Effect of Iron Availability on the Growth and Microcystin Content of Natural Populations of Microcystis spp. from Reservoirs in Central Argentina: A Microcosm Experiment Approach

: The eutrophication of aquatic systems is a problem related to the contribution of excess nutrients—phosphorus (P) and nitrogen (N)—to water bodies, which produces an increase in cyanobacterial blooms. Under eutrophic conditions, P and N concentrations are sufﬁcient for cyanobacteria growth, and some micronutrients are considered to become limiting for population growth. This work aimed to assess the effect of iron on cyanobacteria growth and the content of MCs in natural populations of Microcystis spp. Microcosm setting experiments were carried out with natural samples collected during two bloom events of Microcystis spp., kept under controlled light, temperature and pH conditions. The ﬁrst bloom sample was exposed to different iron concentrations (400, 700 and 1100 µ g Fe · L − 1 ) to determine the optimum concentration for growth. The second was exposed to different iron addition modes (one: T1P, and two pulses: T2P) to imitate the iron increase produced by the downward migration of Microcystis spp. colonies. Our results show that iron is a growth-promoting factor and that its optimal range of concentrations for the growth of Microcystis spp. under the experimental setting conditions is between 700 and 1100 µ g Fe · L − 1 . On the other hand, growth rates were not signiﬁcantly different between T1P and T2P; thus, different addition modes did not have an effect on growth. Regarding microcystin content, the MC quota in natural populations of Microcystis spp. did not show a clear relationship with the iron supply. This work contributes to the understanding of the underlying factors affecting cyanobacteria bloom formation and the production of MCs, which in turn would impact the development of management strategies to control cyanobacteria blooms.


Introduction
Anthropogenic eutrophication is a serious water quality problem affecting fresh and brackish waterbodies worldwide [1][2][3]. It is associated with high inputs of phosphorus (P) and nitrogen (N) compounds from human activities in the watershed [4][5][6]. One of the consequences of eutrophication is the massive development of cyanobacteria (phenomena known as cyanobacteria blooms), which alters the normal equilibrium of the aquatic ecosystem [7]. Additionally, some groups of cyanobacteria can produce toxic metabolites that affect animal and human health and represent a serious risk for public health. Cyanobacteria blooms generally occur seasonally, but climate change conditions are expected to exacerbate them, leading to more intense and permanent blooms and to serious impairment of the water quality used as a drinking supply or for recreational purposes [3,[8][9][10].
Microcystis spp. is a unicellular cyanobacteria normally clustered in colonies of different morphologies and capable of producing toxic metabolites known as microcystins (MCs) [11]. Microcystis blooms and MCs are globally distributed, and their presence has been reported in many countries worldwide [12,13]. Argentina is not exempt from this problem, and drinking water reservoirs of important urban areas located in semiarid regions are particularly vulnerable to the impact of toxic Microcystis blooms [14][15][16].
Despite numerous studies carried out to date, the ecological role of MCs and the causes of their production are still unclear. It is well known that the growth and MC content of Microcystis spp. populations are greatly influenced by a number of environmental factors. Numerous studies have examined the effect of light and temperature [17][18][19], pH [20] and macronutrients (P and N) [21][22][23][24] on growth and MC production. Metal ions are other significant chemical factors, of which their comprehensive impact is still being studied [25][26][27][28].
Under optimal light and temperature conditions, it has been observed that Microcystis spp. blooms can be regulated by the concentration of P and N, as well as the N/P ratio [22,29,30]. However, in natural conditions, under eutrophic or hypereutrophic circumstances, the concentrations of P and N, as well as the ratio N/P, are within the required concentrations for cyanobacterial growth, so other elements become important. In the case of cyanobacteria, iron is one of the micronutrients with the highest requirements, and generally, this group has a higher demand for iron than eukaryotic cells [31,32]. This metal is incorporated into cyanobacteria cells in the form of ferrous iron (Fe 2+ ), which is the oxidative status in which iron can be transported through its membrane. Molot et al. [33] proposed that the availability of Fe 2+ is important for the ability of cyanobacteria to outcompete eukaryotic algae and for bloom formation. They suggest that some colonial cyanobacteria (e.g., Microcystis spp.) can move to bottom waters through changes in buoyancy and obtain the iron (Fe 2+ ) normally released from the sediments of anoxic nutrient-rich waters. This mechanism would prevent the deficit of iron inside the cyanobacteria cells, which is an advantage for their growth over the other phytoplankton groups.
Regarding the production of MCs, Dai et al. [26] reviews the environmental factors that influence the synthesis of MCs. Iron has also been postulated to influence the production of MCs in different ways.
One hypothesis supports the idea that MCs would act as an intracellular iron chelator [19] or as an extracellular iron-scavenging molecule [34], and as such, the iron concentration in water would directly regulate the synthesis of MCs [35]. When the iron concentration in water is low, cells would synthesize more MCs [28,[36][37][38], which in turn would act as an iron chelator, either scavenging the available iron and facilitating its uptake or storing it inside the cell.
Extensive research has been performed to study Microcystis spp. bloom formation and the ecophysiological role of MCs, as well as the effect of environmental parameters on the regulation of population growth and the production of MCs. Most of this research has consisted of either culture-based studies using pure cyanobacteria strains under controlled conditions or field studies using natural cyanobacteria populations under natural conditions (see [39] and references therein). Only a few studies have been carried out with natural populations under laboratory controlled conditions. The relevance of these types of studies is that they maintain the colony aggregation of Microcystis spp. and they control some environmental factors in optimal ranges for growth. On the contrary, if these colonies are transferred to the laboratory and isolated, they disaggregate into a unicellular form after some generations [40]. In addition, some authors suggested that colonial Microcystis has a higher tolerance to stress conditions when compared to unicellular phenotypes [41,42].
Hence, by using natural populations, we expect that our experiments may reflect the response of Microcystis spp. in lakes better than those experiments using axenic cultures.
In this work, we used natural populations of Microcystis spp. collected from two eutrophic reservoirs in a semiarid region of central Argentina in order to evaluate the response of local populations to changes in the iron concentration under controlled laboratory conditions. We evaluated the effect of the iron concentration on population growth and MC content. It is known that colonies of Microcystis spp. are able to move downward in the water column for better acquisition of nutrients in the anoxic hypolimnion (e.g., soluble iron-Fe 2+ ; [33]). For this reason, we considered not only the iron concentration in water but also different modes of supplying it as experimental treatments, simulating one or two sudden iron pulses. The iron pulse might reproduce the iron increase produced when Microcystis spp. colonies move to bottom layers during anoxic conditions.
Despite the importance of Microcystis spp. blooms in reservoirs in central Argentina, few experimental studies have been conducted for identifying the main drivers controlling these bloom events. The aim of this work was to assess the effect of iron on cyanobacteria growth and MC content in natural populations of Microcystis spp. The study focused on answering some questions raised in eutrophic systems: (1) which is the optimum range of iron concentrations for the growth of natural populations of Microcystis spp.? (2) Is there a positive effect on growth if iron is supplied in two pulses instead of only one (simulating one or two downward migration events along the population growth)? (3) Might MC content be affected by different iron concentrations and supply modes?
This study attempts to contribute to the understanding of the underlying factors affecting cyanobacteria growth and MC production and their relationship with internal nutrient dynamics in eutrophic reservoirs. In turn, this would contribute to freshwater management and the effective application of strategies to control cyanobacteria blooms.

Microcystis spp. Collection, Sample Characterization and Laboratory Acclimation
Microcystis spp. populations were obtained from the center area of two eutrophic reservoirs in the semiarid region of central Argentina (Pampean Ranges of Córdoba; [43]) during bloom events in March 2017 and February 2018. Simultaneously to the sampling, air and water temperature, conductivity, pH and dissolved oxygen were measured in situ with a multiparametric probe (YSI model 556 MPS). Samples for microcosm experiments and for determination of the phytoplankton composition, total phosphorus, inorganic nitrogen and chlorophyll a were collected at a depth of 0.10 m in 5 L acid-washed (HCl 1N) opaque plastic bottles and transported to the laboratory at 4 • C in dark conditions within 3 h.
The characterization of the phytoplankton community was made at the laboratory immediately by counting an aliquot of the sample with a compound microscope (see more details in Section 2.3). Determination of most of the nutrients was performed at the laboratory following the Standard Analytical Methods (APHA 2017): total phosphorus (TP; SM 4500-P E); nitrites (SM 4500-NO 2 B), nitrates (SM 4110 B) and ammonium (SM 4500-NH 3 A-G). Dissolved inorganic nitrogen (DIN) was estimated as the sum of the three N-fractions.
Microcystis spp. colonies were kept at the laboratory in fish tanks under natural light (~42 ± 10 µE·m −2 s −2 ) and at room temperature for 15-30 days (~22 ± 2 • C) before commencing the experiments for indoor acclimation. The phytoplankton composition did not change during this period and Microcystis spp. continued to be dominant, as confirmed through cellular counts carried out the day before starting the experiments. Cell counting was also performed for calculating a mean volume of the inoculum for reaching a concentration of Microcystis spp. corresponding to natural blooms reported in the region [15,44].

Experimental Design
Given the difficulty of working in the field with natural populations of Microcystis spp., the experimental design tried to reproduce, in the laboratory, the natural physicochemical characteristics of the subsurface layer of the water column during the austral seasons of summer and autumn. The physicochemical characteristics of the experiments were defined by taking into account mean values of the water quality data obtained from the periodic monitoring of the reservoir most studied in the region, the San Roque reservoir (SRr), during the 1999-2018 period ( Table 1). The conditions of temperature, TP and soluble iron (SFe) concentration in SRr during blooms of Microcystis spp. were specially considered. The SRr was also selected as a "model system" because of its eutrophic state, the frequent blooms of cyanobacteria (mainly Microcystis spp. and Dolichospermum spp.) occurring during late spring, summer and autumn seasons and the consequent hypolimnetic anoxia and fish death events [44]. Information on the land use and physical characteristics of the SRr and its watershed can be found in [45]. We used an "enclosed ecosystem" (hereinafter microcosm) as a biological model that simplifies and simulates complex and heterogeneous natural ecosystems ( [46] and references therein). Inside microcosms we isolated and incubated a water parcel collected from the lake under conditions similar to those found in the natural environment. This setting made it possible to conduct controlled incubations of natural populations of Microcystis spp., regulating nutrients, temperature and light, in optimal ranges for growth. Natural conditions are not completely simulated in these cases; for example, water movements are restricted. In these types of microcosm experiments, phytoplankton biomass tends to increase rapidly and therefore often reproduces bloom conditions [47].
After Microcystis spp. bloom events occurred during March 2017 and February 2018, two microcosm experiments were carried out during April 2017 (first experiment, sample-2017) and April 2018 (second experiment, sample-2018). For both experiments, an aliquot (20-50 mL) of Microcystis spp. natural populations maintained under room conditions (as described in Section 2.1) were incubated in 2 L transparent polyethylene bags (microcosms) that contained filtrated and autoclaved sterilized water from the SRr. The microcosms were placed in a plastic tank (50 L) with tap water under controlled temperature at 27 ± 0.35 • C maintained by a water heater (Aqua Zonic AZ Easy Heater) and also under a photosynthetically active radiation intensity of 43.5 ± 3.2 µE·m −2 s −2 provided by a fluorescent array with a light/dark regimen of 12 h:12 h. Lamps were located above the incubation tank for avoiding the shade effect and for obtaining an homogeneous light field. The experiment design consisted of a unique tank containing three microcosms (replicates) of each of the experiment treatments and control. Microcosms were shaken every day to obtain a homogeneous sample, and they were rearranged randomly to reduce the minor differences in photon irradiance. Initial cell densities for both experiments were selected to be inside the range of Microcystis spp. natural blooms historically reported in SRr ( [44]) and observed during the bloom events of 2017 (3.0 × 10 7 -1.5 × 10 8 ; [48]).
The first experiment was carried out during April 2017 for testing the optimal iron concentration for Microcystis spp. growth. Four iron treatments were established: (a) Control, without iron addition: natural iron concentration in the water used in experiments = 27 ± 5 µg Fe·L −1 (0.5 µM); (b) TI: +400 µg Fe·L −1 (7 µM); (c) TII: +700 µg Fe·L −1 (12 µM); (d) TIII: +1100 µg Fe·L −1 (20 µM). The three iron concentrations are representative of natural conditions and were selected considering concentrations above the median (120 µg Fe·L −1 ) and maximum values (950 µg Fe·L −1 ) in the SRr subsurface layer during the summer season ( Table 1). The initial inoculum of Microcystis spp. added to each microcosm made it possible to reach an initial cell abundance of 1.3 ± 0.4 × 10 8 cells·L −1 . The experimental period lasted for 26 days and included the initial lag stage and the exponential stage of algal growth. The end of the exponential stage was confirmed based on the optical density (OD, absorbance at 750 nm, Shimadzu spectrophotometer model UV-1700, Japan) measured daily over the whole experiment ( [49]).
The second experiment was carried out during April 2018 for testing the response of cyanobacterial growth to different iron addition modes (one or two pulses). The iron concentration was selected based on the most favorable concentration of iron for cyanobacterial growth in the first experiment. The two-pulse design aims to reproduce the increase in iron due to the downward migration of cyanobacteria colonies to deeper anoxic layers in eutrophic systems (e.g., median and maximum total iron values at SRr's hypolimnion during summer season: 290 and 2600 µg·L −1 , respectively; Table 1). The treatment setting was as follows: (1) Control, without iron addition: natural iron concentration measured in water used in experiments = 179 ± 25 µg Fe·L −1 (3 µM); (2) Treatment 1 Pulse (T 1P ): +700 µg Fe·L −1 (12 µM) at day 1 (D1); Treatment 2 Pulses (T 2P ): the same amount divided in 2, +350 µg Fe·L −1 (6 µM) at D1 (first pulse) and 350 µg Fe·L −1 at day 9 (D9; second pulse). The second pulse was set on day 9 since we considered it close to the start of the exponential growth. On the first experimental day, each microcosm was inoculated with an aliquot of Microcystis spp., and the initial cell abundance was 2.5 ± 0.7 × 10 7 cells·L −1 .
The experimental period lasted for 25 days. As in the previous experiment, the second experiment included lag and exponential stages of algal growth.
For both experiments, the iron solution was prepared by mixing 10 mL of FeCl 3 (0.2 M) in HCl (pH: 2) with 20 mL of disodium EDTA solution in NaOH (pH: 7; 0.2 M). The final pH of the solution of FeCl 3 in EDTA was 6. Under these conditions, EDTA and the iron cation Fe 3+ form a strong coordination complex and iron does not precipitate ( [50]). To avoid phosphorus limitation due to the consumption by cells over the experiment, K 2 HP0 4 was added to each microcosm to reach a final concentration of 100 µg P. L −1 , according to the median total TP value in the SRr subsurface layer during the summer season (Table 1).

Physicochemical and Biological Determinations of Microcosm Samples
During the experiments, physicochemical parameters (water temperature, conductivity, pH and dissolved oxygen) were measured in situ using a multiparametric probe (YSI model 556 MPS) in each microcosm. After mixing the whole content of each microcosm, 70 mL of water was collected with a syringe to determine soluble reactive phosphorus (SRP), SFe, cell abundance and microcystin (MC) content. Sampling was carried out early in the morning during experimental days 1, 5/4, 8, 14/15 and 25/26 during the first (sample-2017) and second experiment (sample-2018), respectively.
Samples for determining SRP and SFe were previously filtered using a Millipore cellulose acetate membrane filter (pore size 0.45 mm; Ø = 47 mm) and preserved at −20 • C and 4 • C, respectively. Samples for determining SFe were preserved with HNO 3 (0.1%; v/v). SRP was determined following the Standard Analytical Methods ( [51]; 4500-P E), and absorbances were measured in a spectrophotometer (Shimadzu UV-1700) at 880 nm. The SFe concentration was measured by adjusting the volume of each sample to 25 mL with Milli-Q water and analyzing it via atomic absorption spectrometry (AAS), using a Perkin Elmer Spectrophotometer Model AA 3110.
Microcystis spp. samples for cell abundance determination were fixed in buffered formalin (0.4% final concentration, v:v), and maintained in the dark at 4 • C. For quantification, samples were mixed well, and an aliquot of 1 mL was taken and dispensed into a Sedgewick-Rafter chamber. Duplicate samples from each experimental replicate were observed with a compound microscope (Zeiss model D-7082, Germany) using 200× magnification. Counts were made along transects until at least 20 colonies were observed. The height, width and thickness of each colony were measured and assimilated to known geometric forms in order to estimate the volume ( [52]). The individual volumes of 10 cells were also measured for each sample. The cell abundance was estimated as follows: where CA sample x is the cell abundance of sample x, CoV sample x and CeV sample x are the colony and the cell volumes, respectively, and VoT corresponds to the total volume of transects. The growth of populations of Microcystis spp. throughout the experiments was simultaneously followed using the absorbance of samples in vivo, for identifying the start and the end of the exponential growth phase. Absorbance was measured each sampling day using a spectrophotometer (Shimadzu UV-1700) at 750 nm ( [49]).
Samples for total MC analyses were taken as 2 mL sub-samples of the bottle collected for phytoplankton analysis. Sub-samples were collected after complete homogenization of the sample and stored in plastic vials at −20 • C. Total MCs were determined via an immuno-enzymatic assay (ELISA Kit, Abraxis, detection limit = 0.15 µg·L −1 ), which detects all variants of MCs due to its capacity to bind specifically to the ADDA moiety present in all MC congeners. Sub-samples were subjected to a pretreatment of three thaw-freeze cycles (to lyse the cells and release the MCs) and then filtered through acetate-cellulose membrane filters (pore size, 0.45 µm; Ø = 13 mm) to remove detritus (remaining broken cells) and to recover the soluble MCs. As MCs are intracellular peptides and are not actively excreted by cyanobacteria cells, using this procedure, it is possible to calculate the average amount of MCs per cell (MC quota). The MC quota was then calculated as the total MC concentration over Microcystis spp. cell abundance.
where MC quota sample x is the average amount of MCs per cell or the quota of sample x, [MCs] sample x is the MC concentration, and CA sample x is the cell abundance, as in Equation (1).

Data and Statistical Analysis
Iron treatments were performed in triplicate. All data are reported as the mean and standard deviation.
The specific growth rate (µ) of Microcystis spp. populations was calculated from the cell abundance during the exponential phase according to the following: where N f and N i are the cell abundance on the last and first day, respectively, during the exponential growth phase and (t f − t i ) is the duration of this phase in days. The rates of SFe and SRP decrease were determined from the slope of a linear regression throughout the whole experiment.
A one-way ANOVA test was performed to determine differences in the cell abundance, growth rate, SRP and SFe decrease, MC concentration and MC quota among iron treatments for each day, using a significance level (α) of 0.05 for all comparisons ( [53]). Subsequently, a Tukey test was performed to distinguish the treatments with significantly different effects. The normality and homoscedasticity of data were checked using the Shapiro-Wilk test and Breusch-Pagan test, respectively. In all cases, R v. 3.6.3 software ( [54]) was employed.

Characterization of Phytoplankton Community during Microcystis spp. Blooms
The sample obtained from the bloom during March 2017 was mostly composed of Microcystis spp. (95.2%); other species of Dinophyceae, Diatomeae, Chlorophyceae and Cryptista were also present and represented a low percentage of the total phytoplankton community ( Table 2). This population presented a very low concentration of MCs (range: 0.1 to 0.6 µg·L −1 ) and also low MC quota values (0.27 ± 0.05 × 10 −2 pg·cell −1 ).
The sample obtained during February 2018 mostly constituted Microcystis spp. (97.3%); other species of Cyanobacteria, Diatomeae, Chlorophyceae, Zygnematophyceae and Cryptista were also detected but contributed to a low proportion of the total phytoplankton community ( Table 2). Regarding the potential toxicity, the sample-2018 presented both high concentrations of MCs (5.7 ± 0.9 µg·L −1 ) and high MC quota values (0.18 ± 0.02 pg·cell −1 ), which were within the range of other studies with natural Microcystis spp. populations ( [39,55]). In both samples, the inspection using an optic microscope made it possible to confirm that the cells were in good condition.

Physicochemical Conditions during Microcosm Experiments
Environmental conditions inside the microcosms throughout the experiment were maintained in ranges between minimum and maximum values measured at the subsurface layer of the SRr during the summer season (1999-2018; Table 1). During both experiments, pH varied from 7 to 8, conductivity ranged from 0.15 to 0.28 Ms·cm −1 and dissolved oxygen displayed values between 6 and 8 mg·L −1 . These conditions were similar for all the treatments and are representative of SRr s subsurface layer during the summer season ( Figure 1 and Table 1). Experimental conditions were in accordance with those optimal for the growth of Microcystis spp. previously reported by several authors at different locations: temperatures within a range of 25-32 • C, alkaline pH values and light intensities of 40 to 280 µE·m −2 ·s −1 [17,18,[56][57][58]. Elevated temperatures are frequently associated with Microcystis spp. blooms with stratification conditions of the water column, which is more favorable for the growth of populations than the mixing condition, as observed in the SRr [57]. Water column stratification frequently occurs at the SRr during the summer [44].

Microcosm Experiment Testing Different Iron Concentrations
Natural populations of Microcystis spp. collected in March 2017 exhibited differences in growth when supplied with different iron concentrations (TI: 400 µg Fe·L −1 = 7 µM, TII: 700 µg Fe·L −1 = 12 µM and TII: 1100 µg Fe·L −1 = 20 µM) or when compared to that with no addition (control treatment: 27 µg Fe·L −1 = 0.5 µM). In particular, Microcystis spp. populations supplied with iron reached higher cell abundance by the end of the experiment than those without an iron supply (Figure 2A). Moreover, cells supplied with 700 and 1100 µg Fe·L −1 exhibited significantly higher specific growth rates throughout the exponential phase than cells under control and TI (400 µg Fe·L −1 ) conditions (p ≤ 0.05; Table 3). Similar results were reported by other studies that observed higher M. aeruginosa growth rates under iron-replete (28-100 µM) conditions in comparison with those under iron-depleted conditions (0.1 µM) [59,60].
In accordance with Microcystis spp. growth, the SFe and SRP concentration decreased throughout the experiment with the three iron addition treatments ( Figure 2B), but the rates of decrease were not significantly different among them (Table 3; p > 0.05). The decrease in SRP and SFe concomitant with an increase in Microcystis spp. growth would indicate the consumption of these elements by the cells. As Fe 3+ forms a very strong 1:1 molar coordinating complex with EDTA, in neutral-alkaline conditions, the precipitation of Fe 3+ in water is discarded. Experimental conditions were in accordance with those optimal for the growth of Microcystis spp. previously reported by several authors at different locations: temperatures within a range of 25-32 °C, alkaline pH values and light intensities of 40 to 280 µE·m −2 ·s −1 [17,18,[56][57][58]. Elevated temperatures are frequently associated with Microcystis spp. blooms with stratification conditions of the water column, which is more favorable for the growth of populations than the mixing condition, as observed in the SRr [57]. Water column stratification frequently occurs at the SRr during the summer [44].  According to our results and to iron values registered at subsurface layers of the SRr during the seasons with most frequent blooms (e.g., summer:~50 to 950 and autumn:~50 to 310 µg Fe·L −1 ; Table 1), iron may not be at an optimal concentration for the growth of Microcystis spp. most of the time. Instead, the greater availability of iron at deeper layers (~50 to 2600 and 50 to 1570 µg·L −1 for summer and autumn, respectively; Table 1) might be an important resource for the growth of cyanobacteria populations. This may be caused by the release of Fe 2+ from the sediment to the water column, a process that is favored by the anoxic conditions present at the deepest layer during spring, summer and autumn (see dissolved oxygen values at hypolimnion in Table 1). The variability in the iron concentration in the water column can be exploited by species with the ability to move to different depths (e.g., M. aeruginosa; [61,62]). Hence, Microcystis colonies may take advantage of sudden increases in the availability of iron by moving to anoxic bottom waters [33]. According to our results and to iron values registered at subsurface layers of the SRr during the seasons with most frequent blooms (e.g., summer: ~50 to 950 and autumn: ~50 to 310 µg Fe·L −1 ; Table 1), iron may not be at an optimal concentration for the growth of Microcystis spp. most of the time. Instead, the greater availability of iron at deeper layers (~50 to 2600 and 50 to 1570 µg·L −1 for summer and autumn, respectively; Table 1) might be an important resource for the growth of cyanobacteria populations. This may be caused Control, without iron addition (27 ± 5 µg Fe·L −1 ); TI: +400 µg Fe·L −1 ; TII: +700 µg Fe·L −1 ; TIII: +1100 µg Fe·L −1 . Values are the mean of triplicates, and vertical lines above symbols are standard deviations. Asterisks are located above the treatments that showed a significant difference in cell abundance in comparison with that in the other treatment groups (A). Table 3. Microcystis spp. natural samples collected from the 2017 bloom: specific growth and soluble iron decreasing rates in microcosms exposed to different iron treatments: control, without iron addition; TI, 400 µg·L− 1 ; TII, 700 µg·L− 1 ; TIII, 1100 µg·L− 1 . The mean and standard deviation of growth and decreasing rate values resulted from triplicate microcosms. Different letters indicate significant differences among treatments. SFe: soluble iron; SRP: soluble reactive phosphorus; n.d.: not detectable.  In this line, a subsequent experiment testing the response to differential iron addition modes (one or two pulses) was conducted. For this purpose, based on the results of the first experiment showing that cells supplied with 700 and 1100 µg Fe·L −1 exhibited similar growth rates, 700 µg Fe·L −1 was chosen as the total iron concentration for testing different pulses.

Microcosm Experiment Testing Different Iron Pulses
Samples of Microcystis spp. collected during February 2018 were supplied with 700 µg Fe·L −1 in different ways. Treatment 1 (T 1P ) consisted of one pulse of 700 µg Fe·L −1 , while treatment 2 (T 2P ) consisted of two pulses of 350 µg Fe·L −1 each. During the exponential phase, populations of Microcystis spp. displayed a significantly higher specific growth rate in the iron addition treatment groups in comparison with that in the control (p ≤ 0.05; Table 4). However, there were no significant differences in the specific growth rate between the iron addition treatments (T 1P and T 2P ; Figure 3 and Table 4). Along with population growth, SFe and SRP displayed a decrease of their concentrations, as observed based on their negative decreasing rates ( Table 4). The decreasing SRP rates measured throughout the experiment did not show significant differences between control and iron treatments. These results, along with those observed in the first experiment, confirmed that P was not completely consumed in either of the two experiments, so it is considered a non-limiting factor for the growth of the Microcystis spp. populations. As was also observed in the first experiment, SFe decreasing rates were not significantly different between iron treatments, in this case, considering the rates estimated before the second pulse (D1 to D8; Table 4).
Extensive research has demonstrated that, in general, nutrient-rich waters have anoxic sediments, conditions favored by warm temperatures and stable water columns [63,64]. Hypolimnetic anoxia was reported to promote cyanobacteria blooms in eutrophic waters by promoting Fe 2+ release from sediments and the concurrent availability for cyanobacteria. As Fe 2+ transported upward from anoxic sediments remains in a reduced form until it reaches oxygenated waters-where it is rapidly oxidized to Fe 3+ -cyanobacteria must migrate downwards into anoxic waters below the mixed layer to acquire internally loaded Fe 2+ before its reoxidation [33]. Indeed, the downward migration of cyanobacteria into anoxic waters has been observed ( [65] and references therein), and M. aeruginosa colony velocities have been measured as 52 ± 13 m·day −1 [61]. However, other hypotheses have been formulated for algal buoyancy in natural systems and appear to be much more dependent on light than on nutrients. In this case, physical mechanisms may provide the sufficient replenishment of nutrients at epilimnion from hypolimnion layers, explaining the maintenance of high growth rates of epilimnetic phytoplankton populations [66]. Table 4. Microcystis spp. natural samples collected from the 2018 bloom: specific growth and soluble iron decreasing rates in microcosms exposed to different iron treatments: control, without iron addition; with different iron addition treatments: T 1P , one pulse of 700 µg·L −1 ; T 2P , two pulses of 350 µg·L −1 each. The mean and standard deviation of growth and decreasing rate values resulted from triplicate microcosms. Different letters indicate significant differences among treatments. SFe: soluble iron; SRP: soluble reactive phosphorus; n.d.: not detectable. Period considered: * D1-D25; ** D1-D8 (before the second iron pulse).

Total MC Concentration and MC Quota under Different Iron Addition Treatments
Data shown in this section only correspond to the samples collected during 2018 (second experiment) because of the negligible values of the MC concentrations in samples-2017 (first experiment; see Section 3.1). The total MC concentration and MC quota showed a distinctive response along the growth population curve. Moreover, different responses were observed according to treatments; Microcystis in the control showed a decreasing trend in the total MC concentration throughout the experiment (from 5.0 ± 0.01 to 1.38 ± 0.63 µg·L −1 ; Figure 4A), coincident with a drop in the cell abundance ( Figure 3A). On the other hand, Microcystis exposed to iron addition treatments (T 1P and T 2P ) displayed an increment in the total MC concentration (from 5.18 ± 0.23 to 32.68 ± 14.65 or to 25.98 ± 10.05 µg·L −1 respectively; Figure 4A) along with a rise in cell abundance ( Figure 3A). In particular, T 1P showed a significantly higher MC concentration (17.88 ± 2.15 µg·L −1 ) than T 2P at day 15 (5.07 ± 0.59 µg·L −1 ; p ≤ 0.05; Figure 4A).
Regarding the MC quota, within the first week of the experiment (lag phase), no significant differences among iron treatments were observed, while cells under control conditions displayed significantly higher values at day 8 (p ≤ 0.05; Figure 4B). However, we acknowledge that the high MC quota in the control set could be caused by a low cell division rate and MC accumulation, as we observed a significantly lower specific growth rate in comparison to that with iron treatments (Table 4). On the other hand, during the exponential phase, the MC quota decreased with all treatments at day 25 and was not significantly different among them. The latter is in accordance with [67] who observed a pronounced decline in intracellular MC after day 15 with all iron treatments. However, cells under different iron treatment conditions showed significant differences in the MC quota among each other over the exponential phase; cells treated with one iron addition pulse displayed values significantly higher at day 15 (T 1P ; 0.40 ± 0.03 pg·cell −1 ) in comparison with those in cells receiving a second iron pulse (T 2P ; 0.13 ± 0.04 pg·cell −1 ; p ≤ 0.05; Figure 4B). Indeed, cells under T 2P conditions showed a significant decrease in the MC quota from day 8 to day 15, after the second iron pulse at day 9 (p ≤ 0.05; Figure 4B). These results may indicate that different iron additions applied during the exponential phase (T 1P : no pulse and T 2P : one pulse at day 9) could influence the MC content, reducing the production with an iron supply.   Regarding the growth of Microcystis spp. populations under low iron concentrations, MC synthesis has been proposed to give a selective advantage, as it may play a role in intracellular storage or as a transport peptide in iron metabolism [19,38]. In this way, toxic Microcystis strains might have a more efficient iron uptake system than the nontoxic Microcystis strains [36,37,59].
Although our results are in accordance with [36,38], who reported that the M. aeruginosa cells exposed to low iron levels displayed a higher intracellular concentration of MCs than cells under high iron conditions, we recognized that the differences in iron concentration between T1P and T2P were not so pronounced at day 15 (378 ± 16 and 540 ± 26 Regarding the growth of Microcystis spp. populations under low iron concentrations, MC synthesis has been proposed to give a selective advantage, as it may play a role in intracellular storage or as a transport peptide in iron metabolism [19,38]. In this way, toxic Microcystis strains might have a more efficient iron uptake system than the nontoxic Microcystis strains [36,37,59]. Although our results are in accordance with [36,38], who reported that the M. aeruginosa cells exposed to low iron levels displayed a higher intracellular concentration of MCs than cells under high iron conditions, we recognized that the differences in iron concentration between T 1P and T 2P were not so pronounced at day 15 (378 ± 16 and 540 ± 26 µg Fe·L −1 respectively). Hence, differences in the MC quota observed in our study could be not directly associated with the iron factor.
It is known that not all Microcystis strains are capable of toxin production and that during a bloom, a single species can shift the proportion of toxic and non-toxic cells [68]. Our results did not distinguish if the increase in the MC concentration under low levels of SFe was due to an increase in the proportion of toxic Microcystis in comparison with that of nontoxic Microcystis strains, instead of an increase in the MC quota. However, the cell abundance and MC concentration were positively correlated (R = 0.58, p < 0.001, n = 23), which would indicate that significant differences in MC concentrations among treatments might be caused mainly by the MC quota. In addition, many studies with Microcystis and Planktothrix natural populations showed that only 54% of the variation in MC concentrations could be explained by changes in the proportion of MC-producing cells, suggesting that a considerable part of the MC concentrations was also due to variations in the MC quota [69].

Conclusions
Our data showed that natural populations of Microcystis spp. under conditions of high temperature and sufficient light in the water column (e.g., austral summer and autumn season conditions), as well as under a high P supply (e.g., eutrophic reservoirs), appear to have growth responses driven by other factors, such as the iron supply. From our results of a Microcystis spp. natural population grown under optimal light, temperature, pH and P conditions and considering the range of iron concentrations in the subsurface layer of an eutrophic system, the San Roque reservoir, we conclude the following: 1. iron is a growth promoting factor; 2. the optimal range of iron concentrations for growth is between 700 and 1100 µg Fe·L −1 ; 3. growth is not significantly different when iron is supplied in different modes (i.e.,:700 µg Fe·L −1 supplied in a single pulse at first day or in two pulses at the start and at the half of the experiment); 4. a clear relationship between the iron supply and MC quota is not demonstrated.
From our data, we recognized the iron supply (e.g., from internal sources) as an important factor determining the growth of populations of Microcystis spp. in eutrophic systems. Thus, it is essential to continue more studies related to the role of trace metals in cyanobacterial growth, bloom formation and the production of MCs in natural populations, in order to gain a better understanding of the potential effects of micronutrient internal sources on eutrophic systems.  Institutional Review Board Statement: Ethical review and approval were waived for this study due to "Not applicable" for studies not involving humans or animals.