a, b; Nogueira et al., 2008; Jorcin and Nogueira, 2009; Nogueira et al., 2010; Henry, 2014, Pomari et al., 2018). These reservoirs are considered as strategic for the environmental policy of the State of São Paulo, the most populous and industrialized in the country, where deterioration of inland water resources is of great concern.
The results obtained by Pomari et al., (2018), comparing multiple-use indices and diverse approaches to assess reservoir water quality, reaffirmed the resilience of this ecosystem in maintaining its good water quality. Different responses were obtained with the application of distinct water quality/trophic state indices - the Water Quality Index - WQI results exhibited Excellent conditions for 70% of the determinations during the wet season and 95% in the dry season; TSI for tropical/subtropical reservoirs resulted as Ultraoligotrophic, in 77% of the determinations, followed by Oligotrophic, 17.6%. The Phytoplankton Community Index classified in the categories Excellent, 52%, and Good, 48%; the Zooplankton Community Index, with 73% Good, 22% Regular, and 5% Bad and P-IBI with Excellent, 75% and Good, 25%.
In the P-IBI development process metric selection is essential to achieve an effective index. The measures of integrity, according to Kay and Schneider, (1991) should reflect the two aspects of the organizational state of an ecosystem: function and structure. For this reason, all the selected metrics in this study agree with the seven levels suggested by Sven et al., (2005): (1) application of specific species; (2) ratio between classes of organisms; (3) specific chemical compounds; (4) trophic levels; (5) rates; (6) composite indicators included in E.P. Odum’s attributes and various indices; (7) holistic indicators as, for instance, biodiversity and resistance; (8) thermodynamic indicator.
Studies carried on along the Paranapanema River reservoir cascade have shown that phytoplankton and zooplankton abundances are positively associated along the cascade. It has also been demonstrated that their composition and structure, even in major taxonomical categories, can be good indicators of distinct trophic conditions in the reservoir cascade (Nogueira et al., 2008, Sartori et al., 2009; Nogueira et al., 2010; Perbiche-Neves and Nogueira, 2010).
In general, the phytoplankton community did not exhibit any "bloom" events of cyanophytes (cyanobacteria), or of any other class of algae, but rather punctuated and sporadic occurrences of higher densities. In general, the density values were relatively low. The predominance of species of the class Cryptophyceae is characteristic of reservoir systems (Nogueira et al., 2010). It is also worth mentioning the presence of the Ceratium furcoides, an invasive species, which, despite a considerable increase in recent years, still did not present high densities that could result in the formation of "blooms".
The zooplankton community was composed of 89 taxa represented by Rotifera with 52 taxa (58%), Copepoda with 13 taxa (15%), of which 5 taxa belonging to the Order Calanoida and 8 of the Order Cyclopida, and Cladocera with 24 taxa (27%). In addition, taxa considered tolerant were not typically prevalent. This could be attributed to inherent variety of transient conditions involving hydrological, limnological, and biological features, typical of reservoir ecosystems. An additional point to be considered is that in this basin the range of trophic conditions is not very variable. Zooplankton are valuable as indicators of changes in water quality and trophic conditions and can thus give us signs that the environment is undergoing changes before it become drastic or irreparable.
Along with TSItsr, four metrics were selected by the discriminant analysis (DA) as suitable to distinguish among trophic status classes, reflecting levels of degradation as proposed by Kane et al., (2009). The determination of the Trophic State Index (TSI) is a traditional approach for evaluation of the water quality in freshwater systems. The TSI is based on limnological variables that are relatively simple to measure, easy to calculate, and simple to understand and explain. In our study we adopted the regional proposal of Cunha et al. (2013) for tropical/subtropical reservoirs (TSItsr) which considers relevant ecological aspects for an appropriate assessment, including geographic positioning (tropical/subtropical region) and it was already tested for a large number of reservoirs of São Paulo State for validation purposes.
The variation of TSItsr for the selected Paranapanema River reservoirs was relatively low. The classification varied between Ultraoligotrophic and Mesotrophic, with the prevalence of the first condition (Table 2). These results are in accordance with other studies carried out in this watershed (i.e Pomari et al., 2018), and also with the P-IBI we have developed. In our study the reservoir scores (mean values) were very close and comprised the categories Meotrophic, Oligotrophic, and Ultraoligotrophic and the reservoirs were classified either as ultraoligotrophic or oligotrophic (qualitative categories Excellent or Good) (Figure 3).
The agreement between the two considered methodological approaches, TSItsr and the developed P-IBI, was “slight” (Cohen’s Kappa statistic) (Table 3), which was not unexpected given the low number of samples collected in a single year and the relative stability of this ecosystem. Nevertheless, there is the possibility to revise and improve the P-IBI in future analyses, as systematic monitoring of Jurumirim, Chavantes and Capivara reservoir continue.
Few IBIs have been developed for reservoir cascades. However, previous to our study, a phytoplankton IBI was developed for a reservoir cascade in the Lancang-Mekong River, Southwest China (Li et al. 2013). The index used phytoplankton metrics (but not zooplankton metrics) and the results corroborated the known longitudinal spatial pattern. Variability was mainly related to increasing abundance and biomass of phytoplankton assemblages in reservoir areas.
Despite using data of only two sampling periods, it is important to consider that the selected periods (March and October 2011) are representative of the regional environmental variability, as they correspond to the most contrasting seasonal periods - end of summer/rainy season and end of winter/dry season, respectively. The type of data used to develop our P-IBI is comparable with those of Kane et al., (2009) and other authors who developed IBIs (i.e. Carpenter et al., (2006), Rothrock et al., (2008), et al., (2009), Li et al., (2010), Silva et al., (2010), Esteves and Alexandre, (2011), Gómez et al., (2012), Wu et al., (2012a), Wu et al., (2012b), Ruaro et al., (2013), Li et al., (2013), Kane et al., (2015)).
The ideas proposed by Karr (1981) have contributed to the management of water resources and biological conservation (Ruaro et al., 2013). The original Karr IBI is one of the most frequently used efficient assessment of freshwater ecosystem integrity. It allows the identification of priority sites for conservation (Lyons et al., 1995), integrates measurements of biological condition and associated resources that are easily understood by public and official agencies, and allows the comparison of both individual and cumulative effects of a variety of human activities (Karr and Chu, 2000).
The P-IBI model proposed in this study incorporates direct evaluation of phytoplankton and zooplankton communities, making it a more robust tool to assess the ecosystem health and improve monitoring and management of Brazilian reservoirs.
This study demonstrated that the P-IBI is a potential tool for monitoring programs of large subtropical reservoirs, and reflects differences in ecosystem health over space and time. The use of metrics associated with plankton assemblages are appropriate since these organisms are sensitive to environmental changes and integrate distinct temporal and spatial scales.
P-IBI scores can be understood and used by nonscientists involved in the making, planning, and management decisions at an appropriate level for the multiple uses of subtropical reservoirs. This P-IBI generally agrees with previous evaluations of the study reservoirs and can be easily integrated into the water quality monitoring program of the Paranapanema River reservoir cascade.
The authors would like to thank CAPES Foundation for the first author financial support (PDSE fellowship proc.002422/2015-08), Defiance College, and Dr. Robert Michael McKay (BGSU) for an international internship opportunity and access to the studies in which this article was based on. The BGSU immigration office, especially Maorong Lancaster and Márcia Salazar-Valentine for the documentation support and for everything they did to facilitate a study and life abroad experience. We also thank The Ohio State University’s F.T. Stone Laboratory for computer and library access. Finally, we thank Dr. Raoul Henry and other researchers involved in the Limnological and Water Quality monitoring program of the Paranapanema River reservoir cascade, and Duke Energy Generation Paranapanema and Department of Zoology – State University of São Paulo for supporting laboratory analyses and field work.