Introduction

Freshwater phytoplankton communities are highly variable in space and time, and respond rapidly to changes in their physical, chemical and biological environment (Varol and Sen 2018). Highly diverse, low biomass phytoplankton communities are indicative of healthy freshwater systems (Shao et al. 2019). Conversely, high biomass, less diverse phytoplankton communities often dominated by bloom-forming cyanobacteria tend to persist in systems anthropogenically modified through increased nutrients or flow restriction (Mitrovic et al. 2003; Bormans et al. 2004; Dignum et al. 2005; O’Neil et al. 2012). Many cyanobacteria can produce biologically active secondary metabolites, known as cyanotoxins, which can have severe ecological, economic and human health impacts (Bowling 1994; Bormans et al. 1997; Falconer 2001; Pearson et al. 2010; Rastogi et al. 2015).

Optimal growth of phytoplankton depends on the availability of several key nutrients. Among these, phosphorus (P) and nitrogen (N) are required in the largest quantities and are commonly a growth-limiting factor in freshwater systems (Paerl and Otten 2013). The role of these macronutrients in stimulating cyanobacterial blooms is well documented (Hunt and Matveev 2005; Paerl and Otten 2013; Mueller and Mitrovic 2014; Schindler et al. 2016). Micronutrient trace metals also play key roles in a multitude of biological processes and are cofactors in numerous cyanobacterial proteins (Baptista and Vasconcelos 2006; Facey et al. 2019). There is emerging evidence they can influence cyanobacterial growth alone or in combination with macronutrients (Lukac and Aegerter 1993; Downs et al. 2008; Molot et al. 2010; Harland et al. 2013; Polyak et al. 2013; Sorichetti et al. 2014).

The availability of macronutrients and micronutrients play a key role in structuring the phytoplankton community (Vyverman et al. 2007). Nutrient requirements within the phytoplankton community are highly variable, leading to interspecific competition for nutrient resources (Sourisseau et al. 2017). As different phytoplankton groups have distinct nutrient requirements and means of nutrient acquisition, the addition of a nutrient can cause differential responses in different segments of the phytoplankton community. This process, termed ‘community colimitation’, can cause an alteration to the overall structure of the community (Arrigo 2005). For example, Molot et al. 2014 proposed that iron regulates the ability of cyanobacteria to compete with eukaryotic algae and cyanobacterial dominance can be supressed in P-loaded systems by reducing Fe2+ availability. Further, the growth of heterocystous cyanobacteria will likely be more dependent on molybdenum availability than non-heterocystous cyanobacteria due to its role in the assimilation of inorganic nitrogen (Glass et al. 2012).

While the importance of trace metal micronutrients for phytoplankton growth is becoming more evident, little is known about the extent of micronutrient limitation in freshwaters or how increased concentrations of micronutrients may alter phytoplankton community structure. Identifying how phytoplankton communities, particularly those that include toxin-producing cyanobacteria, respond to different macronutrient and micronutrient regimes is crucial to making informed, effective catchment management decisions. Our research had two aims, firstly, to understand the extent of micronutrient limitation and/or colimitation of cyanobacterial growth in some South Eastern Australian freshwater systems. Secondly, to understand how various phytoplankton communities change in response to micronutrient amendments and to observe which conditions favour cyanobacteria. We hypothesise that (1) micronutrients will be a limiting factor of cyanobacterial growth in some freshwater systems and (2) changes in phytoplankton community structure will occur with increased micronutrient concentrations. We chose to use a mixture of iron, cobalt, copper, manganese, molybdenum and zinc as these are required by some or all phytoplankton at a biochemical level (Facey et al. 2019).

Materials and methods

Study sites

Seven sites were selected across New South Wales and Victoria, Australia. Study sites were chosen because they were known to have varying levels of cyanobacteria present in warmer months. They comprised of lakes, rivers and creeks and are summarised in Table 1. Sampling occurred between the months of November to February when water temperatures and light intensities were not limiting.

Table 1 Summary of study sites, sampling dates and locations
Full size table

Microcosm enrichment assays

In situ nutrient enrichment microcosms were conducted to determine which nutrients were limiting phytoplankton growth and to test for any nutrient-driven changes in community composition after a 7-day incubation period, similar to Mueller and Mitrovic (2014). Approximately 60 L of surface water was filtered through a 63 µm plankton net into a large plastic container. Water was filtered to exclude zooplankton grazers. 1.0 L clear PET bottles were filled from the container, leaving some air space at the top. Nutrient additions were conducted according to the six treatments outlined in Table 2. All treatments were conducted in triplicate.

Table 2 Summary of treatments and nominal concentrations of the target nutrient additions
Full size table

Following the nutrient additions, bottles were mixed by rotation and tied together in random order. They were suspended at the same depth within the euphotic zone using floats (approximately 90 % surface irradiance). Concentrations of nitrate and phosphate were selected to alleviate any macronutrient limitation while remaining within levels typically found in natural Australian systems. They resembled those used by Mueller and Mitrovic (2014) (500 µg/L N, 200 µg/L P), as they effectively stimulated growth and had no toxic effects. Micronutrient additions resembled the concentrations of the cyanobacterial growth medium, MLA (Bolch and Blackburn 1996) and were low enough to avoid any toxic effects. Samples for micronutrients, nitrate/phosphate, physiochemistry, chlorophyll a and phytoplankton enumeration were taken in triplicate from the filtered water at the onset of the experiment. Nitrate/phosphate and micronutrient samples were also taken from surrogate bottles with added nutrients and micronutrients to determine the total concentration of the addition plus the ambient concentration. Samples for chlorophyll a and phytoplankton enumeration were taken after 7 days from each sample bottle.

Nutrient sampling and analysis

In the field, 50 mL of water sample was filtered through a prerinsed 0.45 µm cellulose acetate syringe filter (Sartorius) and frozen immediately. Bioavailable nitrate and phosphate concentrations were determined photometrically using Flow Injection Analysis on a QuikChem 8500 Lachat nutrient analyser. For analysis, frozen samples were slowly thawed to room temperature. Soluble reactive phosphorus (srP) was measured by the reduction of ascorbic acid using the molybdate blue method (Murphy and Riley 1962). Nitrate and nitrite (NOx) was determined following reduction by a cadmium column using the sulphanilamide method (APHA 1998).

Trace metal micronutrient analysis

In the field, 25 mL of water sample was filtered through a 0.45 µm cellulose acetate syringe filter (Sartorius) prerinsed with 50 mL of 10 % nitric acid followed by 100 mL milli-Q water. Samples were collected in 50 mL falcon tubes and refrigerated. Falcon tubes had been soaked overnight in an acid bath (10 % nitric acid v/v) and rinsed repeatedly with Milli-Q water. Within 24 h of collection, samples were acidified with ultra-pure nitric acid to 0.2 % v/v. The concentrations of micronutrients in the filtered solution were analysed by inductively coupled atomic emission spectrometry (ICP-AES) (Varian 730 ES). The spectrometer was operated according to the standard operating procedures outlined by the manufacturer. The instruments were calibrated using matrix-matched standards. At least 10 % of samples were conducted in duplicate to ensure the precision of the analyses. To check for potential matrix interferences at least 10 % of samples had spike recoveries performed.

Phytoplankton identification and enumeration

200 mL grab samples were taken at the beginning of the experiment (day 0) from the large container and from individual microcosms on day 7 after homogenization by mixing and preserved with Lugol’s Iodine solution (~ 0.25 % v/v). Samples were identified and enumerated at 200 times magnification using a light microscope (Olympus BX41) and Sedgwick-Rafter counting chamber. If required, samples were concentrated 5× prior to counting by settling in 50 mL measuring cylinders for 24 h. The upper 40 mL was removed after checking all phytoplankton had settled and were no longer present in the upper layer. Phytoplankton taxa were identified to a genus level using identification literature by Prescott (1978), except for potentially toxic cyanobacteria which were identified to species. Counting precision was performed to ± 10 % with at least 100 units of the dominant taxa counted following Hötzel and Croome (1999). Biovolumes were calculated using the most appropriate conversion factors from Newcombe (2012) and Olenina et al. (2006).

Chlorophyll a analysis

200 mL of sample water was filtered on site via vacuum filtration onto GFC glass fibre filters (Whatman) and frozen for preservation. Chlorophyll a was analysed according to (Mueller and Mitrovic 2014). The glass fibre filters were extracted in 10 mL 90 % ethanol heated in a 75 °C water bath for 10 min. Unwanted filtered material was removed by centrifuging at 3000 rpm for 10 min. The supernatant was analysed immediately using a Varian Cary 50 Bio UV Spectrophotometer at wavelengths 665 and 750 nm.

Statistical analysis

Phytoplankton biovolume, cyanobacterial biovolume and chlorophyll a were analysed using permutational analysis of variance (PERMANOVA) with PRIMER + PERMANOVA software ver. 6 (Anderson 2001) with a significance level of α = 0.05. Community analyses (nMDS, PERMANOVA, SIMPER and Inverse Simpson Diversity Index) were performed using the vegan package (Oksanen et al. 2019). A square root transformation was performed on the community data prior to analysis to reduce the influence of extreme values and plots were created with the ggplot2 package (Wickham 2016) using the software R Version 1.2.1335 (Team 2018). Inverse Simpson Diversity was measured in terms of biovolume (Behl et al. 2011) and used algal data identified to the genus level as this is a useful resolution for assessing changes in community structure (Nielsen et al. 1998).

Results

The effect of nutrient additions on phytoplankton communities was highly variable based on locations. Limitation by either macronutrients or micronutrients are indicated by increases in the biovolume of some or all groups within the phytoplankton community (Table 4). At two locations that had cyanobacterial dominance and high cell concentrations (Burrendong Dam and Mannus Lake), the micronutrient mixture stimulated cyanobacterial growth, suggesting that one or multiple micronutrients were limiting cyanobacterial growth. This was not observed at the other bloom sites on the Murray River at Mildura and Euston (Fig. S2), both of which had very low nitrogen and phosphorus concentrations at the beginning of the experiments (Table 3). Nitrogen, phosphorus or a combination of the two (co-limitation) regularly limited phytoplankton growth, as observed at Morpeth, Windeyers Creek, Lake Lyall, Burrendong Dam, Mildura and Euston (Table 4).

Table 3 Ambient concentrations of macronutrients and micronutrients
Full size table
Table 4 Summary of results from seven nutrient amendment bioassays across South-Eastern Australia
Full size table

Cyanobacterial biovolume was strongly influenced by the addition of micronutrients at Burrendong Dam and Mannus Lake (Fig. 1). At Mannus Lake there was no significant difference between cyanobacterial biovolume in the control (C), nitrogen (N treatment), phosphorus (P treatment) or nitrogen + phosphorus (NP treatment) treatments (PERMANOVA: p value > 0.05). However, in the micronutrient treatments (M and NPM) cyanobacterial biovolume was significantly greater than the control and all other treatments (PERMANOVA: NPM vs. control p value = 0.028, M vs. control p value = 0.011). Similarly, at Burrendong Dam the addition of micronutrients alone increased cyanobacterial biovolume relative to the control (p value = 0.001). Nitrogen alone also had a stimulatory effect on cyanobacteria relative to the control (p value = 0.001). There was no significant difference between chlorophyll a results across the different treatments at either Burrendong Dam (PERMANOVA: p value = 0.188) or Mannus Lake (p value = 0.448) (Fig. S1).

Fig. 1
figure 1

Total phytoplankton and cyanobacterial biovolume in Mannus Lake and Burrendong Dam microcosms. Asterisk represents significant difference compared to the control (PERMANOVA, p value < 0.05). The nutrient concentrations added for each treatment are listed in Table 2. Error bars are standard error of the mean, n = 3

Full size image

At Mannus Lake the phytoplankton community composition was also significantly affected by macronutrient additions (Fig. 2) (PERMANOVA: p value = < 0.001). In the P treatment, growth of green algae was stimulated. However, phosphorus additions did not cause a significant change in cyanobacterial biovolume relative to the control (PERMANOVA: p value = 0.325). Conversely, the addition of the micronutrient mixture (M), even in the presence of phosphorus (NPM), increased the growth of cyanobacteria, which was already dominant. There was a clear distinction between phytoplankton communities at Mannus Lake in treatments with the micronutrient mixture and those without it (Fig. 2). SIMPER analysis demonstrated that the largest contributor to the differences between all treatments was Chryosporum ovalisporum. The increase in C. ovalisporum in the M treatment contributed up to ~ 95 % of dissimilarity compared to the control, while the reduction in C. ovalisporum in the NP treatment contributed 71 % of the dissimilarity compared to the control. Mougeotia and Dictyosphaerium were the key genera of green algae that responded to phosphorus addition in the P and NP treatments. Similarly, at Burrendong Dam the M treatment had a higher proportion of cyanobacteria compared to the control while the phosphorus addition favored a reduction in the proportion of cyanobacteria and a higher diversity community (Fig. 2). SIMPER analysis demonstrated that Microcystis and Radiocystis were the largest contributors to differences between all treatments, while Scenedesmus, Cryptomonas and Chlamydomonas were the largest non-cyanobacterial responders to the NP addition compared to the control.

Fig. 2
figure 2

Proportion of community made up of several key phytoplankton groups at Mannus Lake and Burrendong Dam (left). Shannon Diversity Index (middle) and nMDS plots (right) illustrating differences in phytoplankton community structure between treatments. A square root transformation was performed on the community data for nMDS. Stress < 0.2. Error bars are standard error of the mean, n = 3

Full size image

Discussion

In situ nutrient bioassays were conducted at seven locations throughout South Eastern Australia to assess the extent of trace metal micronutrient limitation of cyanobacterial growth and to identify how increased micronutrient availability influences phytoplankton community structure. Of the seven freshwater systems examined, two exhibited signs of micronutrient limitation of cyanobacterial growth: Mannus Lake and Burrendong Dam. At Mannus Lake, a dense cyanobacterial bloom had established which was dominated by the heterocystous cyanobacteria Chrysosporum ovalisporum, a producer of the toxin cylindrospermopsin (Shaw et al. 1999; Quesada et al. 2006; Yilmaz et al. 2008; Fadel et al. 2014). Cyanobacterial biovolume significantly increased in treatments containing the micronutrient mixture (NPM and M treatments) (Fig. 1) primarily driven by increased growth of the bloom forming C. ovalisporum. The addition of nitrogen and phosphorus alongside micronutrients (NPM treatment) did not increase the effect size as there were no significant differences to the phytoplankton response to micronutrients alone. Interestingly, the Murray River at Euston and Mildura experiments were also undergoing a bloom of a filamentous, nitrogen-fixing cyanobacteria, Dolichospermum crassum, but the response from the micronutrient addition was not observed at either location on the Murray River. Instead, phosphorus was the limiting factor for cyanobacterial growth at Mildura, and to a lesser extent at Euston (Fig. S2). A similar study conducted by Sterner et al. (2004) found phosphorus limitation at Lake Superior and did not observe any limitation of algal growth by micronutrient trace metals (manganese, iron or zinc). However, Sterner proposed that the system was on the cusp of micronutrient limitation but suggests this may have been clouded by the simultaneous limitation of phosphorus (North et al. 2007).

Chrysosporum ovalisporum is often dominant in low nitrogen concentrations where heterocystous cyanobacteria have an advantage over other phytoplankton (Fadel et al. 2014). Nitrogen fixation requires high levels of iron (Sterner et al. 2004; Molot et al. 2014), molybdenum (ter Steeg et al. 1986; Paerl et al. 2006) and cobalt (Rodriguez and Ho 2015), as the N2 fixing enzyme nitrogenase contains metal cofactors. This causes heterocystous cyanobacteria to require some trace metals in higher amounts than other phytoplankton (Schoffman et al. 2016) and may make them more prone to micronutrient limitation (Kustka et al. 2002; Molot et al. 2010; Romero et al. 2013) observed significant increases in nitrogen fixation upon addition of both iron and molybdenum and suggested co-limitation involving trace metals is common in lakes. A similar phenomenon may have caused the increase in C. ovalisporum growth in Mannus Lake upon the addition of the trace metal micronutrient mixture. C. ovalisporum had already established a dense bloom so it is possible that nutrient constraints were beginning to come into effect. Given that iron was relatively available at the onset of the Mannus Lake experiment (Table 4), molybdenum and cobalt are more likely to be the limiting micronutrients. Both were below detection limit.

At Burrendong Dam, which was dominated by the microcystin-producing genera Microcystis aeuginosa and Radiocystis sp. (Vieira et al. 2003; Rastogi et al. 2015), the micronutrient treatment (M) had a slightly higher proportion of cyanobacteria than the control, and the NPM treatment had a higher proportion of cyanobacteria than the NP treatment. This indicates that cyanobacteria may be more successful competitors in the phytoplankton community with higher micronutrient concentrations. The addition of micronutrients alone (M) and nitrogen (N) stimulated cyanobacterial growth relative to the control. Although the NPM treatment was higher, it was not statistically different (p value > 0.05) to the control (Fig. 1). M. aeruginosa and Radiocystis remained dominant under all treatments. The large stimulatory effect of nitrogen on cyanobacteria at Burrendong Dam was not observed at Mannus Lake where the heterocystous Chrysosporum ovalisporum dominated. It has been suggested that reduced nitrogen input will cause an increase in the proportion of N2 fixing cyanobacteria (Schindler et al. 2008). The relatively low availability of NOx at the onset of the Mannus Lake experiment was likely a contributing factor to the dominance of C. ovalisporum and given its ability to fix atmospheric nitrogen, nitrate is unlikely to limit C. ovalisporum growth. Conversely, Microcystis and Radiocystis depend on dissolved nitrogen for growth, which had become limiting by the onset of the Burrendong Dam experiment.

As Microcystis and Radiocystis, the dominant cyanobacterial genera at Burrendong Dam, are non-nitrogen fixing, the limitation of growth by micronutrients in this system was unlikely to be related to nitrogen fixation. Iron is also required for the reduction of nitrate to ammonia prior to assimilation (via nitrate and nitrite reductase) (Schoffman et al. 2016). Sub-optimal iron availability appears to be able to limit nitrate uptake in natural waters (DiTullio et al. 1993). At low iron concentrations, and without the presence of highly bioavailable ammonia, the phytoplankton community can be co-limited by iron and nitrogen (Saito et al. 2008; Schoffman et al. 2016). For example, North et al. (2007) suggested that iron enrichment reduced nitrogen limitation by allowing NO3 assimilation in nutrient enrichment bioassays. However, this is not supported by our results as the addition of nitrate alone in the N treatment stimulated cyanobacterial growth at Burrendong Dam, suggesting there was sufficient Fe in the ambient water to allow for nitrate reduction and assimilation. The simultaneous limitation of the community by nitrate and micronutrient trace metals at Burrendong Dam, combined with the lack of response in the NP and NPM treatments, is difficult to elucidate.

These results demonstrate that micronutrient trace metals can stimulate cyanobacterial growth in-situ and may act as an important regulator of the severity of cyanobacterial blooms. This study joins a growing list that have observed an important role of micronutrients in structuring phytoplankton communities and increasing cyanobacterial growth in physically and chemically diverse freshwater systems. For example, Downs et al. (2008) noted a stimulation of the cyanobacterium Anabaena flos-aquae upon addition of cobalt, copper, manganese and a trace metal mixture, while a number of studies have observed iron limitation of cyanobacteria growth (Wever et al. 2008; Molot et al. 2010; Harland et al. 2013; Fujii et al. 2016).

Phosphorus‐driven changes in community structure

In both experiments the addition of phosphorus promoted higher diversity in the phytoplankton community composition. Green algae, diatoms and dinoflagellates made up a larger proportion of the community in P and NP treatments (Fig. 2). This trend is surprising as the addition of P decreases the N:P ratio, which is generally expected to favour cyanobacterial growth (Tew et al. 2014; Li et al. 2018). However, the opposite effect was observed at both Burrendong Dam and Mannus Lake. The change in community composition may indicate that the systems were also phosphorus limited at the time and green algae and diatoms were able to respond faster to the sudden phosphorus pulse due to their faster growth rate compared to cyanobacteria (Lürling et al. 2013; Deng et al. 2014). Alternatively, each species is likely to have different nutrient requirements and therefore some species can be nutrient limited whereas others are not (Baptista and Vasconcelos 2006; Mueller and Mitrovic 2014). This may explain why phosphorus limitation was not evident when assessing total phytoplankton biomass. This trend is particularly evident in Burrendong Dam where phosphorus concentrations were very low. Interestingly, when the phosphorus addition (P or NP) was coupled with the micronutrient mixture (NPM) the communities were composed of a notably higher proportion of cyanobacteria, particularly at Mannus Lake. This suggests that micronutrients impart a competitive advantage to cyanobacteria over other components of the phytoplankton community even under high phosphorus conditions. This may be because of specific micronutrient requirements of cyanobacteria or a result of a more efficient metal uptake system (Baptista and Vasconcelos 2006; Sunda 2012), for example via the production of metallophores (Kraemer et al. 2015).

Implications for management and research

Micronutrient trace metals appear to be an important regulator of the severity of cyanobacterial blooms in some freshwater systems. Improving our understanding of how specific micronutrients influence phytoplankton community structure and cyanobacterial growth could be an important aspect of catchment management plans and may be critical to securing freshwater resources into the future. In both micronutrient limited sites, high-density cyanobacterial blooms had established by the onset of the experiment. Limiting micronutrient inputs may help to reduce the severity of such blooms. All the micronutrients used in this study are common additions to many fertilizers (Molina et al. 2009). Over application of fertilizers and subsequent runoff may be a significant source of trace metals in freshwater systems as well as N and P. This risk could be minimized through more targeted application of fertilizers or by increasing vegetation in the riparian zone to act as a buffer for micronutrient inflows, which are already effective measures for reducing macronutrient inflows (Aguiar et al. 2015).

Many trace metals (such as Co, Cu, Fe, Mn and Zn) can be released from sediments under anoxic conditions caused by thermal stratification (Shipley et al. 2011). These micronutrients can become available to cyanobacteria who may vertically migrate to nutrient-rich hypolimnial waters (Bormans et al. 1999; Wagner and Adrian 2009; Molot et al. 2014), particularly in shallow reservoirs such as Mannus Lake. Further, when the water column mixes after periods of thermal stratification upwelling occurs, increasing the availability of micronutrients in surface waters. Breaking down or supressing the formation of thermal stratification via maintaining high flow velocities in rivers or by installing mixers (such as fans or bubble plumes) are commonly used to manage blooms in systems where cyanobacterial buoyancy mechanisms are a primary driver of their dominance (Mitrovic et al. 2011; Visser et al. 2016; Bormans et al. 2016). These mixers may also be effective in reducing sediment-derived micronutrients in systems prone to cyanobacterial blooms by preventing anoxic conditions at the water-sediment interface.

Conclusions

This study has provided insight into the extent of micronutrient limitation of cyanobacterial growth in Australian freshwater systems and how the phytoplankton community changes in response to micronutrient additions. We hypothesised that micronutrients will be a limiting factor of cyanobacterial growth in some freshwater systems. Two sites out of seven exhibited signs of micronutrient limitation. Both of these sites had high cyanobacterial biovolume at the onset of the bioassays, suggesting that micronutrients may become limiting during high competition for nutrient assimilation during bloom events. This suggests that micronutrient trace metals can regulate the severity of cyanobacterial blooms in some freshwater systems. Micronutrients also influenced phytoplankton community structure, supporting our second hypothesis. At both sites showing micronutrient limitation of cyanobacteria, the addition of the micronutrient mixture resulted in higher proportion of cyanobacteria compared to the control, suggesting that micronutrients can not only influence cyanobacterial biovolume but also their ability to compete with other phytoplankton. These results may have important implications for the management of micronutrients and cyanobacterial blooms in freshwater systems.