Raphidiopsis mediterranea (Nostocales) exhibits a flexible growth strategy under light and nutrient fluctuations in contrast to Planktothrix agardhii (Oscillatoriales)

Raphidiopsis mediterranea is a freshwater cyanobacterium that forms toxic blooms in eutrophic water bodies. Factors controlling its proliferation have not been explored in detail. We investigated R. mediterranea autecology by (i) analyzing its dynamics in a hypertrophic shallow lake dominated by Planktothrix agardhii (Oscillatoriales) and its relationship with environmental factors; and (ii) studying the effect of light intensity and phosphate availability on R. mediterranea isolates growing in mono or in co-cultures with P. agardhii. The redundancy analysis demonstrated that water temperature, light, and phosphate concentrations were important driving factors for the seasonal succession of the two species. When grown together with P. agardhii, R. mediterranea growth was strongly promoted under the highest light intensity treatment. On the other hand, in monoalgal cultures under phosphorus starvation, both strains exhibited a significant increase in total alkaline phosphatase activity, and changes in the expression of homologs to phoA-like and phoD genes (members of the Pho regulon). However, R. mediterranea showed higher phosphatase activity than P. agardhii, suggesting greater tolerance to phosphate limitation. Taken together, we conclude that physiological features of R. mediterranea play an important role in the coexistence with P. agardhii under environmental changes.


Introduction
Cyanobacterial blooms in freshwater ecosystems pose a serious ecological and health problem worldwide (Paerl & Otten, 2013). Cyanobacteria tend to dominate in eutrophic lakes due to their eco-physiological traits such as buoyancy, efficient light harvesting, and strategies to face nutrient limitation (Dokulil & Teubner, 2000;Carey et al., 2012). Light and nutrients, in particular phosphorus, are considered primary factors that govern cyanobacterial proliferation (Scheffer et al., 1997;Schindler et al., 2008). Ecophysiological adaptations to face phosphorus limitation include their ability to store it in polyphosphate granules (Gomez-Garcia et al., 2013) and to uptake phosphate from extracellular organic sources after the action of external alkaline phosphatases (Orchard et al., 2009).
Raphidiopsis mediterranea Skuja (Nostocales) is a filamentous species that frequently blooms in eutrophic lakes and reservoirs (Rzymski and Poniedzialek, 2014;Wilk-Woźniak et al., 2016). Due to its ability to produce potent cyanotoxins such as cylindrospermopsin and anatoxin-a, it is regarded as a potential health risk (Namikoshi et al., 2003;McGregor et al., 2011). In eutrophic shallow lakes, R. mediterranea regularly co-occurs with the microcystin producer Planktothrix agardhii (Gomont) Anagnostidis & Komárek (Oscillatoriales) (Kurmayer & Gumpenberger, 2006;Yamamoto & Nakahara, 2009;Aubriot et al., 2011), a resilient and shade-tolerant species, whose dominance in eutrophic freshwater systems is commonly attributed to their ability to maintain net growth under self-shading conditions (Scheffer et al., 1997;Bonilla et al., 2012). In temperate regions, P. agardhii commonly alternate in dominance with Nostocales under extreme low underwater irradiance and high nutrient concentrations (Havens et al., 2003;Nixdorf et al., 2003;Toporowska et al., 2016). Planktothrix agardhii has been studied together with Raphidiopsis raciborskii (Woloszynska) Aguilera, Berrendero Gómez, Kȃstovský, Echenique & Salerno (basionym Cylindrospermopsis raciborskii (Aguilera et al., 2018)) either in field or in laboratory experiments (Kokociński et al., 2010;Ammar et al., 2014). However, factors affecting the dynamics of R. mediterranea in shallow lakes have been little explored (Aubriot et al., 2011;. The aim of this study was to investigate the factors controlling the dominance of R. mediterranea and contribute to a better characterization of the autecology of the species. We investigated the dynamics of R. mediterranea in a turbid and hypertrophic shallow lake dominated by P. agardhii located in the Pampean region of central eastern Argentina. Light is particularly relevant in Pampean shallow lakes (Allende et al., 2009;Izaguirre et al., 2015). Field data were complemented with laboratory assays to test whether these two cyanobacterial species have different responses to changes in light and phosphorus availability.

Materials and methods
Field work: sampling and physicochemical parameters The field study was conducted in Los Patos shallow lake, located in Ensenada city, Province of Buenos Aires, Argentina (34°50 0 42 00 S, 57°57 0 23 00 W). The region has a temperate humid climate, with 16°C annual average temperature and 1,000 mm annual mean rainfall (Díaz & Mormeneo, 2002;Martínez et al., 2006). Los Patos is an artificial and hypertrophic freshwater body (area 0.25 km 2 , maximum depth 1 m), with dense phytoplankton accumulations dominated by filamentous cyanobacteria. The water body is used for recreational activities and fishing (Bauzá et al., 2014).
Sampling was performed fortnightly from June 2012 to May 2014 in two stations in the littoral zone. Water samples were taken 30 cm below the water surface using a van Dorn bottle and preserved in Lugol's iodine solution (1%). Surface water temperature, pH, and conductivity were measured in field using portable pH and conductivity meters (Parsec). Water transparency was estimated with a Secchi disk. Meteorological data (air temperature, rainfall, wind direction, and intensity) were provided by the Departamento de Sismología e Información Meteorológica (SIM), Facultad de Ciencias Astronómicas y Geofísicas, Universidad Nacional de la Plata, Argentina.
Phytoplankton counting was performed following the Utermöhl (1958) methodology with an inverted microscope (Carl Zeiss AXIOVERT 40C). At least 100 individuals of the most frequent species or 400 individuals in total were counted in random fields (accuracy of 10%) (Lund et al., 1958). Individual volume (lm 3 ) was calculated by measuring at last 30 individuals of each taxa and following volumetric formulae (Hillebrand et al., 1999). Biovolume was calculated by multiplying abundance per individual volume (mm 3 l -1 ). Species were considered dominant when represented at least 30% of the total biovolume in a particular sample.

Co-cultures and strain isolation
The co-culture of P. agardhii and R. mediterranea was obtained from an environmental sample taken in October 2012 by means of serial dilutions in multiwell plates containing sterile MLA medium (Bolch & Blackburn, 1996). The medium was supplemented with cycloheximide to a final concentration of 50 mg l -1 to inhibit eukaryotic algae.
Isolation of R. mediterranea FCC LP and P. agardhii FCC LP strains were carried out from samples collected on October 2014 from single filaments, and using micropipettes under microscope. The identification of the strains was verified after 16S rRNA and cpcBA-IGS sequencing (flanking regions of the cpcB and cpcA genes of the phycocyanin operon) (Aguilera et al., 2018). Isolates are kept in the Fundación para Investigaciones Biológicas Aplicadas (FIBA) Culture Collection (FCC) in Mar del Plata, Argentina.

Co-culture experiments
Co-cultures (R. mediterranea ? P. agardhii) were subjected to two phosphate concentration conditions: (i) low phosphate (Pi), called P5 (MLA containing 5 lM P-PO 4 ) and (ii) high Pi called P200 (MLA, containing 200 lM P-PO 4 ). To obtain the low Pi concentration we proceeded as follows: cultures in MLA media were diluted 1:1 (volume:volume) with fresh MLA without phosphorus. When cells reached the stationary phase (approximately 10 days later), cultures were diluted 1:1 (volume:volume) with fresh MLA containing 5 lM Pi till reached the next stationary phase. Then, dilution with fresh MLA containing 5 lM Pi was repeated twice a week for 2 months and kept under this Pi-limiting condition until the beginning of the experiments. The coexistence of P. agardhii and R. mediterranea in co-cultures was tested at two Pi concentrations (P5 and P200) and two light intensities (40 and 80 lmol photon m -2 s -1 ), L L and H L , hereafter. Experiments were carried out with 80-ml cultures grown for 18 days, and by triplicate. Mean light intensity perceived by organisms (I est ) was calculated from absorbance at 440 nm (Aguilera et al., 2017). During the course of experiments, the cocultures were refreshed by addition of one volume of MLA containing the respective Pi concentration (5 or 200 lM P-PO 4 ) every 3 or 4 days. Culture flasks were gently shaken several times a day and grown at 24 ± 2°C under a light/dark (16 h/8 h) cycle. That temperature mimics the conditions found in Los Patos shallow lake during spring or autumn.
Trichome abundance and biovolume for each species were determined every 2 days using a 1-ml Sedgwick-Rafter chamber (McAlice, 1971). Changes in trichome features in response to light and Pi treatments were analyzed since this information can be useful for understanding the variability of natural populations and the flexible responses to specific environmental variables (Shafik et al., 2003;Miner et al., 2005).

Experiments with mono cultures
To examine the effect of Pi deprivation in R. mediterranea FCC LP and P. agardhii FCC LP growing separately, exponentially growing cells (optical density, OD 750 nm * 0.4) were filtered through 0.45-lm cellulose membrane filters (Millipore). Cells were resuspended in MLA or MLA -P (P-PO 4 free) media, respectively. Media MLA -P was supplemented with 200 lM KCl to maintain osmotic balance and avoid potassium deficiency. The experiment was carried out at 24 ± 2°C under a 16 h/8 h light/dark cycle (40 lmol m -2 s -1 /dark) in an orbital shaker (120 rpm) and in triplicate. Regular microscopic inspection revealed that the biomass of the contaminating bacteria never exceeded 1% of the cyanobacterial biomass in the course of experiments.
Alkaline phosphatase activity was measured daily by using p-nitrophenyl phosphate (pNPP, Amresco) as substrate, following Shen & Song (2007). Enzyme activity was expressed in terms of lmol of pNP released lg -1 chlorophyll a h -1 . The flasks were sampled at 24 and 48 h after the start of the experiments for qPCR analysis.
Total RNA was extracted from the cell pellet using the TRIzol Reagent (Invitrogen) according to the manufacturer's instructions. After digestion with RQ1 RNase-free DNase (Promega, USA), RNA (1 lg) was retrotranscribed using random hexamers (Promega, USA). Then, cDNAs were used in qPCR amplifications to quantify the relative abundance of alkaline phosphatase homologous genes (phoA-like and phoD, members of the Pho regulon) and of homologs to genes involved in polyphosphate synthesis and degradation (ppk and ppa, respectively). Target genes were identified in the genome of Raphidiopsis brookii Hill (strain D9), Planktothrix agardhii NIVA-CYA 126/8 and P. agardhii NIVA-CYA 15 using the integrated microbial genomes portal at the Joint Genome Institute (http://img.jgi.doe.gov/) based on homology. PCR primers were designed using Primer-BLAST (www.ncbi.nlm.nih.gov/tools/primer-blast) ( Table S1). The qRT-PCR reactions were performed in a Step One real time PCR system (Applied Biosystems) using a Micro Amp Fast Optical 48-well reaction plate with 15 lL reaction volume containing 1 9 Power Sybr Green PCR Master Mix (Thermo Fisher), 0.2 lM of each primer, and 1.5 lg of cDNA. Cycling program was: one cycle of 95°C for 10 min, 40 cycles of 95°C for 15 s followed by 40 cycles of 54°C for 1 min. A melting curve analysis was conducted to verify the formation of a single unique product and the absence of potential primer dimerization. To standardize the amount of total RNA in each reaction, 16S rRNA was co-amplified as is it considered one of the most suitable internal reference genes for cyanobacteria (Pinto et al., 2012). Samples were run in biological duplicates and technical triplicates. For each primer pair, a standard curve was established by ten-fold dilutions of a PCR template. Linear regression analysis between PCR product concentration and the cycle number (Ct-value) was used to determine primer efficiency and for relative quantification of each gene transcript.

Data analysis
One-way ANOVA was carried out to test whether the environmental and biological parameters were statistically different among sampling stations during the whole sample period (P \ 0.05 was considered significant). A preliminary detrended correspondence analysis (DCA) was conducted to know the length of the gradient of species distribution. Because the DCA did not exceed four units of standard deviation, we performed a redundancy analysis (RDA) suitable for species that respond to a linear distribution model (ter Braak & Šmilauer, 2002). The RDA with all the samples (n = 37) was conducted to determine how much variance of the biovolume of dominant species (frequency at least 30% in one sample) is explained by the environmental variables. Species biovolumes were transformed [ln(x ? 1)] and environmental data were standardized (ter Braak, 1986). Variables which had a variance inflation factor \ 10 were retained in the analysis. The overall significance of the ordination and the significance of the first two axes were tested by a Monte Carlo permutation test (P \ 0.05) using unrestricted permutations. The significance of the environmental variables was assessed by forward selection (ter Braak & Šmilauer, 2002). The relationship between the biological variables and the physical-chemical parameters was also analyzed using Spearman's correlation.
Morphological differences in the strains growing as co-cultures were analyzed with two-way ANOVA, with light intensity and phosphate concentration as factors. The Tukey Test was used for pairwise multiple comparison. Differences in alkaline phosphatase activity and gene expression in the MLA and MLA -P treatments were determined at each time point using a paired t test. Statistics analyses were performed with SigmaPlot version 11 (Systat Software, Inc).

Relationship between R. mediterranea dynamics and environmental factors
During 2012 to 2014, the climate pattern in Los Patos was typical for the region, and characterized by defined seasonality (Fig. 1a). Low transparency and extremely high values of TP, TN, and Chl were registered during the sampling period (Fig. 1b, c; Table 1), with no significant differences between the two stations (P [ 0.05). TP and SRP showed a strong positive correlation with water temperature (r = 0.81 and r = 0.71, respectively, P \ 0.01). Dissolved inorganic nitrogen concentrations were higher from July 2012 to March 2013 (Fig. 1c). Total and dissolved N:P ratios did not vary significantly over the two years (P [ 0.05). Total N:TP and DIN:SRP ratios were lower than 10 and 15, respectively (Fig. 1d).
Water transparency values were associated with cyanobacterial blooms. This parameter showed lower values during P. agardhii blooms than during mass development of Nostocales (Fig. 3). R. mediterranea biovolume increased in parallel to a greater light intensity and higher water temperature (summer and early autumn) (Figs. 1a, 3). RDA analysis was used to study the relationship between the abundance of dominant species (Nostocales and P. agardhii) and environmental variables (Fig. 4). Significant variables according to Monte Carlo permutation test (P \ 0.05) were temperature, Secchi depth, SRP, and N-NO 3 -. The analysis showed that the higher water temperatures and SRP were more important for the proliferation of Nostocales than for P. agardhii. Spearman's correlation analyses confirmed the importance of water temperature, TP, and SRP for the development of Nostocales species (Table S2).
Effect of light and phosphate concentration on P. agardhii and R. mediterranea co-cultures The coexistence of P. agardhii and R. mediterranea was evaluated in the laboratory after exposing cocultures to two light intensities and two phosphate conditions. Species growth responded differentially to treatments. At the onset of the experiment, P. agardhii biovolume in the co-culture was three-fold higher than that of R. mediterranea (Fig. 5). Planktothrix agardhii grew under high phosphate concentration and light intensity (P200/L L ), whereas R. mediterranea notably decreased its biovolume (Fig. 5a). A similar pattern was observed under low phosphate and low light intensity (P5/L L ), albeit both species showed lower biovolumes compared to P200/L L (Fig. 5a, b). Highlight treatments significantly promoted the proliferation of R. mediterranea, regardless of the phosphate concentration (Fig. 5c, d). Light intensity changes (I est ) inside the cultures are also shown in Fig. 5.
Light intensity and Pi availability treatments led to changes in trichome features. While the average length and volume of P. agardhii trichomes significantly increased under P200/L L (Tukey's test, P \ 0.05), trichome shortness was observed under H L and trichome fragmentation was noticed under P5/ H L (Table 2). Notably, R. mediterranea trichomes were significantly longer and bigger when cultivated under H L compared to L L , and were found to be the shorter and smaller under P5/L L (Tukey's test, P \ 0.05) ( Table 2).

Responses to Pi deprivation and changes in gene expression in R. mediterranea and P. agardhii
The ability of R. mediterranea FCC LP and P. agardhii FCC LP to cope with Pi deprivation was analyzed by resuspending monoalgal cultures in MLA P-free medium. Growth curves of R. mediterranea FCC LP were similar in MLA (200 lM total phosphorous) and MLA -P over 12 days (P \ 0.05) (Fig. 6a), whereas cultures of P. agardhii FCC LP grown without Pi displayed slightly slower growth rates compared to MLA (Fig. 6b). In addition, total phosphatase activity was determined along the growth curves. Higher enzyme activities were measured in cultures of both strains growing in MLA -P compared to MLA. However, the time course of the enzyme activity differed between species (Fig. 6a, b). Tenfold higher activity was measured in R. mediterranea FCC LP at day 3 under Pi deprivation compared to MLA, and the activity reached maximal levels a b c d Fig. 1 Seasonal variation of environmental factors and nutrient concentrations in Los Patos shallow lake over the studied period. a Daily mean values of rainfall and air temperature, and water temperature measured in situ when sampling; b total nitrogen (TN) and total phosphorus (TP); c dissolved inorganic nitrogen (DIN) and soluble reactive phosphorus (SRP); d TN:TP and DIN:SRP ratios  Aphanizomenon gracile and Sphaerospermopsis aphanizomenoides were grouped together as ''Aphanizomenon-Sphaerospermopsis (Apha-Spha) complex'' for the quantitative analyses since the diagnostic features (position and form of the akinetes) needed to distinguish between them were not always present (99.7 ± 11.3 lmol lg -1 chlorophyll a h -1 ) at day 4 (Fig. 6a). On the contrary, alkaline phosphatase activity in P. agardhii FCC LP was five-fold higher at day 10 compared to the control, while the activity reached a maximum at day 11 (Fig. 6b).
The expression of genes involved in phosphorous metabolism was up-regulated after cell transfer to MLA -P . Transcripts from phoA-like and phoD (members of the Pho regulon, encoding alkaline phosphatases) were approximately 20 and 50% higher after 24 h of Pi starvation, respectively (Fig. 7a, b). In regards to polyphosphate metabolism genes, the highest transcript levels were observed 48 h after transferring cells to MLA -P , with a 1.5-fold change for ppk transcripts (polyphosphate kinase) in R. mediterranea FCC LP (Fig. 7c, d). Transcripts for the ppa (pyrophosphatase) in R. mediterranea FCC LP and P. agardhii FCC LP increased within 24 h of Pi starvation (Fig. 7e, f).

Discussion
This study shows that physiological features of R. mediterranea may play an influential decisive role in coexistence with P. agardhii under environmental changes. This is particularly important because P. agardhii is considered to be successful in creating an environment in which tends to outcompete other cyanobacteria (Kurmayer et al., 2016).
The long-standing dominance of cyanobacteria in the turbid and hypereutrophic Los Patos lake was characterized by perennial blooms of P. agardhii (Oscillatoriales) that alternate with blooms of R. mediterranea and other Nostocales species. The increase in P. agardhii biomass from late summer led to a decrease in light availability that, together with the fall in water temperature, could have avoided further development of R. mediterranea in autumn (Figs. 2, 3). Similar cycles of Oscillatoriales and Nostocales were observed in temperate lakes from Europe (Kokociński et al., 2010;Toporowska et al., 2016). Interestingly, the occurrence of R. mediterranea was also registered in autumn and winter albeit in low proportions, indicating that the species was not completed outcompeted by P. agardhii under conditions of low transparency and low temperature.
Analyses integrating physiology and ecology of cyanobacterial species are needed to better predict their dynamics. In this work, complementing field studies and laboratory work highlighted the importance of changes in light, temperature and phosphorus availability in the dynamics of R. mediterranea and P. agardhii. We only used one strain of each species, so the conclusions of the co-culture experiments cannot necessarily be generalized for both species.
When co-cultures were exposed to L L (40 lmol photon m -2 s -1 ), independently of Pi concentration, P. agardhii growth avoided R. mediterranea proliferation, probably by creating light-limiting conditions (Fig. 5). Similar results were reported in co-cultures of P. agardhii and R. raciborskii, where P. agardhii grew faster and outcompeted the latter under 20 lmol photon m -2 s -1 (Ammar et al., 2014). Notably, in our experiments, R. mediterranea managed to persist in L L treatments (Fig. 5a, b). On the other hand, R. mediterranea grew under both H L treatments regardless the initial higher density of P. agardhii (Fig. 6c,  d). Taken together, these results reveal that R. mediterranea has certain tolerance to low and medium light intensities. Although little is known about optimal light requirements for R. mediterranea, the phylogenetically related R. raciborskii has been shown to have a high level of flexibility with respect to light and tolerance to a wide range of light  intensities (Pierangelini et al., 2015;Burford et al., 2016). In addition, in field studies, R. mediterranea has also been found as the dominant species in turbid, light deficient environments and also in deep water layers (Moustaka-Gouni et al., 2010;Padisák et al., 2009). Raphidiopsis mediterranea and P. agardhii growing in co-cultures differed in their morphological responses to different light and Pi environmental conditions. Trichomes lengthened and presented highest volumes under P200/LL, the most favorable growth condition for P. agardhii. The shorter trichomes observed under H L could be due to high-light intensity stress. Furthermore, physiological stress was enhanced under Pi limitation, which led to trichome fragmentation (Table 2). Accordingly, trichome shortening has been observed in Planktothrix cultures when increasing light intensity and decreasing phosphorus concentration (Hašler et al., 2003). On the other hand, R. mediterranea showed larger and bigger filaments under H L and low Pi availability (P5 treatments), conditions that did not affect trichome development. These results suggest a phenotypic plasticity, which has not previously been reported for this strain ( Table 2). The greater variation in the morphology of R. mediterranea could imply a higher tolerance to environmental conditions, as suggested for R. raciborskii (Xiao et al., 2017).
Additionally, laboratory experiments showed that R. mediterranea exhibited flexible features to overcome short periods of phosphate depletion. Even though both strains increased alkaline phosphatase activity, the response of R. mediterranea was faster and higher. This could imply an advantage over other organisms under environmental Pi starvation. Accordingly, R. mediterranea was shown to coexist with P. agardhii due to the rapid regulation of Pi uptake systems under fluctuating Pi availability (Aubriot et al., 2011;. The expression of genes related to the hydrolysis of phosphomonoesters (phoA-like and phoD) was regulated by external Pi availability in R. mediterranea FCC LP and P. agardhii FCC LP (Fig. 7), as previously reported in other strains (Orchard et al., 2009;Harke et al., 2012;Bai et al., 2014). On the other hand, the expression of ppk and ppa genes (polyphosphate synthesis and utilization, respectively) under Pi deprivation suggests that those strains can acclimate to environmental Pi fluctuations by modulating the polyphosphate granule metabolism. To our best knowledge this is the first report on the capacity of R. mediterranea to cope with Pi deprivation via phosphatase alkaline activity and rapid expression of pho genes. Recent reports showed that R. mediterranea is able to coexist with P. agardhii over long periods of strong Pi-deficiency, even at Pi concentrations in the nanomolar range (Aubriot & Bonilla, 2018).
Many lake management plans seek to reduce total phosphorus concentrations to mitigate algal blooms (Schindler et al., 2008). Furthermore, the collapse of perennial P. agardhii blooms was reported after a decrease in Pi concentrations as a consequence of water management practices performed in a French lake (Catherine et al., 2008). Our results indicate that P. agardhii and R. a b Fig. 6 Response of Raphidiopsis mediterranea FCC LP (a) and Planktothrix agardhii FCC LP (b) under two contrasting phosphorus availability conditions. Growth (circles) and total alkaline phosphatase activity (triangles) of cells cultured in phosphate-sufficient (black) and phosphatelimiting (white) MLA media. Error bars represent the standard deviations of the means of three independent experiments with two replicate samples mediterranea respond differently to light and Pi availability. Furthermore, the persistence and dominance of R. mediterranea may occur after a change in the environmental conditions, for example, better illuminated water column and low Pi. Our findings indicate that management plans to control cyanobacterial blooms should take into account the dominant species and the changes in their dominances according to their eco-physiological traits.

a b
c d e f Fig. 7 Quantification of relative transcript accumulation in Raphidiopsis mediterranea FCC LP (left) and Planktothrix agardhii FCC LP (right) following Pi starvation. Target genes include phoA-like and phoD (from the Pho regulon) and ppa and ppx (involved in polyphosphate metabolism). a, b transcript levels of phoA-like and phoD, respectively; c, d transcript levels of polyphosphate kinase genes (ppk); e, f transcript levels of inorganic pyrophosphatase genes (ppa). For each gene, relative transcript levels, presented as a ratio under MLA -P (0 lM P-PO 4 ) and MLA ?P conditions (full MLA, 200 lM P-PO 4 ), were determined at 24 and 48 h following the transfer of cells to the new growth medium (MLA -P and MLA ?P , respectively). The qPCR results show the means and standard deviations (error bars) for data from three independent experiments with three replicate samples