Diatom assemblages ( Bacillariophyta ) in six tropical reservoirs from southeast Brazil : species composition and spatial and temporal variation patterns Assembléias de diatomáceas ( Bacillariophyta ) em seis reservatórios tropicais no sudeste do Brasil : composição de espécies e padrões de variação

Aim: Relationships between species composition and variation patterns in diatom assemblages over six tropical reservoirs located in Southeast Brazil were explored. Methods: Surface-sediment and phytoplankton diatom assemblages were determined and Canonical Correspondence Analysis was used to verify the set of environmental variables that best explain the species composition variation among sites. Results: A total of 28 diatom taxa representing 20 genera were identified using light and scanning electron microscopy. Information on their ecological preferences was also provided. Humidophila biscutella is reported for the first time in Brazil whereas Sellaphora sassiana and Humidophila brekkaensis are reported for the first time in São Paulo State. Three groups of potential water quality indicators were delineated: the first suggests oligotrophic conditions, the second is related to cold waters with low luminosity in a mixing regime, and the third is a small group of tolerant species occurring in water with high conductivity, pH and total phosphorus conditions. Conclusion: present study provided the first insight into the general diatom communities over six tropical reservoirs in Southeast Brazil and provided information on their ecology and distribution aiming bioassessment. The potential of water quality indicator diatom groups here outlined are in line with those reported in the literature and reinforce the importance of PEJU for the maintenance of ecological quality of reservoirs and reference conditions for the Metropolitan Region of São Paulo water sources.


Introduction
Diatoms are routinely used in environmental status assessment because of their importance in food webs and biochemical linkages, and due to their sensitivity to physical, chemical and biological disturbances (Stenger-Kovács et al., 2007;Bolla et al., 2010;Chételat et al., 2010;Kireta et al., 2012;B-Béres et al., 2014).Diatoms respond predictably to many water chemistry variables, besides having a relatively well known ecology (Soininen et al., 2007).Factors such as choice of sampling site and methods for preparing and processing samples and identifying taxa can be crucial to the assessment results (Besse-Lototskaya et al., 2006) as their use as environmental indicators requires high taxonomic precision (Birks, 1994).
The composition of diatom communities reflects an entire complex of ecological parameters at a particular site (van Dam et al., 1994;Resende et al., 2005).In this sense, floristic surveys support not only applied research projects, such as water quality monitoring, but also a wide range of basic research issues from taxonomic revisions and monographs, phylogenetic reconstruction, biogeographical studies, as well as ecological, physiological, restoration, and conservation biology research programs (Kociolek, 2006).
da Fumaça and Serraria) are connected with the same water course (Juquiá River), whereas the other three (Jurupará, Salto do Iporanga and Paineiras) are located in three different rivers (Rio dos Peixes, Assungui and Turvo rivers, respectively) (Figure 1).Mean elevation range varied from 17 to 996 m a.s.l.
All study sites were located in well preserved areas, most of them located in areas of the PEJU State Park (Parque Estadual do Jurupará).Presently, reservoirs were classified mostly oligotrophic to mesotrophic (Lamparelli, 2004) with no obvious signs of human impact.The area is a protected representative of Atlantic forest in the São Paulo Atlantic Plateau, in which 77 species of fauna and flora currently endangered and 182 endemic species besides more than 300 new species records were recorded (São Paulo, 2010).However, the diatom flora is virtually unknown in this area.

Sampling
Samplings covered two distinct habitats (plankton and surface sediments) from 20 sampling sites distributed along the reservoirs during austral summer and winter of 2014.Plankton samples were obtained with van Dorn water sampler along a vertical profile of the reservoirs.Surface sediments were taken with a gravity core (UWITEC) and the top 2 cm sections were saved for analyses.

Slides preparation and counting
Diatom samples were processed by hot digestion using hydrogen peroxide (H 2 O 2 ) and HCl (37%).Peroxide and acid were removed through a series of dilutions.Subsequently, the samples were dried onto cover glass and mounted in Naphrax (R.I. = 1.74) according to Battarbee (1986) Acta Limnologica Brasiliensia, 2018, vol. 30, e201 Universidade Federal do Paraná and Instituto de Botânica.For scanning electron microscope (SEM) observations, cleaned samples were filtered with additional deionized water through a 3 μm isopore polycarbonate membrane filter (Merck Millipore).Filter was mounted on aluminum stubs and coated with platinum using a Modular High Vacuum Coating System BAL-TEC MED 020 (BAL-TEC AG, Balzers, Liechtenstein).An ultrahigh-resolution analytical field emission (FE) scanning electron microscope Hitachi SU-70 (Hitachi High-Technologies Corporation, Japan) operated at 5 kV and 10 mm working distance was used for the analyses.SEM images were taken using the lower (SE-L) and upper (SE-U) detector signal at the LIST.Also, a subsample of the oxidized material was placed on aluminum stubs and coated with gold at 1 kV for 5 min in a Balzers Sputtering/SDC030 sputter coater, and SEM observations were made with a JEOL JSM 6360LV, operated at 15 kV at 8 mm distance at the UFPR as well as with a Phillips 20XL operated at 10 kV at the Instituto de Botânica.Micrographs were digitally manipulated and plates containing light and scanning electron microscopy images were created using CorelDraw X7.Morphometric information is provided for all taxa [D: diameter (μm); L: length (μm); W: width (μm); SH: semi cell height (μm); S: striae in 10 μm; A: areolae in 10 μm].
Taxonomy and nomenclature followed classic works and new publications (e.g.Krammer, 2000;Metzeltin et al., 2005;Lange-Bertalot et al., 2011) and the on-line catalogue of valid names (California Academy of Sciences, 2011).Classification systems followed Medlin & Kaczmarska (2004) for supra-ordinal taxa, and Round et al. (1990) for subordinal taxa, except for genera published subsequently to this work.To account for the species distribution in Brazil and the State of São Paulo, literature with illustrations or sufficient taxonomic description of the species were considered.
Diatom quantification was made at 1000× magnification using a Zeiss Axioskop 2 microscope.At least 400 valves were counted per slide (Battarbee, 1986) with a minimum sampling efficiency of 90% (Pappas & Stoermer, 1996).Species abundances were calculated and expressed as a percentage of the total diatom count in each sample.
All slides used for diatom identification and enumeration were deposited at the "Herbário Científico do Estado Maria Eneyda P. Kauffmann Fidalgo" (SP), São Paulo State Department of Environment, Brazil.

Statistical analyses
We used the unimodal-based method based on Detrended Correspondence Analysis (DCA) by checking the length of the first DCA axis (length of gradient 2.5) (Birks, 2010).Final ordination was based on a canonical correspondence analysis (CCA) in order to extract major gradients among combinations of explanatory variables in a dataset.We used a stepwise function selection to obtain a subset of explanatory variables from the set of all variables available for the constrained ordination.We further performed a PERMANOVA two-way for two different groups: (i) reservoirs and (ii) habitats (surface sediments and phytoplankton for both seasons) in order to evaluate which factor structures the diatom community.
We downweighted our data of rare species by applying a 2% cut off on the relative abundance and eliminated all taxa occurring in less than two samples, because individual samples with rare species may distort the results of the analyses (Birks, 2010).Also, much of the observed species were rare and composed singletons making the species richness very high, consequently obstructing the taxonomic identification.The environmental variables were standardized and species abundances were Hellinger-transformed.Hellinger distance is the recommended measure for clustering or ordinating species abundance data (Legendre & Gallagher, 2001).Diatom names were coded according to the OMNIDIA software (Lecointe et al., 1993).
Ordination techniques were performed in software R v. 3.0.2(R Core Team, 2014) with the vegan package (Oksanen et al., 2016).

Taxonomy
A total of 28 species of diatoms belonging to 20 genera was identified for the planktonic and surface sediments communities of the six sampled reservoirs.Only three species (Discostella stelligera (Cleve & Grunow) Houk & Klee, Aulacoseira tenella (Nygaard) Simonsen and Spicaticribra kingstonii J.R.Johansen, Kociolek & R.L.Lowe were considered dominant (> 50% in relative abundance) (Figures 2 A, B and C) and three groups of potential water quality indicators were delineated (Figure 3).

Canonical correspondence analysis
The CCA biplot with the selected variables are illustrated in Figure 3. PERMANOVA two-way analysis showed significant among-group differences both among reservoirs and habitats (F=5.36;P>0.05) and habitats (F=4.98;P>0.05).Habitat: reservoirs interaction values were non-significant (F=0.88;P<0.5).
The stepwise selection of environmental variables retained six significant variables in decreasing importance, which influenced diatom distribution: TP, conductivity, pH, SSR, Secchi and TN.These variables accounted for about 36% of the biological variation in the first two CCA axes.The intraset correlations indicated that TP and conductivity were the most significant contributors to axis 1, whereas pH and Secchi were the most significant contributors to axis 2.
There was a trend in the separation of the sites according to the hydrological connectivity between the rivers of the systems (Figure 3A), and the ordination plot distinguished four main groups: the first group was represented by reservoirs located at the same watercourse (Cachoeira do França, Serraria and Cachoeira da Fumaça).This group had the highest abundance of Spicaticribra kingstonii     2.
Table 2. List of diatom taxa identified from the reservoirs (>2% relative abundance).Morphometric and meristic limits and ecology according to the literature.D: diameter (μm); L: length (μm); W: width (μm); SH: semi cell height (μm); S: striae in 10 μm; A: areolae in 10 μm; LN: lineolae in 10 μm.125,128).These species are known for their planktonic habit and they showed the greatest relative abundance during the winter, in which the water column is in a mixing regime.

Discussion
Studies about limnological characteristics for the studied region are extremely rare, especially those considering seasonal and spatial influence.Most of them are restricted to reports on water sources that focus on the characterization of the conservation unit (CETEC, 2000) and to the PEJU Management Plan (Fundação Florestal do Estado de São Paulo), which covers the water sources inserted in the Park area.Among others, there are some studies regarding the surface water quality (Cetesb, Companhia Ambiental do Estado de São Paulo) besides reports organized by the Companhia Brasileira de Alumínio (CBA), which operates eight plants in the Ribeira de Iguape basin.Concerning the Paineiras reservoir, only one survey was found: an unpublished doctoral dissertation whose theme is the adaptation of the fish assembly index in reservoirs along the Turvo River (SP) (Ferreira, 2011).
The currently measured limnological characteristics of the reservoirs indicate that the systems are under relatively protected conditions, with low anthropogenic impact, promoted by their insertion in the PEJU preservation area.Low concentrations of nutrients in their dissolved forms as well as TN and TP values measured in this study are typical of oligotrophic and mesotrophic environments (Tundisi, 2006;Vercellino & Bicudo, 2006).However, despite the nutrients played an important role in the diatom community variation, the hydrological connectivity role should not be neglected.
According to the currently calculated TSI, reservoirs in the same watercourse included in the first group showed a better water quality (ultraoligotrophic and oligotrophic) when compared to the second and third groups (mostly mesotrophic) (Table 1).However, it is important to note that there was not a large trophic gradient in the study area, and these eutrophic sites are probably  Figs 111,112,119,121,122: 5 μm;Figs 113,115,117,118: 2 μm;Fig. 114: 4 μm;Fig. 116: 1 μm.Marquardt, G.C. et al. Acta Limnologica Brasiliensia, 2018, vol. 30, e201 a result of the system anthropic management during the sampling period.
Regarding ecological preference trends of species, our results agree with those found in the literature, with most of the species having been reported in Brazilian reservoirs and oligotrophic environments.Occurring on Cachoeira do França, Cachoeira da Fumaça and Serraria reservoirs (first group), Achnanthidium tropicocatenatum was found in alkaline waters of good quality and with low electrolyte content (Marquardt et al., 2017b).
Because of the wide morphological variation, the Spicaticribra kingstonii species complex might be a species flock (Tuji et al., 2012).Also, S. rudis is frequently reported to the Brazilian southern, southeastern and northeastern regions, found in high temperatures and low conductivity waters (Ludwig et al., 2008); this species occurred in the mesotrophic Paraitinga reservoir (Rocha, 2012), and was abundant in the ultra-oligotrophic Jaguari reservoir (Nascimento, 2012) as well those located along the Paranapanema River (Fontana & Bicudo, 2009).Also, Staurosira construens is considered an alkaliphilous and meso-eutraphentic taxon (van Dam et al., 1994;Morales, 2006), registered in the plankton and periphyton in rivers from the Southern region of Brazil (e.g.Flôres et al., 1999;Landucci & Ludwig, 2005) besides an oligotrophic reservoir in São Paulo State (Barbosa, 2012).A further species related to the first group, Discostella stelligera is considered tolerant to nutrient enrichment (Stoermer, 1978).In Brazilian reservoirs, the species was recorded by Faustino et al. (2016) in the Guarapiranga reservoir in mesotrophic to super-eutrophic conditions in 37% of samples.Recent studies suggested that this taxon is part of a key group of diatoms that are frequently dominant members of phytoplankton communities in low to moderate-productivity lakes, and that processes that modify light availability (such as water transparence and water column stability) and nutrient concentration are likely to play a major role in controlling the growth of small centric diatoms in Arctic lakes (Saros & Anderson, 2015).In addition, Gomphonema naviculoides is particularly prevalent in the Tropics and also in North America; to date, this species has been identified as G. gracile Ehrenberg (Reichardt, 2015).According to the latest revision of the complex involving Gomphonema gracile, the species tolerates low salt concentration (Reichardt, 2015).Aulacoseira tenella, typically associated with oligotrophic and oligo-mesotrophic reservoirs (Bicudo et al., 2016), has been recorded in Brazil lotic ecosystems (Landucci & Ludwig, 2005), with low conductivity reservoirs (Raupp et al., 2006;Eskinazi-Leça et al., 2010;Silva et al., 2010;Nascimento, 2012), as well as in acidophilic (Camburn & Charles, 2000) and oligotrophic environments (Siver & Kling, 1997).Fragilaria billingsii is commonly found in São Paulo State reservoirs, living in slightly acidic waters with moderate to high nutrient enrichment (Wengrat et al., 2016).Finally, Eunotia waimiriorum is a common colonial planktonic diatom from oligotrophic pristine rivers and streams in the Amazon hydrographical basin (Wetzel et al., 2010).
Among the diatoms related to the Salto do Iporanga reservoir (second group), Navicula notha is considered a cosmopolitan species, present in oligotrophic environments with low conductivity, slightly acid to near neutral water (Lange-Bertalot, 2001).Geissleria punctifera was considered tolerant to the pollution conditions on rivers and streams in Maringá (Paraná State, Brazil), and was an abundant periphytic diatom in three streams with different anthropic influences (Moresco et al., 2011).Centric diatoms as Melosira varians prefer low light availability conditions.This species is frequent and abundant in environments with vertical column mixing water (Reynolds et al., 2002), and has high nutrient requirements and high disturbance tolerance (B-Béres et al., 2014).
Humidophila brekkaensis is reported for the first time in São Paulo State, and had strong correlation to the pH and conductivity vectors.This species was registered in lotic environments from southern Brazil (Oliveira et al., 2002).
Mostly taxa correlated to the Jurupará and Paineiras reservoirs (third group) were 'araphid' and 'centric' ones.Our studies corroborate Bicudo et al. (2016) in an ecological study focused on Aulacoseira from São Paulo State.In this study, species such as Aulacoseira ambigua and A. granulata were also associated to the water mixing and low light conditions (Houk, 2003;Taylor et al., 2007) showing higher abundances in colder (winter), slightly acid waters.In the same way, A. granulata var.angustissima was associated to environments with higher TN concentration.The araphid Fragilaria longifusiformis subsp.eurofusiformis is distributed in freshwater localities in at least two continents from the northern hemisphere, including ponds, lakes, reservoirs and rivers over a wide range of trophic conditions (Morales & Manoylov, 2006).In USA, latter taxa were lacking from the most acidic habitats in both regions, and were most often observed in relatively dilute waters with a specific conductivity below 200 μS cm −1 (Morales & Manoylov, 2006).Finally, Frustulia crassinervia, considered a species that occurs in oligotrophic environments (van Dam et al., 1994) is a characteristic member of phytobenthos in acid, peaty waters (such as Sphagnum bogs), and many ephemeral habitats (Krammer & Lange Bertalot, 1986;Round et al., 1990;Veselá et al., 2012).Encyonema neomesianum, widely reported in São Paulo State (Marquardt & Bicudo, 2014), is considered an oligotrophic taxon, with an optimum in alkaline waters (Moro & Fürstenberger, 1997).
It is important to note that despite the limited understanding from theory of how factors interact to affect species distribution, there are two important aspects of the landscape spatial structure susceptible to impact the strength of species-sorting: the connectivity matrix (i.e. the spatial arrangement among localities and dispersal rate among them) and the environmental heterogeneity (i.e.variance and range of environmental conditions and their spatial autocorrelation) (Ai et al., 2013).Approaches concerning the significance of relationships among the biological, environmental and spatial datasets in the systems currently studied can be found in Marquardt et al. (2017a).

Conclusions
This study provided new information on the ecology and distribution of tropical diatoms.Three potential water quality indicator diatom groups were outlined, which indicated oligotrophic conditions, water mixing with low light conditions, and species with higher nutrient requirements occurring in higher conductivity and pH waters, in agreement with those already reported in the literature.We reinforce the Parque Estadual do Jurupará (PEJU, Jurupará State Park) importance for the ecological quality maintenance of the reservoirs and reference conditions for the São Paulo Metropolitan Region water sources.

Figure 1 .
Figure 1.Study area and sampling sites.