Intra-annual variation in planktonic ciliate species composition ( Protista : Ciliophora ) in different strata in a shallow floodplain lake Variação intra-anual na composição de espécies de ciliados planctônicos ( Protista : Ciliophora ) em diferentes estratos de uma lagoa rasa de planície de inunda

Aim: We aimed to evaluate the variation in planktonic ciliate species composition in different strata of the Guaraná Lake, encompassing high and low water periods, at the Upper Paraná River floodplain. Methods: Samplings were collected monthly between March 2007 and February 2008, from the epilimnion, metalimnion and hypolimnion. Ciliates samples were filtered using a plankton net of 10μm mesh size and identified in vivo under an optical microscope. Results: Among 112 species identified, 13 were found exclusively during the high water periods and 39 during the low water period. Results of nonparametric extrapolation indices evidenced that the observed richness represented between 70% and 90% of the estimated richness. Regarding the variation in species composition, Beta1 index showed that the alteration in composition between strata during the low water period was slightly greater than that registered during high waters. Cluster analysis evidenced a higher dissimilarity in ciliate species composition between periods than among the different strata. The greatest variation in species composition was verified during the distinct hydrological periods, whereas no significant differences were observed for the different strata analysed. Conclusions: We found that in the pelagic compartment, ciliate species composition changed significantly between hydrological periods, and a higher similarity in species composition among strata was observed during the high water period. Therefore, alterations in the vertical distribution seem to be related to the homogenizing effect of the floods in the water column stability.

Distinct ciliate communities are found in specific compartments within the same environment (Finlay & Esteban, 1998b;Zingel & Ott, 2000) showing clear patterns of vertical distribution (Mieczan, 2008).Oligotrichs usually dominate in the epilimnion in lakes of temperate regions, although individuals belonging to Prostomatida, Peritrichia and Haptorida are also frequently found in this water layer.Scuticuciliates, prostomatids and haptorids are usually more numerous in the intermediate layer (metalimnion) and scuticuciliates, heterotrichs and prostomatids are more representative in the hypolimnion (Mieczan, 2008;Zingel & Ott, 2000).On the other hand, in tropical lakes, which are usually shallower than temperate lakes, studies approaching ciliate vertical distribution are still scarce.Thus, we evaluated the variation in ciliate species composition in different strata and two hydrological periods in the Guaraná Lake, at the Upper Paraná River floodplain.
We hypothesized that planktonic ciliate species composition would be different among strata of the water column, considering that the lower strata would be more dissimilar due to the contribution of species associated with the sediment (not typically planktonic).Moreover, we predict that species composition among strata during high waters would be more similar than during low waters, based on seasonal alterations determined by the hydrological regime and considering that the flood pulse is the main factor structuring ciliate communities.

Study area
Guaraná lake (22º43'S-53º18'W) is located in the Upper Paraná River floodplain, Brazil (Figure 1).This lake has 4.2 ha area and 3.2 m mean depth, it is usually covered by floating aquatic macrophytes

Introduction
River-floodplain systems are characterized by a great diversity of water bodies, including lotic, semi-lotic and lentic environments (Junk et al., 1989).This great environmental diversity, together with the connectivity among those habitats, and the exchange of organisms between the river and the adjacent floodplain, provide a high biodiversity, due to periods of low and high waters that result in a unique environmental condition, determined by the flood pulses (Neiff, 1990).The hydrological connectivity resulting from the flood pulse act as a homogenizing factor in the habitats of the Upper Paraná River floodplain (Thomaz et al., 2007).In fact, several studies show the importance of the flood pulse in structuring aquatic communities in floodplains (Algarte et al., 2009;Arrieira et al., 2017;Lansac-Tôha et al., 2009;Train & Rodrigues, 2004).
Ciliates are important components of food webs in floodplain environments, acting as a link between bacteria and flagellates and higher trophic levels (Azam et al., 1983;Segovia et al., 2015), besides feeding on phytoplankton organisms (Sherr & Sherr, 2002).However, as opposed to temperate regions, where temperature is the main factor structuring ciliate communities (Graham et al., 2004;Müller et al., 1991), in floodplain habitats of tropical regions the hydrological regime constitutes the major force influencing those protists (Pauleto et al., 2009).This temporal alteration of the hydrological regime provides the occurrence of species that are adapted to this condition, which is facilitated by mechanisms to persist in the environment during unfavorable periods (Lytle & Poff, 2004).For example, ciliate encystment allows some species to tolerate physical and chemical stresses and reduce their metabolic losses and risk of predation (Taylor, 1981).

Sampling and laboratory analysis
Sampling was performed in the Guaraná Lake from March 2007 to February 2008.Two-liter water samples were taken at three sampling sites in three different depths: epilimnion (approximately 20cm below the water/air interface), metalimnion (determined as the half of the total depth of the water column at the time of sampling) and hypolimnion (approximately 20 cm above the substrate), since thermal stratification was not observed in any occasion.

Data analyses
Samples were categorized in two hydrological periods according to the average depth registered throughout the study period (Figure 2).We considered the months which exceeded the average depth (March, April and December 2007;January and February 2008) as high water period, whereas the other months were considered as low water period (May to November 2007).
We used a nonparametric procedure for the extrapolation of the species accumulation curve, based on incidence data: Jackknife 1, Jackknife 2 and Bootstrap, to estimate ciliate species richness and analyse which portion of the expected total species richness was registered in our study.Estimators were calculated using software EstimateS (Colwell, 2006).Ciliate species were arranged according to their preferential habits (following Berger & Foissner, 2003) as planktonic (adapted to the pelagic compartment) or non-planktonic (associated to a substrate, from the littoral or benthic zones).The frequency of occurrence of ciliate species was calculated through the percentage of samples in which those species occurred (Fr=n*100/N, where n = species occurrence in the analysed samples and N = total number of analysed samples).According to their frequency of occurrence, species were then grouped in: constant (76% to 100% of samples), frequent (51% to 75% of samples), accessory (26% to 50% of samples) and accidental (less than 25% of samples) species.
Nonmetric multidimensional scaling (MDS) was performed to summarize patterns in species composition, based on Jaccard distance.Significant differences were tested through a Permutational Multivariate Analysis of Variance (PERMANOVA) determined by 999 permutations.MDS and PERMANOVA were performed in software R (R Core Team, 2013) using "vegan" package (Oksanen et al., 2016).
We used Beta 1 diversity index (Harrison et al., 1992) to quantify the turnover in ciliate species composition regarding the vertical distribution and hydrological periods, which was estimated through the expression: β-1 = {[(S/α)-1]/(N-1)}×100, where S is the total number of ciliate species registered in each sampling site, α is the mean number of species found in the samples and N is the number of sampling units.
Cluster analysis (Hammer et al., 2001), based on Jaccard similarity coefficients, were performed to verify similarities in vertical and temporal distributions of ciliate species.Cophenetic Correlation Coefficient was calculated to estimate the representativeness of the dendrograms relative to the original data.

Results
In our study, 112 ciliate species were identified (Table 1).Nonparametric extrapolation indices results showed that the observed richness represented between 70% and 90% of the estimated richness.Bootstrap (124 species) was the index that better reflected the observed species richness (Figure 3).
Among the identified species, 13 were registered only during the high water period, whereas 39 species where registered only during the low water period.Nine species occurred exclusively at the epilimnion, four species only at the metalimnion and 21 species were registered only at the hypolimnion.
A total of 32 species were registered in all strata and both hydrological periods.Halteria grandinella Mueller, 1773, Urotricha farcta Claparède & Lachmann, 1859 and Tintinnidium cf.pusillum Entz, 1909, were registered in over 50% of the samples in each stratum, both during high and low water periods (Table 1).
Considering the number of ciliate species in each of the analysed months, Prostomatida, Oligotrichida    and Scuticociliatida showed the greater number of species.Haptorida and Peritrichida also showed a high number of species (Figure 4).Regarding the distinct strata, Prostomatida showed the highest number of species in each vertical compartment, especially at the metalimnion.Scuticociliatida, Haptorida and Oligotrichida were also important in all strata, whereas Nassulida occurred only at the metalimnion, Pleurostomatida occurred only at the hypolimnion, and Synhymeniida was absent at the epilimnion (Figure 5).
Over 70% of the ciliate species (82 species) could be considered non-planktonic and associated with some type of substrate, whereas only 26.8% (30 species) are truly planktonic.Thus, there was a dominance of non-planktonic species, mainly in March and May, when those organisms represented over 66% of the total species number.(Figure 6).
Considering the two hydrological periods, non-planktonic species remained dominant, mainly during the low water period when 71 non-planktonic species (71.7%) and 28 planktonic species (28.3%) were registered.In the same way, during the high water period 49 non-planktonic (67.1%) and 24 planktonic (32.9%) species were found (Figure 7A).
Regarding the vertical distribution of the preferential habitat, a greater number of non-planktonic species was registered in all strata.The highest relative contribution of non-planktonic species percentage was found in the hypolimnion (70 species, corresponding to 72.2% of total species), followed by the epilimnion (54 species, or 67.5% of total species) and metalimnion (47 species, equivalent to 65.3% of total species) (Figure 7B).Ciliate species turnover indicated that the alteration in composition among strata was slightly higher during the low water period (26%) than in the high water period (20.2%) (Figure 8A).Considering the turnover in composition in each stratum between hydrological periods, changes were more pronounced.Higher turnover rate was registered in the hypolimnion (33.8%), followed by the metalimnion (26.3%) and the epilimnion (21%) (Figure 8B).
Cluster analysis (Cophenetic Correlation Coefficient = 0.87) showed a higher dissimilarity in ciliate species composition between hydrological periods than among strata.Therefore, the analysis distinguished two groups coinciding with the hydrological periods.Moreover, there was a higher similarity between the epilimnion and the metalimnion (0.64) during the low water period, and a higher similarity between the metalimnion and the hypolimnion (0.71) during the high water period (Figure 9).
MDS results showed that sampling units were separated only regarding the two hydrological periods, but not regarding the different strata (Figure 10).PERMANOVA statistically confirmed significant differences in ciliate composition between the hydrological periods (Pseudo-F=4.0236;p<0.001), whereas no significant differences were found in the pairwise comparison between epilimnion and metalimnion (Pseudo-F=0.31467;p=0.996), between epilimnion and hypolimnion (Pseudo-F=0.80554;p=0.713) nor between metalimnion and hypolimnion (Pseudo-F=0.6299;p=0.903).

Discussion
The high number of ciliate species registered in the Guaraná Lake not only exceeds the gamma diversity found in freshwater environments both in Brazil (Bossolan & Godinho, 2000;Cardoso, 2007;Dias et al., 2008;Gomes & Godinho, 2003)    and other regions of the globe (Carrick, 2005;Mayer et al., 1997;Muki et al., 2005;Song, 2000a;Wiackowski et al., 2001), but is also similar to the number of species found by Madoni & Braghiroli (2007) in six sampling sites in an Italian river system, and by Pfister et al. (2002) in 58 north German lakes of distinct trophic status.
Previous samplings performed by Pauleto et al. (2009) in the Guaraná Lake found only 36 ciliate species, however, their sampling design was different: one sample was taken during the high water and one during the low water period.Meanwhile, in our study, in which samplings were performed monthly during a whole year at this same lake, 112 ciliate species were found.This supports the idea that long-term approaches with high sampling effort are necessary in order to determine the total diversity of dynamic environments, which encompasses both active and passive diversities (Finlay & Esteban, 1998a).
Our results evidenced the great biodiversity of floodplain lakes which is, in part, determined by the contribution of non-planktonic organisms from the littoral region and the sediment (Lansac-Tôha et al., 2009).Indeed, non-planktonic ciliates constituted most of the species richness registered in the pelagic compartment of Guaraná Lake.Ciliate species richness is, in general, favoured by the occurrence of macrophyte banks (Song, 2000b), especially the more diverse ones, which support higher species richness and abundance of non-planktonic organisms (Karus et al., 2014), contributing to the increment of species of the planktonic community.The massive contribution of non-planktonic organisms in the pelagic compartment was also observed in studies approaching the zooplankton community in the Upper Paraná River floodplain (Alves et al., 2010;Bonecker et al., 1998;Lima et al., 1998), which evidenced that the faunal exchange between lake compartments occurred mainly during floods due to greater habitat connectivity, leading to the presence of both planktonic and non-planktonic species within the community.
Most of the truly planktonic species are, in general, cosmopolitan, and are thus commonly registered in lake habitats (Pfister et al., 2002).Although Urotricha farcta typically inhabits mainly the benthic compartment, it is also found in the pelagic region of lakes and rivers of reduced water flow rates (Berger & Foissner, 2003).Organisms belonging to this genus, despite their small size, are efficient predators and highly adaptable and tolerant, and are one of the few to be present even in extremely acidic lakes (pH = 3.0) (Packroff, 2000).
Halteria grandinella and Tintinnidium cf.pusillum, besides other species that were common in our study, such as Rimostrombidium humile and Pelagostrombidium mirabile, belong to the order Oligotrichida and are characteristic of the pelagic compartment of lentic environments (Mieczan, 2007;Müller, 1989).These organisms are filter-feeders and feed on bacteria and/or small sized algae, besides being totally adapted to the pelagic environment (Foissner & Berger, 1996).We found a high number of species belonging to Prostomatida and Scuticociliatida, which are important in the metalimnion and the hypolimnion of stratified lakes (Zingel, 2005), reinforcing the influence of littoral and benthic regions in ciliate species composition of the pelagic compartment of the lake.The constant presence of those ciliate orders, which are mostly omnivorous or bacterivorous, suggests a continuous input of organic matter from the marginal zone to the central region of the lake, facilitated in environments with reduced dimensions such as the Guaraná Lake.
We found a greater turnover in species composition of the distinct strata during low waters than during high waters.Similarly, the greater temporal turnover found at the hypolimnion suggests the development of a distinctive community at the bottom of the lake at certain stages of the year.These findings are corroborated by the Cluster results (see Figure 10), which indicates the lowest similarity between samples from the hypolimnion and the other strata during the low water period.During low waters, the wind action is more pronounced because lakes are shallower, inducing water circulation and destratification (Thomaz et al., 2004), which lead to a higher similarity in the ciliate species composition between the epilimnion and metalimnion strata.Meanwhile, in the hypolimnion we found the highest number of exclusive ciliate species, which indicates that a unique community was formed in this water layer.On the other hand, during high waters, there is a relatively stable thermal stratification (Thomaz et al., 2004), which likely resulted in the distinction of the epilimnion from the other two water layers.
Although we found differences in ciliate species composition among strata, a higher dissimilarity was found between hydrological periods in Guaraná Lake.In fact, the flood pulse is recognized as the main factor structuring aquatic communities (Junk et al., 1989;Neiff, 1990), and was also found to be the major factor influencing the ciliate community in the Upper Paraná River floodplain lakes (Pauleto et al., 2009).
In summary, our hypotheses were corroborated, since we found that, in the pelagic compartment, ciliate species composition changed significantly between hydrological periods, and a higher similarity in species composition among strata was observed during the high water period.Therefore, alterations in the vertical distribution seem to be related to the homogenizing effect of the floods in the water column stability.Intra-annual variation in planktonic... Acta Limnologica Brasiliensia, 2017, vol. 29, e107

Figure 1 .
Figure 1.Study area showing the sampling sites at the Guaraná Lake.

Figure 2 .
Figure 2. Average water depth registered at the Guaraná Lake during the study period (black dots = mean, bar = standard error).

Figure 3 .
Figure 3. Nonparametric estimators for ciliate species richness in the Guaraná Lake.

Figure 4 .
Figure 4. Composition by order (%), registered in the analysed months at Guaraná Lake.

Figure 6 .
Figure 6.Species richness of planktonic and non-planktonic ciliates, analysed monthly at Guaraná Lake.

Figure 9 .
Figure 9. Cluster analysis based on ciliate species occurrence in distinct strata (Epi=epilimnion, Meta=metalimnion, Hypo=hypolimnion) in two hydrological periods (High water and Low water periods) at Guaraná Lake.

Table 1 .
List of ciliate species registered in distinct strata (Ep = Epilimnion; Me = Metalimnion; Hy = Hypolimnion) in two hydrological periods (High water and Low water) at Guaraná Lake (H = Preferential Habitat; P = planktonic species and N = non-planktonic species).