Introduction

Temporal variability, structure and dynamics of phytoplankton community are of utmost importance to aquatic ecosystems’ metabolism. As responses to the interactive processes of physical, chemical and biological variables, abundance and specific composition reorganization of the phytoplankton community are frequently observed.

Phytoplankton community has already been very adequately identified for detection of sporadic events and also long term variations (Barbosa & Padisák, 2004) not only regarding species composition, but also in terms of biomass. Use of species or above species phytoplankton categories for water quality determination has already a long history (Padisák et al., 2006).

Reynolds et al. (2002) proposed a functional classification scheme of phytoplankton species, which sorts species with morphological, physiological and ecological similarities into functional groups labelled by alpha-numeric characters. Many studies have already proved the explanatory and predictive manner of functional groups (e.g. Marinho & Huszar, 2002; Kruk et al., 2002; Mischke & Nixdorf, 2003; Albay & Akçaalan, 2003; Morabito et al., 2003; Nixdorf et al., 2003; Naselli-Flores & Barone, 2003; Naselli-Flores et al., 2003; Mischke, 2003; Rojo & Álvarez-Cobelas, 2003; Silva, 2004; Crossetti & Bicudo, 2005; Fonseca & Bicudo, 2008) by describing the abiotic characteristics of the environments in which they occur. Following this approach, tests using phytoplankton have been developed to determine the water quality (Padisák et al., 2006).

Padisák et al. (2006) developed the Q index that provides 5 degrees of water quality through phytoplankton functional groups (Reynolds et al., 2002). Q index was developed to assess the ecological status of different lake types established by the Water Framework Directive (2000; WFD hereafter) proposed by the European Community. WFD is a broad concept that aims at developing sustainable management strategies for ground and surface waters in Europe (Padisák et al., 2006). The index takes into account the relative weight of functional groups in the total biomass, as well as a factor number determined for each type of water body. Although the proposal is very recent and until now no study has been carried on to show the Q index utilization, it can be applied for environment ecological evaluation with no geographic limitation (Padisák et al., 2006).

Differently from any other existing one, the Q index of functional groups gives no preference to any kind of human impact in particular, which considerably increases its application spectrum (Padisák et al., 2006). Finally, the Q index is also adequate to test monitoring ecosystems based on habitat concepts (Padisák et al., 2006).

The present study aimed at testing the Q index as a tool for water quality evaluation using a phytoplankton long term series (8 years) in Garças Reservoir, thus contributing to the discussion on phytoplankton as an indicator group of water quality.

Study area

Garças Reservoir is located in the Parque Estadual das Fontes do Ipiranga Biological Reserve located in the southeast region of the municipality of São Paulo (23°38′08″S and 23°40′18″S; 46°36′48″W and 46°38′00″W). Its mean altitude is 798 m and the total area is 526.4 ha (Fernandes et al., 2002). The reserve is one of the few remaining patches of Atlantic Rain Forest in the midst of a densely urbanized region of the city of São Paulo and one of the largest metropolitan green areas in Latin America.

Although locally called Garças Pond, the ecosystem is, in fact, a reservoir recently classified as eutrophic (Bicudo et al., 2002b). With its 4.7 m maximum depth, the reservoir is the most man affected ecosystem in the reserve. Its surface area is 88,156 m2, maximum length is 512 m, maximum width is 319.5 m and the mean retention time is 71 days (Bicudo et al., 2002a). Garças Reservoir has a single exit and seven tributaries, four of which carry sewage without any kind of treatment and only one is considered oligotrophic (Carmo et al., 2002).

The Garças Reservoir’s mixture type is discontinuous warm polymictic according to Lewis’ (1983) model (Bicudo et al., 2002b).

A recent temporal series (8 years) study of limnological data of Garças Reservoir identified 3 phases characterized by their physical, chemical and biological features, considering the water hyacinth cover influence on the system (Bicudo et al., 2006).

  1. 1.

    Phase I (from January 1997 to March 1998): characterized by a gradual increase of the water hyacinth cover, which reached 10–20% of the water surface, with cyanobacterial blooms restricted to the spring (Fonseca & Bicudo, 2008), and high chlorophyll a and pH values in the surface (Bicudo et al., 2006).

  2. 2.

    Phase II (from April 1998 to August 1999): characterized by the rapid macrophyte proliferation, that covered 40–70% of the reservoir surface, favored by the increasing of allochthonous nutrients input (N, P). Extensive macrophyte banks accounted for a very serious mosquito proliferation (Mansonia sp.) that called for mechanical removal of the macrophytes. This phase was characterized by low algal biomass, increasing transparency and DO (dissolved oxygen), SRP (soluble reactive phosphorus) decrease and the increased physical stability of the ecosystem (Bicudo et al., 2006).

  3. 3.

    Phase III (from September 1999 to December 2004): characterized by the water hyacinth cover reduced to only 10% of the water surface, retained by wire screens. Removal totalled 3,100 m3 of the macrophyte (Bicudo et al., 2007). During this phase, an abrupt limnological change in Garças Reservoir was registered, represented by an increase of chlorophyll a, pH, TP and SRP values, and a dramatic decrease of transparency, free CO2 and low DO values, mainly at the deepest layers of the reservoir (Bicudo et al., 2006).

Material and methods

Samplings were performed monthly over eight consecutive years (1997–2004), at five depths (subsurface, 1, 2, 3 m and ~20 cm from the bottom) always in the morning at the deepest site of reservoir. Water samples (n = 2) were collected with a van Dorn sampler and transferred to acid-rinsed bottles. Samples were filtered for the dissolved nutrients analyses under low pressure (< 0.3 atm) using Whatman GF/F membrane filters. In the field, water temperature, pH and conductivity were measured using standard electrodes. Water transparency was determined using a Secchi disk.

The following variables were also determined at the sampling day: dissolved oxygen (DO) (Winkler modified by Golterman et al., 1978), alkalinity (Golterman & Clymo, 1971), free CO2, HCO3 − and CO −3 (Mackereth et al., 1978), soluble reactive phosphorus (SRP) and total dissolved phosphorus (TDP) (Strickland & Parsons, 1960), total phosphorus (TP) (Valderrama, 1981), nitrite (NO2 −) and nitrate (NO3 −) (Mackereth et al., 1978), ammonium (NH4 +) (Solorzano, 1969), total nitrogen (TN) (Valderrama, 1981) and soluble reactive silica (SRS) (Golterman et al., 1978).

Trophic state index values were taken from Bicudo et al. (2006). Indexes used were Carlson’s Trophic State Index (TSI) modified by Toledo (1990) and Lamparelli (2004). The two latter authors adapted the original Carlson’s TSI for tropical systems. Toledo (1990) basically promoted alteration on the transparency, chlorophyll a and total phosphorus equations, considering them equivalent in the mean TSI. Lamparelli (2004) altered only chlorophyll a and total phosphorus TSI equations. Lamparelli (2004) used a large database supplied by CETESB (State of São Paulo Environmental Technology and Sanitation Company), which refined the index (Bicudo et al., 2006) and was, consequently, used for the Garças reservoir trophic evaluation.

Biological data from the year 1997 were taken from Fonseca & Bicudo (2008).

For determination of chlorophyll a concentration corrected for phaeophytin, pigment extraction was accomplished with ethanol 90% as an organic solvent (Sartory & Grobbelaar, 1984; Wetzel & Likens, 2000; Pápista & Böddi, 2002; Wasmund et al., 2006; Huang & Cong, 2007).

Phytoplankton, and especially cyanoprokaryotes, were identified with the possible most recent taxonomic literature (e.g. Azevedo et al., 1996; Azevedo & Sant’Anna, 2003; Komárek & Anagnostidis, 1986, 1989, 1999, 2005; Komárek & Azevedo, 2000; Komárek & Fott, 1983; Komárková-Legnerová & Cronberg, 1994; Sant’Anna & Azevedo, 2000; Sant’Anna et al., 1989).

Phytoplankton quantification followed Utermöhl (1958) and sedimentation time was set according to Lund et al. (1958). Biomass (mg l−1) was estimated using the biovolume calculated values and multiplying each species’ density by the mean volume of its cells considering, whenever possible, the mean dimension of 30 individuals of each species following Sun & Liu (2003), Hillebrand et al. (1999) and Fonseca (2005).

Biological data were integrated arithmetically (mm3 m−2). Reynolds’ functional groups approach was applied to each species biomass (mg l−1). Using biological data, Q index (Padisák et al., 2006) was applied considering the following 5 degrees classification: 0–1: bad; 1–2: tolerable; 2–3: medium; 3–4: good and 4–5: excellent.

For comparison among limnological phases, variance analysis (one-way ANOVA) was employed using the software MINITAB (version 14.1).

Results

Abiotic variables

Temporal and spatial study of water temperature clearly showed alternation between stratification (rainy periods) and mixing periods (dry periods) (Fig. 1). However, during phase II, in which the macrophytes almost entirely covered the reservoir surface, the greatest water transparency and depth of the mixing (Zmix) and euphotic zones (Zeu) were registered. From the beginning of phase III, these values considerably decreased, except for the ones registered during the winter months of 2004. Considering the other limnological features, phases were characterized as follows:

Fig. 1
figure 1

Depth–time diagram of water temperature (°C) during the period 1997–2004 at Garças Reservoir. Dashed lines indicate limits of the three limnological phases

Full size image

  • PhaseI: greatest DO gradients (Fig. 2) were registered during the stratification months (mainly September 1997). A substantial increase of CO2 was observed in the entire water column in the end of phase I as well as the greatest values of N-NH4+, especially during the period of thermal stratification, and anoxia occurred at the bottom of the reservoir. Molar total N:P and dissolved N:P ratios oscillated and long periods of P limitation were observed.

  • PhaseII: DO distribution was more homogeneous; however, during the mixing period and the entire phase II, the lowest values of N-NH4+ were registered. SRP concentrations were, almost always, below the detection limit of the used analytical method during phases I and II. CO2 increase was observed in the entire water column in the beginning of this phase.

  • PhaseIII: in spite of the greatest DO values observed during the spring months (16.5 mg l−1) at the surface in September 2001 (Fig. 2), long lasting periods of anoxia at the deepest layers of reservoir were evident. With the increase of pH, there was also a drastic decrease in the free CO2 availability and an increase in CO3− and HCO3− concentrations. TN substantially increased, especially at the deepest layers of reservoir, its greatest concentration being detected at the bottom of the lake in April 2001 (74,207 μg l−1). After macrophyte removal, a steady increase of SRP was registered in the deepest layers of the reservoir, mainly during the thermal stratification periods (Fig. 3). The highest concentrations of the nutrients above were 703.1 and 549.9 μg l−1, respectively, at the bottom of reservoir in December 2003 and at 3 m depth in November 2003. TP followed a similar trend, however, with an increase in its concentration at the upper layers of the reservoir during the thermal stratification periods. The highest TP concentration was measured at the bottom of the reservoir in December 2000 and at its surface in July 2000 (respectively, 1,588 μg l−1 and 979.7 μg l−1). During some stratification months of phase III and mostly during the spring at the surface of reservoir, N limitation was observed (Table 1).

Fig. 2
figure 2

Depth–time diagram of dissolved oxygen (mg l−1) during the period 1997–2004 at Garças Reservoir. Dashed lines indicate limits of the three limnological phases

Full size image
Fig. 3
figure 3

Depth–time diagram of soluble reactive phosphorus (μg l−1) during the period 1997–2004 at Garças Reservoir. Dashed lines indicate limits of the three limnological phases

Full size image
Table 1 Minimum, maximum and mean values, standard error (SE) and variation coefficient (VC, %) of abiotic and biological variables during phase I (n = 75), phase II (n = 85) and phase III (n = 320) and Q Index (phase I, n = 15; phase II, n = 17; phase III, n = 64) at Garças Reservoir
Full size table

Biological variables

Chlorophyll a values showed a clear temporal pattern along the reservoir limnological phases, increasing markedly after intervention in the entire water column, and ranging from 218 in phase I to 1,324 μg l−1 in phase III. Chlorophyll a:wet weight ratio among the phases were similar considering the whole water column values (0.31, 0.25 and 0.30, respectively). However, considering the surface values, the third phase presented the lower mean value (0.16) (Table 1). Integrated values of chlorophyll a (Table 1) increased 2 times from phase I to phase II, and 3 times from phase II to phase III. Chlorophyll a values were considered significantly different among the three limnological phases (ANOVA ‘one way’; F = 22.8; P < 0.05) (Table 1, Fig. 4).

Fig. 4
figure 4

Depth–time diagrams of (A) total biomass, (B) chlorophyll a and (C) Cyanobacteria during the period 1997–2004 at Garças Reservoir. Dashed lines indicate limits of the three limnological phases

Full size image

Total biomass also showed a huge increase after water hyacinth removal, ranging from 148 mg l−1 in phase I to 4,236 mg l−1 in phase III. The greatest value of integrated biomass was documented in November 2000 (5,539.537 mm3 m−2). The three limnological phases were considered significantly different in relation to the integrated total biomass values (ANOVA ‘one way’; F = 6.3; P = 0.000) (Table 1, Fig. 4).

Cyanobacterial biomass increment was observed during phase III. Maximum values of its total biomass increased almost 29 times from phase I to phase III, and almost 58 times from phase II to phase III (Table 1, Fig. 4). Cyanobacterial integrated biomass increased almost eight times from phase I to phase III and about 21 times from the period the reservoir was covered by macrophytes to the one after their removal (Table 1). During the first phase, the greatest contribution was registered in September 1997, reaching 330.104 mm3 m−2. In the presence of the macrophyte, other algal groups contributed most to the biomass observed. After intervention and beginning of phase III, blue greens dominated during the entire period, reaching its maximum in November 2000 (5,523.969 mm3 m−2). Considering the cyanobacterial integrated biomass values, the three phases were considered significantly different (ANOVA ‘one way’; F = 6.8; P < 0.05).

According to Reynolds et al. (2002) assemblages approach, 21 functional groups were identified during the entire temporal series study. During phase I, despite the contribution of many functional groups, group M (Sphaerocavum brasiliense) was the most representative in the bloom period (September 1997), meanwhile S N (Cylindrospermopsis raciborskii) presented greater contribution during the other stratification months. At the end of the first phase and during almost the whole second one the functional group Y (Cryptomonas curvata and Cryptomonas erosa) presented higher contribution. In that period, L O (Peridinium spp. and Merismopedia spp.) and K (Aphanocapsa elachista, Aphanocapsa delicatissima and Aphanothece smithii) were also important. Despite W2 (Trachelomonas sculpta and Trachelomonas volvocinopsis) appearance at the end of the studied period, this group was as well registered at the beginning of phase II. H1 (Aphanizomenon gracile and Anabaena planctonica), S N (Cylindrospermopsis raciborskii) and S1 (Planktothrix agardhii, Pseudanabaena galeata and Geitlerinema unigranulatum) presented important biomass contribution during the entire third phase. However, M functional group (Sphaerocavum brasiliense, Microcystis panniformis and M. aeruginosa) was responsible for the greatest biomass values registered in that period (Fig. 5).

Fig. 5
figure 5

Relative integrated biomass (%) of functional groups (sensu Reynolds, 1997 and Reynolds et al., 2002) during the three limnological phases at Garças Reservoir. Legend: D, L O , W2, S1, H1, K, M, Y, S N and others

Full size image

Q Index and trophic status index

Following the steps recommended by Padisák et al. (2006) for the Q Index application, the factor F was determined for each functional group occurring at Garças Reservoir (Table 2).

Table 2 Factor F to the functional groups of Garças Reservoir
Full size table

The Q Index showed differences between the three limological phases. Phase I classification varied from bad to medium. Under the macrophyte influence, Q index pointed to tolerable and medium classifications. After intervention, Garças Reservoir ecological status was during most of the period under bad classification, with exception of the last few months at the end of phase III, in which the classification was tolerable. The three limnological phases were considered significantly different considering the Q Index (ANOVA ‘one way’; F = 119.4; P = 0.000). Table 1 shows the Q Index variation amplitude at Garças Reservoir phases (Fig. 6).

Fig. 6
figure 6

Ecological status evaluation by Q Index Relative during the three limnological phases at Garças Reservoir

Full size image

Different trophic status indices applied to Garças Reservoir by Bicudo et al. (2006) evidenced the three limnological phases. Not only through Carlson’s modified by Toledo (1990) TSI, but also Lamparelli’s (2004) indicated changes among the three phases, considering chlorophyll a, Secchi disk and TP data set or the mean values themselves, especially when comparing phases I and III to phase II (Fig. 7).

Fig. 7
figure 7

Garças Reservoir trophic classification according to Carlson’s trophic status index modified by Toledo (1990) and Lamparelli (2004) (from Bicudo et al., 2006)

Full size image

Discussion

Temporal series analyses done by Bicudo et al. (2007) aiming at determining if intervention had a significant effect on the limnological characteristics of Garças Reservoir classified the limnological responses to the macrophyte removal a ‘permanent abrupt impact’ (sensu McDowall et al., 1980).

On the basis of all data obtained during the long term study of Garças Reservoir, a continuous degradation of the ecosystem after the water hyacinth removal was evidenced. Cyclic anoxic periods previously observed during springs and summers were replaced by a more persistent period of anoxic conditions in the sediment-overlying water (Bicudo et al., 2007). P dynamics, initially driven by allochthonous sources, was replaced by an internal loading. Water hyacinth removal triggered self-eutrophication leading to an irreversible turbid phase (Bicudo et al., 2007).

Alterations above were clearly demonstrated by the phytoplankton responses and the Q Index. Oscillations in the ecological status of Garças Reservoir during phase I evidenced that, although eutrophic, there was a tolerable period in which the ecosystem was inhabited by many ‘good’ functional groups, and another period when the M group dominated, driving the reservoir ecological status to bad.

Q index pointed for a better ecological status during the macrophyte presence, classified in between tolerable and medium. This was due to the contribution of a more diverse composition of functional groups and greatest values of the F factor.

After macrophyte removal, worst ecological status, varying from bad to tolerable, was registered due to the overwhelming dominance of functional groups of bloom forming cyanobacteria.

In 2000, Microcystis panniformis bloom was responsible for the high biomass values (4,236 mg l−1) due to its big colonies. These are characterized by having cells densely aggregated in all the mucilage surface (Sant’Anna et al., 2004) and were already registered as tending to preserve their arrangement after Lugol’s solution addition (Naselli-Flores & Barone, 2003). Also, in that period, high value of chlorophyll a was registered (1,197 μg l−1). In 2004, the highest chlorophyll a value was observed (1,324 μg l−1), with the contribution of M. aeruginosa to almost 100% of the total biomass registered. These bloom episodes were reflected directly in chlorophyll a:wet weight ratio variance observed in phase III. Although the average value found in all phases are under the common ones found in world literature (Vörös & Padisák, 1991; Temponeras et al., 2000), the huge biomass of cyanobacteria found in the third phase would explain not only the ratio’s wide variation (variation coefficient = 203%, n = 320; 75%, n = 64) but also its extreme values observed. Such values of chlorophyll a and biomass were also verified in very impacted lakes and reservoirs in Brazil (Huszar et al., 2000, 2006; Da Silva, 2005) and in other ecosystems around the world (Zohary & Breen, 1989; Zohary et al., 1996; Kotut et al., 1998). In Brazilian fishponds in the metropolitan region of São Paulo, Da Silva (2005) registered a biomass peak of M. aeruginosa reaching almost 20,000 mg l−1. Huszar and collaborators (2006) found a maximum value of 556 μg l−1 chlorophyll a in a set of 192 (among them 79 Brazilian) tropical and subtropical ecosystems. In the Garças reservoir, despite those peaks mentioned above, the mean values of chlorophyll a and total biomass are comparable to the aforementioned selection of lakes (Table 1). Extreme values of chlorophyll a were also registered by Zohary & Breen (1989) when studying Microcystis aeruginosa hyperscums in a hypertrophic South African lake (Hartbeespoor Dam). These authors registered a high peak of chlorophyll a exceeding 100,000 mg m−3.

Cyanobacterial blooms were important integrating factors in the self-establishing ecosystem state (stable degraded state) during phase III, acting in the feedback mechanisms already identified in the lake (Bicudo et al., 2007). In a general way, Q index was inversely proportional to the dominance of M, S N , S1 and H1 functional groups.

Padisák et al. (2006) concluded that, after discussing the Q index application to Balaton Lake with an evaluation made through some categories based on algal fresh weight by Mischke et al. (2002 in Padisák et al., 2006), responses of the first one were more realistic than those given by chlorophyll a and biomass. Latter authors also evidenced that the trophic gradient of Balaton Lake were registered by the index.

Comparing responses of both Garças Reservoir TSI methods applied by Bicudo et al. (2006) to the Q index ones, alterations registered in the three limnological phases could be identified in all of them. Although TSI have shown different results on the trophic classification, both showed better water conditions during phases I and II and a worse one during phase III, as well as did the Q index. However, the only biological characteristic considered in the TSI was chlorophyll a. Q index takes into account not only the algal biomass through the functional groups contribution, but also the species themselves. According to Padisák et al. (2006), the method is differently sensitive to taxonomical misidentifications and cyanobacterial identification requires special skills and profuse taxonomic literature.

Among the weaknesses pointed by the Q Index’s authors is the fact that the factor F is determined exclusively by previous knowledge and experience. Determining factor F was the most difficult step during the present study. However, it was our first intention not to let the eutrophic conditions of the Garças Reservoir influence the weight of each functional group. It was also attempted to imagine what those functional groups would be during the pristine status of the system, although there is no scientific or historical reference about them. In this regard, considering the morphological similarities between the Garças Reservoir (hypereutrophic urban shallow lake) and IAG Reservoir (oligotrophic urban shallow lake), both located in the same park, it was suggested that the same factor F could be applied to the last reservoir that could be considered the opposite or the pristine state of Garças Reservoir.

About the IAG Reservoir, phytoplankton studies were performed (Oliveira, 2004; Ferragut, 2004; Lopes et al., 2005). Lopes et al. (2005) postulated, after evaluating the short term and temporal variation of phytoplankton in this lake, that during the dry period under constant mixture and least nutrient availability functional groups X2 (Chlamydomonas planctogloea, Chlamydomonas sp.) and F (Oocystis lacustris and Kirchneriella pseudoaperta) were relevant. Meanwhile, in the rainy season, period of daily stratification and highest nutrient availability, the same species of functional groups X2 (Chlamydomonas planctogloea, Chlamydomonas sp.), Z (Synechococcus nidulans, Chroococcus minor), F (Oocystis lacustris, Elakatothrix gelatinosa), J (Crucigenia tetrapedia, Tetraedron caudatum) and X1 (small Chlorococcales) were the most representative ones. In the present study, Lopes et al.’s (2005) contribution on functional groups was taken into account regarding the factor F determination. To those assemblages that are supposed to occur under a pristine state in the Garças Reservoir, a higher F value was ascribed.

Padisák et al. (2006) described eight types of lakes in Hungary and for each one ascribed different factor F weights. However, according to Garças Reservoir characteristics, it does not seem adequate to use the same weights defined to the Hungarian lakes. Considering its typology and the fact that the Garças Reservoir is a man made, tropical, urban, small, not exposed to wind action, with a sedimentary rocks ground, and considering the long term phytoplankton expertise and knowledge, it seems that F factors are well determined by phytoplankton of IAG Reservoir examples, since this last one has the same characteristics and is under no substantial human impact.

Among the Hungarian lakes described in Padisák et al. (2006), the closest one to Garças Reservoir is the type 7, formed by oxbows. Nevertheless, when evaluating the factor F of this sort, it seems not reasonable to ascribe weights 3 and 3.5 to, respectively, X1 and X2 functional groups, since previous studies showed the important contribution of these assemblages to what we consider the pristine state of the Garças Reservoir. Consequently, in the present study, factor F has different weights compared to those of the oxbows’ sort.

Finally, Q index presented very good results as the trophic index applied to the Garças Reservoir, indicating that phytoplankton could be successfully used as an indicator of ecological status of the system in monitoring processes.