Samples were collected monthly from December 2010 to March 2011 from four pre-determined sites in the Hartbeespoort Dam (Figure 1), based on four factors:

Algal cells were collected in 500 mL glass bottles from each site during the summer season. Samples were transported to the laboratory in cooler boxes filled with ice blocks and processed the same day by filtration. For algal identifications, samples were treated according to [56,57] and analysed within 24 h of collection.

Algal cells were filtered on pre-weighed GF/C glass-fibre filters (47 mm, 0.45 µM). Filters were washed further by flushing with distilled water to remove any superficial microcystins remaining on cells. Filters were stored at −20 °C before freeze-drying. After freeze-drying, weighed algal cells (1.0 g) were extracted using 70% MeOH_(aq) and then the extracts were subjected to CHCl₃/MeOH/H₂O liquid-partitioning (7/6/3, v/v/v) to remove pigments, co-eluting compounds and other cartridge-blocking material [58] prior to SPE extraction using Waters^TM HLB cartridges [59].

On-site physicochemical parameters (conductivity, temperature, pH and dissolved oxygen) were measured immediately at each site using multifunction meter (YSI^TM 6-Series Sonde, Marion, Germany). The determination of the concentrations of chlorophyll-a, nitrates and phosphates as well as species identification were carried out in a parallel project that dealt with the assessment of nutrient loading in the Hartbeespoort Dam (Zoology Department, University of Johannesburg).

Organic solvents (MeOH, CHCl₃, MeCN, HCOOH) were of high purity analytical grade and/or HPLC grade, >99% purchased from Merck (Johannesburg, South Africa) and/or Alfa Aesar (Karlsruhe, Germany). Microcystin standards (MC-RR, -YR and -LR) were purchased from CyanoBiotech GmbH (Berlin, Germany) and were supplied by Industrial Analytical (Pty) Ltd (Johannesburg, South Africa). SPE cartridges (Waters Oasis^TM HLB cartridge: 60 g, 3 mL) were purchased from Waters Inc. (Milford, MA, USA), and supplied by Microsep (Pty) Ltd, SA (Johannesburg, South Africa). Samples were filtered through Whatman GF/C glass-fibre filters (porosity 0.45 µM, Whatman International Ltd., Maidstone, UK).

The SPE extracts were analysed using a modified method reported in the literature [12]. Briefly, a Waters^TM LC-MS instrument (with a Waters^TM 3100 Mass Detector) coupled with an electrospray ionisation source (ESI) was conditioned for 10 min with the mobile phase shown below. MCs separation was performed on a conditioned C₁₈ column (Waters Symmetry300^TM, 4.6 mm × 75 mm, 3.5 μm) using the Alliance Waters^TM e2695 separation module. The mobile phase consisted of a mixture of 0.5% (FA) in Milli-Q water (A) and 100% acetonitrile (B). The column was operated at room temperature, and the microcystins were eluted within a gradient window of 20 min using a reverse phase system consisting of 25% B (10 min), 70% B (10 min), 25% B (11.1 min), 25% B (20 min). Flow rate and sample injection volumes were set at 0.5 mL/min and 0.5 μL, respectively. The ion source was operated on both positive and negative ESI modes for all experiments. Structural identification of microcystins in samples was based on retention time and fragment ion products relative to respective elution window/fragmentation pattern of authentic analytical standards, results were complemented with literature values.

Reference standard curves were established from linear regression values of respective authentic microcystin standards using Sigmaplot^TM Software (Version 8). Linear regression equations were established giving acceptable R² values for each authentic standard solution. Major intracellular microcystins (MC-RR, -YR, and -LR) isolated from M. aeruginosa cells were quantified based on linear regression equations using peak areas [60,61,62]. All experiments on field samples were performed in triplicate and quantities were reported in mean values (n = 3). Statistical analyses (one-way ANOVA and ANCOVA) were performed on an R2.14.1 statistical package [63] using mean values of the calculated MC concentrations (P < 0.05 denoted a significant difference).

The Hartbeespoort Dam (GPS co-ordinates: 25°43'44.56"S, 27°51'30.35"E) is one of the major water impoundments situated in North-West Province, South Africa, with its water being used for irrigation, domestic and recreation purposes [64]. Thus monitoring of the quality of its water is a primary priority for public health. However, on a daily basis, this reservoir receives millions of litres of treated wastewater that is rich in phosphates and nitrogenous species, this has led to excessive nutrient loading resulting in Hartbeespoort Dams being one of the most heavily eutrophied dams in South Africa [65,66]. Eutrophication of this dam dates back to beyond the 1970s and the water is quite often characterised by foul smell and heavy green pigmentation especially during summer seasons [67]. However, substantial lake restoration control measures are currently underway to reduce nutrient loading (phosphorus/nitrogen), and a shoreline restoration project is being undertaken [66]. Physical removal of algae and water hyacinth after trapping them with floating boom barriers is also an ongoing dam-remediation project managed by DWAF [68]. Sample-collection sites were pre-selected in areas where operational dam restoration and sampling activities would not have any effect/impact on each other. However, sample-collection sites were meant to give a large representative sample area of the dam with different features of interest relevant to public health including recreational and fishing activities and unrestricted/ease of public access (Sites S1–S4) (Figure 1). A pre-survey (and during GPS positioning exercise) showed that Site S1 experienced visible algal cell accumulation during the summer, probably due to the wind direction and effects of currents caused by the outflow of water near the water gates. In addition, this site (S1) was situated within a few metres from the water-purification facility for domestic use as well as to the animal conservation area (known as the Snake Park). Site S2 was selected due to low current and turbulence effects that led to massive algal scum throughout sampling. Algal blooms characteristically occur in calm, nutrient-rich water bodies [69]. Likewise, Site S3 was a relatively calm site, and protected from strong winds by some bushes forming part of a conservation scheme around the dam. The water mass around Site S3 is mainly from Magalies River inflow (Figure 1). Persistent small quantities of algal cells are very common around Site S3, even during the winter season (personal communication from a local resident).

Preserved cells were identified microscopically according to the Utermöhl method as described [70]. Microcystis aeruginosa species was the dominant species throughout the sampling period across all sites with an average dominance rating of over 80% (i.e., 80% to 100% of total cells identified belonged to M. aeruginosa), data not shown. These findings corroborate those reported in other previously published reports which state that over the past two decades M. aeruginosa has been dominant in Hartbeespoort Dam, particularly during the summer season [25,48,50,71,72]. Other algal species identified included M. wesenbergii (<10%), Spirulina (<5%), Planktothrix spp. (<5%), Melosira spp. (<1%) and Nitzschia spp. (<1%). However, although their presence in this dam has been previously reported [25,50,71], none of these minor species have been associated with MC production in Hartbeespoort Dam. Hence it is widely accepted that M. aeruginosa is the major producer of MCs observed in this dam.

Some important physicochemical parameters that prevailed in the dam during the summer are shown in Table 1. The proliferation and persistence of M. aeruginosa as well as MC production are dependent on a number of environmental conditions prevailing in any given water body [2,46,73,74,75]. While an increased N:P nutrient ratio is known to be crucial in this regard [76], other important parameters have previously been demonstrated to play an important role in MC production, including as demonstrated recently, optimal pH and temperature [75], as well as light intensity [25,75]. During the entire period of our study, the average pH levels in the dam ranged between pH 8.8 ± pH 0.21 to pH 7.4 ± pH 0.07. This pH range is suitable for desolvation of phosphates in a water system as demonstrated by Greenwald [77]. The availability of soluble phosphates in water increases as the pH of the solution increases towards alkaline conditions up to a pH < 8.6 [77]. The optimal pH range observed in Hartbeespoort Dam is a characteristic feature of most eutrophic waters [71,78] as it is suitable for metabolic activity and growth of toxic algae [5,71,75,79] at the expense of less favoured algae. Similar to previously reported pH values [25,38,71], floating algal cells were conspicuous on the water surface throughout the dam, including all sampling sites where M. aeruginosa cells were collected (Figure 1). The presence of larger masses of algal cells in some areas (e.g., Site S1 and Site S2), which were later shown to be dominated by M. aeruginosa, indicated that the dam conditions were favourable for M. aeruginosa growth, distribution and dominance as shown earlier [46,50].

Algal growths were observable throughout sampling dates, during which the average temperature in December 2010 through to March 2011 ranged between 24.2 °C and 26.2 °C (Table 1). The observed temperature range is typical for summer seasons as reported elsewhere favouring the growth of Microcystis spp. [4,28,29,33,73,74,80] as is the case in Hartbeespoort Dam [25,50]. There was an increase in the quantities (>5.94 µg/g DW) of all MCs extracted towards mid-summer, particularly by February 2011 during which period the water surface in the dam was completely green, evidenced by higher quantities of chlorophyll-a measured (data not shown). Algal cell multiplications result in an increased algal biomass paralleled by an increased quantity of chlorophyll-a being produced, as well as elevated MC production [76,81]. Higher quantities of MCs were recorded at Site S2 (Figure 1-Figure 3) where chlorophyll-a concentration was the highest, being higher than S1, S3 and S4 (data not shown). A correlation between higher total MC concentrations expressed as MC-LR eq/L and higher chlorophyll-a concentrations have been reported for algal cells collected from this dam [25,47,50]. Areas around Site S1 and Site S2 have been identified as some of the historical algae concentration zones due to their locations and conditions that favour massive algal cell aggregations and growth [82].

A decrease in DO is commonly associated with microbial activity taking place [75]. Towards the end of the summer season there was a tremendous decrease in DO across all sites, an indication that there was higher microbial activity, probably due to bacteria feeding on dying algal cells. A decrease in intracellular MCs extracted from all sites (S1 to S4) in March 2011 supports this observation (compare Figure 2 and Table 2). A decrease in intracellular MC towards the end of the summer is of particular importance in terms of public health since higher releases of extracellular MCs are expected due to cell deaths. Decreasing numbers of visible algal cells in a water column and clear water can be a deceiving water-safety indicator to users, should there be no post-bloom monitoring strategies to determine MCs in recreational water as well as in domestic water, particularly after the summer.

Pooled MC-rich hydroalcoholic extracts from a partitioning step were dried under vacuum at 40 °C, then re-dissolved in 1 mL MeOH and subjected to the SPE protocol [16]. After LC-MS instrument and column conditioning (10 min), retention times (RTs) for pure and mixed standards were established and optimised. Similarly, SPE products of algal samples were injected into the column using the auto-sampler and eluted for 20 min as had been done for the standards. Three major microcystins, namely MC-RR, -YR, -LR and a few minor microcystin variants, namely MC-WR, MC-(H₄)YR and (D-Asp³, Dha⁷)MC-RR were separated and identified based on m/z signals in positive ESI mode. The negative mode did not give good results, so it was not used.

The observed molecular ion masses, fragments and retention times were analysed relative to those of the authentic standards used in this study and compared to the literature values. The assignments were as follows: MC-RR: (m/z 1,039.0 [M+H]⁺, ADDA m/z 135, RT 6.57), MC-YR: (m/z 1,046.3 [M+H]⁺, 523.3 [M+2H]²⁺, ADDA m/z 135, RT 7.54), MC-LR: (m/z 996.3 [M+H]⁺, ADDA m/z 135, RT 7.75) and MC-WR: (m/z 1,069.3 [M+H]⁺, ADDA m/z 135, RT 8.37). The ADDA fragment ion [PhCH₂CH (OCH₃)]⁺ (m/z 135) resulted from α-cleavage of the methoxy group of the ADDA residue from microcystins [83,84] and was used as a diagnostic fragment for MC identification [13,18,85]. Except for MC-LA (not isolated in this study), our results are therefore in agreement with those of previous studies that reported on the occurrence of M. aeruginosa in Hartbeespoort Dam and the presence of various microcystin congeners including MC-RR, -YR, -LR, -LA and -WR [46,48,86]. From our work, further analysis of LC-ESI-MS data showed two other minor MCs from all algal extracts collected from all sites (December–March), which is also an indication that the species producing these MCs dominates during summer and is well distributed in the whole dam. Structures of the minor MCs corresponded to MC-(H₄)YR (m/z 1050, [M +H]⁺, m/z 135, ADDA) [83] and (D-Asp³, Dha⁷) MC-RR (m/z 1010, [M+H]⁺, m/z 135, ADDA) when compared with information gleaned from the literature [87]. These two MCs (i.e., MC-(H₄)YR and (D-Asp³, Dha⁷)MC-RR) have previously been isolated from M. aeruginosa elsewhere [62,86], but there is limited information on their production from the same species found in the Hartbeespoort Dam, thus more investigation is underway to further elucidate their occurrences and to quantify them.

Microcystis aeruginosa cells were identified from all samples collected from December 2010 to March 2011. Freeze-dried algal samples were treated similarly during MC extraction. To minimise instrumental and operational errors quantifications were performed in triplicate to generate mean values for each MC congener (Table 2). Complete removal of pigments and/or organic matter was achieved through liquid-partitioning procedures using CHCl₃ as a non-polar solvent mixed with aqueous methanol prior to SPE [58]. When not properly removed from algal samples, green pigments and organic matter lead to SPE cartridge blockages, prolonged extraction time, and poor LC-MS spectra resulting in problematic quantification [88] and low yields of MCs due to pigment influenced degradation of MCs [89] and signal interferences [59]. Higher recoveries of MC-RR, -LR and -YR have been demonstrated following a chloroform-methanol partitioning step prior to SPE protocol [59] resulting in clean and quantifiable LC-MS spectra. The insolubility of MCs in chloroform is therefore of particular advantage in the isolation and LC-MS quantification of MCs following a partitioning step using chloroform-methanol solvent system prior to SPE protocol. In this study clean LC-MS spectra were obtained following a liquid-partitioning-SPE procedure (data not shown). Microcystin-RR, -LR and -YR were quantified using peak areas substituted on linear regression equations: Y = 5,342x − 4,218: (R² = 0.9963), Y = 4,030x − 2,518: (R² = 0.9946), and Y = 3,098x + 1,172: (R² = 0.9948) of standard curves for MC-RR, -LR and -YR solutions (0.5, 1.0, 2.0, 4.0, 6.0, 8.0 and 10.0 µg/mL), respectively. Monthly mean concentrations (n = 3) were calculated for each MC congener and total MC concentrations were shown (Table 2). Aquatic environments are dynamically non-uniform in nature leading to the existence of micro-environments within any given water system. The existence of micro-environments has been shown to influence and govern factors involved in the development of algal blooms and distributions, MC productions and quantitative variations, both in time and space [79]. Micro-environmental dynamics regulate important growth factors, particularly N/P nutrient ratios, thereby affecting growth rates of MC-producing algal blooms leading to variations in MC content [90,91]. From information given in Table 1 and Figure 2 it was shown that there were visible variations in the amounts of intracellular MCs among sites as well as on a monthly basis indicating that MC production was governed by differences in micro-environments within the dam [79]. Basically, MC-RR accounted for the highest amounts of all MCs quantified at all sites and over the four months of the sampling period (Table 1, Figure 2), however, statistically, a one-way ANOVA showed that there was no significant difference in MC concentrations among months, and however, marginal significant difference among MC congeners was observed (P = 0.62 and P = 0.06, respectively) (Figure 3A and C). On the other hand, a one-way ANOVA showed that there was a significant difference (P < 0.001) in MC concentrations among sites. The highest total MC concentration of about 468 µg/g DW was recorded at Site S2 in February 2011 (Table 2). These results support the information about historical massive concentrations of algae [82] around this area, particularly M. aeruginosa cells. A multivariate analysis to determine effects of interactions between sites and MC congener on MC concentration was performed using ANCOVA (Table 3). Results showed that there was a highly significant influence of the sampling sites (P < 0.001) resulting in quantitative inter-site (spatial) MC variations (Table 2 and Table 3). However, we did not find any significant interaction between period of sampling and site (P = 0.5764), a probable indication that the same species producing MCs was dominant throughout study. Microcystin content variations, including short-term (temporal), inter-site (spatial), perennial as well as seasonal variations, are a common trend in many eutrophic systems due to changing environmental factors and species succession/or dynamics [79]. Publications on intracellular and/or extracellular MC content variations, including temporal/short-term/or inter-site (spatial) variations, have appeared in which either laboratory or actual environmental samples were investigated, including the most recent findings published [44,76,92,93,94,95], etc. Detailed referenced work on this subject was shown in Kardinaal et al. [79] and Briand et al. [91]. Thus, the observed inter-site variations in intracellular MC concentrations in the Hartbeespoort Dam were not uncommon considering the size of the dam. Moreover, this study was arguably done during the summer season immediately before the onset of the colder season when M. aeruginosa undergoes overwintering [96,97], thus it is our assumption that we were able to capture typical trends in intracellular MC profiles from samples collected representing the dam situation during the summer season. Temporal distribution of Microcystis spp. responsible for MC production in the Hartbeespoort Dam has been demonstrated in the paper published by Van Ginkel et al. [47], however, quantitative evaluations of MC dominances were not shown. In her report Van Ginkel [24] reported that cyanobacterial cells increased in a pool with time from the beginning of the summer season and decreased at the onset of the colder season from March towards the winter months, a probable indication that Microcystis cells were overwintering. From the observations made during our study, the MC profiles are evenly distributed across the dam as they were shown to occur in all samples studied, implying that they are produced from one dominant and well-distributed species, M. aeruginosa. Our findings therefore corroborate those of previous studies which showed that M. aeruginosa is a dominant MC-producing species in the Hartbeespoort Dam [25,38,46,48,50,86,98].

From this study, the observed MC-RR concentrations in the dam were found to be between 5.94 µg/g and 268 µg/g DW (Table 2) among sites. This range of concentration is within and above the range detrimental to detoxifying organs found in fish, as described in the literature [99]. Based on mouse assay the toxicity level of MC-RR (LD₅₀ = 200 µg/g to 800 µg/kg bw) is less than that of MC-LR (LD₅₀ = 25 µg/g to 50 µg/kg bw) [100], however, Ito et al. [101] showed that both MC-RR and MC-LR equally inhibit the activities of PP1 and 2A enzymes. The inhibition of PP1 and PP2A activities results in metabolic dysfunction in animals and eventual death in case of higher dosages [31]. In addition, it has also been demonstrated that at elevated concentrations, MC-RR congener had inhibitory effects against antioxidant and detoxification activities of GST on fish gills [99,102] at a dose reaching 10 µg/L. MC-RR is the dominant congener in the Hartbeespoort Dam, toxicologically, however, the fact that its dominance would therefore pose the major health hazard may be ignored, since it is the uptake rate [90,103], synergistic effects of other compounds/or MCs [37,103] and the amount [104] of MCs ingested that are physiologically more important in terms of animal poisoning than toxicity values of any particular pure MC congener. For instance, bioaccumulation of MC-RR in fish and plant tissues resulting from its elevated extracellular concentration in water [104] is of importance, as toxicological effects pose health risks for the public consuming seafood products from this water as well as other uses of the water for domestic, recreational and agricultural activities. Although there is no guideline for MC-RR concentrations in recreational waters due to its low toxicity [100], the WHO has set a limit of 25 µg MC-LR eq/L for total MC concentration in recreational waters [105]. Therefore, from Table 2, it can be deduced that upon cell lysis in Hartbeespoort Dam by the end of the summer season, MC-RR would have contributed the highest proportion to the total MC concentration in recreational water, above the WHO threshold level for recreational waters. However, as described above, upon ingestion or dermatological contact, both MC-RR and MC-LR equally inhibit the activities of PP1 and 2A enzymes, thus elevated concentrations of MC-RR alone could potentially pose a health risk to the general consumers of products or services from the dam.

In addition, Blom and co-workers [104], showed that the toxicity activity of [D-Asp³, (E)-Dha⁷]MC-RR was higher than that of the MC-RR congener to algae grazers. Thus, although further investigation is still needed to fully characterise the structure of (D-Asp³, Dha⁷)MC-RR isolated from the algal extracts we studied, its occurrence in Hartbeespoort Dam is of particular interest to us with regard to what could be its synergistic effect (if any) on the toxicity of the dominant congener, MC-RR and the like.