Introduction

Harmful cyanobacterial blooms are one of the most severe problems in freshwater ecosystems, nowadays. Filamentous Cylindrospermopsis and colonial Microcystis are the most successful bloom-forming organisms in many freshwater ecosystems [44, 46, 48, 66] and are common genera in tropical waters [23, 26].

Because of its potential toxicity and increased frequency of blooms, Cylindrospermopsis raciborskii has become one of the most notorious occurrences in phytoplankton communities. Since early 1990s, C. raciborskii has shown an expansion in its distribution in many aquatic ecosystems worldwide, including tropical [16, 41], subtropical [14, 71], and temperate lakes [22, 64]. This could be attributed to its high phenotypic plasticity and wide ranges of tolerance to key environmental factors [5]. C. raciborskii is particularly known to tolerate low light [8] and its dominance has been attributed to it [6, 47]. It is also known to have high storage capacity and high affinity for phosphorus [27].

In some tropical reservoirs, C. raciborskii is replacing other species of cyanobacteria, such as Microcystis aeruginosa and Anabaena spp., or becoming co-dominant [41, 43, 61]. This shift in dominance has been attributed to changes in availability of nutrients [41, 67] and of light [61]. Although other ecological interactions like grazing or virus susceptibility can contribute to population dynamics, the replacement of several cyanobacteria species by C. raciborskii in tropical aquatic ecosystems may be related to competition for resources, because of the apparently invasive behavior of the species [47].

When competing for limiting resources, species with lower requirements of a critical nutrient (R*) will be the stronger competitor [65]. In a similar fashion, theory of competition for light predicts that the species with lower requirements of critical light intensity (I*) will be the stronger competitor [25]. Fluctuation of physical factors directly influences availability of resources, thus affecting ecological interactions such as competition [35].

On one hand, when interspecific competition for both nutrients and light is considered, grounds for competitive exclusion are possible, because one species can be the best competitor for a limiting nutrient (ex. present the lowest R*) as well as the best competitor for light (lowest I*) and, consequently, will outcompete all others [2, 9, 50, 63]. On the other hand, when competition for multiple resources is considered, coexistence is also possible when one species presents the lowest R* and another presents the lowest I*. Species are thought to have trade-offs in their competitive ability for nutrient vs. light [25, 32, 33]. If a species has a better ability to compete under low nutrient, this will make it a weaker competitor under low light. Trade-offs in the utilization of different resources arises because of limited energy and materials that can be devoted to acquisition and utilization of a particular resource [35].

Theoretical investigation shows that variable supply of resources can promote coexistence and increase diversity [34]. Moreover, experimental studies have shown that many species of phytoplankton can coexist when a limited number of nutrients regulate their growth rate [18]. Sommer [62] distinguished several differing adaptive strategies for contending with variable supplies of nutrients. Species might be relatively velocity adapted, in which high rates of cellular growth and replication are matched by suitably rapid rates of nutrient uptake (elevated maximal uptake rates of nutrients—V max), or else, they may be more storage-adapted, in which rapid, opportunistic uptake rates exceed relatively slow rates of deployment in growth, thereby permitting build up intracellular storage pools for future growth. Species may also show a tendency to be more or less affinity-adapted due to its low half-saturation constant requirement (K s) or low V max/K s [60, 62]. Phytoplankton species can present physiological trade-offs in abilities to acquire and utilize resources. These trade-offs define contrasting ecological strategies of nutrient acquisition [35].

Interestingly, when intraspecific competition is considered, different strains may perform as different species the same way as described above, with strains being the strongest competitors or even presenting trade-offs among them. Differences in competitive abilities between toxin and non-toxin producing strains of Microcystis under light limitation and CO2 concentration have already been demonstrated [29, 68]. Under environmental conditions that limits cell growth, microcystin-producing strains of Planktothrix agardhii were clearly winning out over the non-microcystin-producing ones [7].

From both an ecological and a physiological point of view, some studies have shown that cyanobacteria show high diversity at the intraspecific level [28, 70, 73, 74]. Moreover, it is clear that important differences exist between strains of the same species, particularly in response to nutrients, light, and toxicity [7, 69, 72]. However, many laboratory-based physiological studies focus on only one or a few congeneric or conspecific cyanobacterial strains [27, 31, 39, 40, 53], and this limits our understanding of the role that genetic or physiological variation among strains may play in determining growth and potential of bloom formation [73].

Phenotypic and genetic diversity in physiology (requirements, toxicity) within and between populations is, therefore, a successful ecological strategy for bloom-forming species. Interactions between different populations of these species controlled by nutrient depletion, grazing, chemical warfare, and even virus infections comprise the dynamics of a bloom.

For Microcystis, a recent study found distinct genotypes in ecosystems of two watersheds with significant differences in salinity, soluble reactive phosphorus and dissolved inorganic nitrogen concentrations, pH, and water transparency and concluded that although this cyanobacterium is globally distributed, local microdiversity exists and may be linked with environmental regulation [42]. Other studies indicated the presence of two or three [4, 30] or even six M. aeruginosa genotypes in a single water body [3]. This high level of genetic diversity of Microcystis strains demonstrates the potential for major shifts in dominant genotypes and caution is needed in global generalizations of the physiological or environmental tolerance, toxicity, or bloom dynamics [3].

Genetic variability has also been observed among strains of C. raciborskii [21]. A study with ten strains of C. raciborskii from temperate and tropical areas showed a high variability of growth in response to temperature and light intensity [8]. Other studies suggested the existence of distinct ecotypes, based on differences in morphology (size and shape), ecophysiology (responses to light, phosphate, and temperature tolerance), and genotype of strains, even for those isolated from the same geographical region [10, 51].

In this study, we aimed to evaluate competition outcomes among different strains of C. raciborskii and M. aeruginosa when exposed to conditions of light limitation or phosphate limitation. Our hypothesis is that the critical requirements of phosphorus (P*) and of light (I*) of two strains of each species will differ, and trade-offs (the stronger competitor for phosphorus will be a weak competitor for light and vice-versa) between competitive abilities for phosphorus and light would lead to alternative stable states depending of the pair of strains in competition experiments.

Methods

Species

After a series of pilot experiments with eight strains isolated from different Brazilian water bodies, we selected two pairs of strains that were able to grow in our chemostats under P limitation. The experiments were performed with two C. raciborskii (Woloszýnska), Seenayya et Subba Raju 1972 strains — CP and CS, and two M. aeruginosa (Kützing) Kützing 1846 strains — MIRF and LEA. Strains were obtained from culture collections of Laboratory of Ecophysiology and Toxicology of Cyanobacteria (LETC), Federal University of Rio de Janeiro (Brazil), and of Laboratory of Aquatic Ecology (LEA), Federal University of Juiz de Fora (Table 1). Microcystis strains were grown as single cells, not in colonies. Both strains produced a variety of microcystins (Table 1, determined by LC-MS/MS as described in Lürling and Faassen [38]). Cultures were not grown axenically, but regular microscopic inspection revealed that biomass of heterotrophic bacteria remained well under 1 % of total biovolume. C. raciborskii filaments were straight and only strain CP produced several saxitoxins.

Table 1 Characteristics of the four strains used in this study
Full size table

Culture Conditions

In all experiments, temperature was kept at 20 °C, in a temperature-controlled room. Incident light intensity (I in) was set at 40 μmol photons m−2 s−1 of constant irradiance. Since temperature, light intensity, and medium used in culture collections differed from experimental conditions, the strains were acclimated before the experiments. pH was monitored on alternate days using a Thermo Scientific pH electrode Refillable Ag/AgCl (Orion 9102SC) with model SA720 Ankersmit pH meter. Monocultures under phosphate limitation were performed in Kitasato flasks of 500 mL, and dilution rate was set to 0.12 day−1, illuminated from above. Both monocultures under light limitation and phosphorus and light competition experiments were performed in laboratory-built flat chemostats with an optical path length (“mixing depth”) of 5 cm according to Huisman et al. [24]. Each chemostat consisted of a flat culture vessel illuminated from one side to obtain a unidirectional light gradient. Chemostats had, on average, a 1.8-L of effective working volume and dilution rate was set to 0.12 day−1.

Growth Medium

Continuous cultures were provided with modified WC medium [37] using two different concentrations of phosphate. We used 4.5 μmol L−1 K2HPO4 in experiments under phosphorus limitation and 350 μmol L−1 K2HPO4 in experiments under light limitation. Values followed concentrations used by Passarge et al. [50] to induce phosphate and light limitation. KCl was supplemented to prevent K deficiency owing to reduced K2HPO4. Under light limitation, concentration of nitrate in the medium was 7 mmol L−1 NaNO3. Imposed limitations were confirmed by pilot experiments. Under phosphorus-limited conditions, steady-state population biovolume increased only in response to an increase in the phosphorus supply. Under light-limited conditions, steady-state population biovolume increased only in response to an increase in the light supply.

Light Penetration (I out)

Light intensities (PAR from 400 to 700 nm) penetrating through cultures (I out) were measured on alternate days with a quantum sensor LI-190SA attached to a light meter LI-250 (LICOR, Lincoln, Nebraska, USA) in 15 randomly chosen positions on the back surface of the chemostat vessel.

Chlorophyll a and Efficiency of Photosystem II (ϕ PSII)

Chlorophyll a concentration (microgram per liter) and efficiency of photosystem II (ϕ PSII) of M. aeruginosa and C. raciborskii strains were measured on alternate days with the PHYTO-PAM phytoplankton analyzer (Heinz Walz GmbH, Effeltrich, Germany).

Biovolume and Population Densities

Cultures were sampled on alternate days and biovolume (cubic micrometer per milliliter) were measured in triplicate with an automated cell counter (Casy Cell Counter, Schaerfe System GmbH, Reutlingen, Germany) with a 120-μm capillary, directly after sampling. The cell counter did not discriminate between species. So, in competition experiments, population densities were counted microscopically in a Bürker-Türk hemocytometer. At least 400 individuals were counted (error <10 %, p < 0.05, [36]). The biovolume of each strain was estimated from the product of the population and mean cell volume of each strain. The percentage of contribution was calculated and then applied to the total biovolume to estimate percentage of contribution of each species.

Phosphate Concentrations

Concentrations of dissolved phosphate in the culture medium were measured with a continuous flow analyzer (CFA, Skalar Analytical BV, Breda, The Netherlands).

Maximal Phosphorus Uptake Rates (ν max)

The strains growing in phosphorus-limited steady-state chemostats were transferred to phosphorus-free batch cultures with the same light intensity and temperature. After addition of a saturating pulse of phosphorus (4.5 μmol L−1 K2HPO4), the inorganic phosphorus concentrations in the batch cultures was measured each 10 min during 4 h. ν max was determined as the initial linear slope of the curve of phosphorus concentrations vs. time.

Critical Requirements of Phosphorus (P*) and Light (I out*)

The critical requirement of phosphorus (P*) and light (I out*) was determined in monocultures as the steady-state concentrations of dissolved phosphorus and as the steady-state light penetration [50]. Pairs of competitors were defined based on their ecophysiological characteristics and competitive abilities for phosphate (comparing P*) and light (comparing I*). We considered MIRF and CP the strongest competitors and LEA and CS the weakest competitors (see results below). We tested two pairs, MIRF × CP and LEA × CS, under phosphate and light competition.

Statistical Analysis

The P*, I out*, and biovolume were tested for differences between strains using one-way ANOVA in the statistical tool pack PASW® Statistics version 17.0 (SPSS Inc.). Means that were significantly different were distinguished by a Tukey's post hoc comparison test (P < 0.05).

Results

Critical Requirements of Phosphate and Light and Ecophysiological Predictions

The observed variability for the strains' traits measured in the monocultures was high not only between P-limited and light-limited conditions, but also when we evaluate the interspecific and intraspecific differences (Table 2). The differences among strains of M. aeruginosa were higher than those observed for C. raciborskii strains. MIRF showed the lowest value of I* and CP the lowest value of P*. When we compare the P* and I*, we can even find a trade-off among strains for P and light critical requirements (Fig. 1). Based on this result, we can infer that under P limitation, Cylindrospermopsis strains will win the competition, and under light-limited conditions, the Microcystis MIRF will be the strongest competitor.

Table 2 Biovolume, light penetration, dissolved phosphorus concentration, cellular chlorophyll a content (mean ± SD, n = 5), cell surface:volume ratio, and maximum uptake rate of phosphorus (ν max) measured at steady-state in monocultures under phosphorus and light limitation
Full size table
Fig. 1
figure 1

Critical requirements of phosphate and light and ecophysiological predictions of the outcome of competition for phosphorus and light between the strains of Microcystis aeruginosa and Cylindrospermopsis raciborskii. The solid lines represent the zero net growth isoclines (ZNGI) of the strains

Full size image

Competition for Phosphorus

In the competition experiments under phosphorus limitation, we observed different results for the two pairs of strains tested. For the pair LEA × CS, the total biovolume decreased (from 0.42 mL L−1) until the eighth day and then stabilized (0.17 mL L−1 on average) until the end of experiment (Fig. 2a). This variation in biovolume was followed by a change in the relative contribution of competitors. Initially, Microcystis (LEA) showed an increase in dominance coinciding with biovolume reduction. From the eighth day, there was a gradual shift in dominance, and Cylindrospermopsis (CS) displaced Microcystis. The pH remained in the range of 7.7 to 8.1. Although the competitive exclusion was not observed in the time span of the experiment, Cylindrospermopsis contributed with more than 97 % of the total biovolume by the end of the experiment (Fig. 2a).

Fig. 2
figure 2

Competition experiments under phosphorus-limited conditions. a LEA (Microcystis) × CS (Cylindrospermopsis). b MIRF (Microcystis) × CP (Cylindrospermopsis). Bars represent the percent of contribution of each strain to total biovolume. Solid line—total biovolume in the chemostat vessel. Dotted line—concentration of dissolved phosphate (the arrows indicate the days that the chemostat was re-inoculated—details see text)

Full size image

In the competition between MIRF × CP under P limitation, the system did not reach a steady state (Fig. 2b). The pH varied slightly in the range of 7.5 to 8.0. A gradual decrease in biovolume occurred (from 0.31 ml L−1), and 6 days after the start of the experiment, it represented only 27 % of the initial biovolume (0.08 mL L−1). The physiological status of the cells, evaluated through chlorophyll content and the efficiency of photosystem II, revealed a poor condition (average ϕ PSII = 0.01). The culture system was re-inoculated in the 7th and 21st days (Fig. 2b). In both times, a repeated pattern was observed, with decline in 80 % of biovolume in the next 6–7 days after the inoculation. But at these times, the photosynthetic activity indicated a healthy condition of the cells (average ϕ PSII = 0.13). The chemostat was inoculated one more time (31st day), with a lower biovolume (0.22 mL L−1), and the biovolume dropped again. The Cylindrospermopsis strain (CP) was always quickly washed out from the chemostat. In 3–4 days after the inoculation, its contribution decreased to less than 10 %. Microcystis strain (MIRF) was always also washed out, although it took more time (Fig. 2b).

Competition for Light

Similar to P-limited competition, the results of the competition experiment under light limitation showed that Cylindrospermopsis can dominate or can be displaced by Microcystis depending on the strain (Fig. 3). In the experiment with the pair LEA × CS, the total biovolume increased quickly until the 10th day, but only reached the steady-state after 30 days (on average 6.5 mL L−1). The increase in biovolume resulted in reduction of the light penetration in the chemostat. The I out rapidly dropped to values around 2.0 μmol photons m−2 s−1. As the biovolume increased and reached carrying capacity, the I out decreased in parallel with a rise in the percentage of Cylindrospermopsis (CS) contributing to total biovolume. It displaced Microcystis (LEA) and became the dominant species, although the complete exclusion could not be observed in the time span of the experiment (Fig. 3a).

Fig. 3
figure 3

Competition experiments under light-limited conditions. a LEA (Microcystis) × CS (Cylindrospermopsis). b MIRF (Microcystis) × CP (Cylindrospermopsis). Bars represent the percentage of contribution of each strain to total biovolume. Solid line—total biovolume in the chemostat vessel; dotted line—light penetration through the chemostat vessel (I out)

Full size image

In the competition experiment between MIRF × CP under light limitation, the biovolume reached a steady-state (on average 2.33 mL L−1) only after 19 days (Fig. 3b). The light penetration was strongly reduced in few days after the start of the experiment and remained < 2.0 μmol photons m-2 s−1 until the 13th day. The I out showed values around 2.2 μmol photons m−2 s−1 from day 15 to 27 and was > 3.5 μmol photons m−2 s−1 in the end of the experiment. In the first 7 days of the experiment, Microcystis strain (MIRF) increased its contribution to around 80 % of the total biovolume. Cylindrospermopsis (CP) was displaced but was not excluded and remained in the chemostat representing ≅20 % of the total biovolume (Fig. 3b). In both experiments, the pH varied slightly (range 8.4–8.9), and was always below carbon limited conditions ∼10, indicating that the experiments were performed under light limitation.

Discussion

Competition for resources will be possible only when a resource is limiting. Species or strains with lowest critical phosphorus concentration (P*) will be the superior competitor under phosphorus limitation [65]. When light is limiting, the superior competitor will be the species with the lowest critical light intensity (I*) [24]. The results of this study clearly demonstrate that the competition for nutrients and light between M. aeruginosa and C. raciborskii cannot be predicted accurately without taking intraspecific variability into consideration.

Competition for Phosphorus

According to P* values, C. raciborskii (strains CP and CS) should be a better competitor for phosphorus than M. aeruginosa (strains LEA and MIRF). These results are in agreement with other studies that pointed out C. raciborskii as a good competitor for phosphorus [27]. Although Cylindrospermopsis CS excluded Microcystis LEA, the expected outcome of phosphorus competition between Microcystis MIRF and Cylindrospermopsis CP was not confirmed by our results. Cylindrospermopsis CP did not win and both strains were washed out (Fig. 2b). This was an unexpected result since both strains could reach easily a steady-state when growing in monocultures in P-limited chemostats.

The inoculums of the studied strains came from monocultures growing under the same P-limited condition and were already adapted to both limitation and continuous culture conditions. Hence, we can rule out differences in the physical environment, leaving two other possibilities to explain the highly unexpected, but repeatable outcome. With the inoculums, some compounds with allelopathic effects might have been introduced. Some recent studies have found evidence of allelopathic interactions between cyanobacteria [17, 19, 45]. In fact, based on these observations, Souza et al. (in prep.) performed experiments to test if exudates of Microcystis (strains MIRF and LEA) and Cylindrospermopsis (strains CP and CS) can affect the growth of each other. In contrast to the prediction derived from the chemostat experiment, no growth inhibitory effects for the MIRF/CP pair were found.

Another possibility is related with costs for the uptake of phosphorus under competition stress. When cyanobacteria are deficient in phosphorus, they synthesize periplasmic or extracellular phosphatases and exhibit an increased capacity to take up phosphate [15, 20, 58]. At phosphorus concentration similar to our experiments (6.5 μmol L−1), Microcystis increase alkaline phosphatases activity, implying that P at this concentration was not sufficient for adequate growth [59]. The total biovolume at steady state could be indicative of the lower growth. The average biovolume at steady state in the competition experiments between LEA × CS (0.17 mL L−1) was lower than that observed for CS in monocultures experiments (0.21 mL L−1). These results also could indicate in the direction of energetic costs of competition. Both organisms were capable maintaining stable population densities in monocultures under identical P limitation at the same and even at higher dilution rates. Apparently, costs associated with the interspecific competitive environment were much higher than in environments with a single genotype. Overall, this experiment clearly revealed that the interaction between two organisms might yield unexpected results. More research is needed to unravel the underlying physiological changes and determine the mechanisms for the additional costs.

As we started the experiments with a high biovolume, the initial phosphorus concentration in the culture medium (4.5 μmol P L−1) was rapidly reduced to nanomolar concentrations. Considering the maximal phosphorus uptake rates (ν max) estimated for Cylindrospermopsis and Microcystis strains (Table 2) and the initial biovolume of each strain (155 mm3 L−1), we can estimate a maximal uptake of phosphorus ≅20 μmol h−1. Even if we consider that the uptake rate of phosphorus was half of ν max, we can infer an uptake of ≅10 μmol h−1, which is about two times the concentration in the medium. In both experiments, the uptake system had to be adjusted and cells had to invest more energy to get the scarce resource and slow the growth [15]. This means that competition for phosphorus was rapidly intensified.

Our hypothesis is that Cylindrospermopsis strains took more time to adapt to this situation and initially Microcystis could be favored by its higher ν max. Considering its lower P* values, Cylindrospermopsis could be considered an affinity strategist when compared to Microcystis. Hence, compared to Cylindrospermopsis, Microcystis could be considered more storage-adapted. There is some evidence of a time lag between the uptake and increased cell quota of phosphorus and the increase in growth rate [1, 12]. Although ν max can be variable and dependent on ambient conditions [55, 56], a higher ν max could be more advantageous than higher maximum growth rates in competition under phosphorus limitation [1, 13]. Growth strategists have higher maximal uptake rates, which is an adaptation to ensure high uptake rates of nutrients. Combined with storage strategy of building up intracellular storage pools for future growth, such a capability would be advantageous in nutrient-depleted habitats [64].

Competition for Light

C. raciborskii could be considered as low light-adapted species [8, 47] and M. aeruginosa as a species adapted to high light conditions [11, 57]. The differences in light interception among these species are mostly due to morphological characteristics: the slender, attenuated forms of filamentous Cylindrospermopsis are superior to large colonial Microcystis at harvesting light [54]. On the other hand, M. aeruginosa is known by its resistence to photoinhibition and to high light due to protective pigmentation [49, 54]. However, according to I* values, M. aeruginosa (strain MIRF) could also be considered a good competitor for light, and in some cases, even better than the C. raciborskii (strains CP and CS). Our estimations of I* were made on strains of Microcystis growing in single cells and were similar to data obtained in another study (2.3–2.8 μmol photons m-2 s−1) [24] and for other species of cyanobacteria [24, 50]. Furthermore, Microcystis LEA also grew as single cells but presented the highest I* in this study. Although single cell growth is not prevalent in nature, single cells of M. aeruginosa were very common in a tropical reservoir where MIRF strain was isolated [61]. So the morphology of growth did not influence the low values of I* obtained for Microcystis strains.

In addition of the adaptive mechanism that makes the elongated shapes good light antenna, the cells can simply increase the cell-specific light-harvesting capacity by the synthesis of more chlorophyll. Microcystis is known to respond to light availability with photoacclimation. Under low light (20–40 μmol photons m-2 s-1), it can have three times more cellular Chl a content than when growing under high light intensities [52]. The I* values were inversely related with the cellular Chl a content (Table 2) and may be a response to low light availability.

The results of competition experiments confirmed the predictions based on I* values from monocultures. CS strain outcompeted LEA and MIRF-displaced CP. But in the last case, the Cylindrospermopsis strain was still contributing with 20 % of biovolume on average and both species coexisted. This was possible because I* values of MIRF and CP were very close to each other.

In synthesis, our data show that under light limitation, competition between the weakest predicted competitors led C. raciborskii to dominate. Between the strongest predicted competitors, the opposite was observed; M. aeruginosa displaced C. raciborskii, but both strains co-existed in equilibrium. Under phosphate limitation, competition between the weakest competitors led C. raciborskii to exclude M. aeruginosa, and between the strongest competitors, the opposite was observed, M. aeruginosa displaced C. raciborskii. However, the system did not reach an equilibrium and both strains were washed out. Hence, the outcome of the competition depended on the pair of competing strains and not only on species or on type of limitation. We conclude that existence of different trade-offs in traits between strains and between species underlie our results showing that C. raciborskii can either dominate or be displaced by M. aeruginosa when exposed to different conditions of light or phosphate limitation.

By not showing constancy of competitive abilities among strains (different rates of growth, nutrient uptake, or toxin production), such intraspecific variability grants environmental plasticity to the species, but it also challenges clear-cut conclusions about behavior of organisms and outcomes of ecological interactions among species. We expected that forecasting based on one strain studies are not fully considering plasticity of species and may lead to narrower hypothesis about population dynamics of these organisms.