1. Introduction
Cyanobacteria are a well-known group of photosynthetic bacteria which can be found globally, being distributed from polar regions to the tropics [
1,
2]. Within this broad classification of organisms there exists a range of species which can cause problems for humans, via mechanical clogging of filter equipment [
3], creation of hypoxic water bodies leading to death of aquatic organisms [
2], reduction in recreation and tourism [
4] and via the production of toxins capable of affecting humans directly [
4,
5,
6,
7]. Factors influencing cyanobacteria growth include light intensity, water temperature and nutrient (phosphate and nitrate) availability [
8,
9,
10,
11], with the latter typically resulting from increased levels of precipitation linked to agricultural and industrial nutrient run-off promoting eutrophication [
12]. Of those genera of cyanobacteria containing species known to produce toxins,
Microcystis occurs commonly around the globe with issues caused by toxic
Microcystis having been reported from Australia, Brazil, China, Portugal, Sweden and the USA [
4,
5,
11,
12,
13] amongst others. In the UK, in addition to
Microcystis spp., other cyanobacteria genera were reported, including
Oscillatoria,
Plantothrix,
Anabaeana,
Pseudoanabaena,
Aphanizomenon,
Snowella and
Gomphospaeria [
8,
14,
15].
Microcystins (MCs) are the primary toxins associated with cyanobacteria, including
Microcystis spp., and they represent one of the most common and most studied groups of cyanotoxins [
1,
4]. MCs are cyclic heptapeptides with more than 240 known analytes which vary in toxicity [
4]. In humans and other mammals these toxins act as hepatotoxins, inhibiting protein phosphatases [
11] and are known to be tumorigenic [
16], neurotoxic [
17] and genotoxic [
18]. Routes of exposure for humans include intravenous injection [
5], skin contact [
19], inhalation [
19] and ingestion, either directly in the form of drinking or unintentionally through recreational water activities [
20,
21], or potentially via a food vector [
19,
22]. More detailed information on human intoxications is presented in the recent review by [
4], based on a wealth of global cases. Testing methods have been developed to allow for the detection of these toxins, primarily in water. In turn this has led to the World Health Organisation (WHO) recommendation of a drinking water guideline value for microcystin-LR (MC-LR), of 1 μg/L for life long consumption [
23]. To date this remains the only cyanotoxin group which has received such guidance in part due to being well studied. Consequently, the MCs have become the toxin group which has received the most attention as a framework for assessing risk.
Within the United Kingdom (UK) there are few reported cases of microcystin related intoxications originating from cyanobacteria. These cases primarily relate to intoxications of animals, such as sheep and dogs [
13,
24], after drinking
Microcystis contaminated lake water. However, cases of human intoxications have been reported following exposure to
Microcystis and
Plantothrix blooms [
6,
20].
Although, to date, cyanobacterial blooms in the UK have not resulted in widespread intoxications of humans through direct exposure or via drinking water, the presence of these harmful species within UK water bodies does not rule out the possibility of future contaminations. Work conducted in the late 1980’s showed a prevalence of microcystins in 68% of 91 bloom sites surveyed within the UK [
13], suggesting that the potential for microcystin intoxications exists across the country. The current stance by Water UK is that microcystins pose a limited threat to UK drinking water supplies [
25,
26]. Direct contact through recreational use of contaminated water bodies remains a potential route for exposure. In recent years, open-water swimming has become a highly popular sport, with thousands of swimmers involved in more than 170 mass events each year in the water bodies around the country [
27,
28]. Over the past twenty years, various agencies have carried out assessments of freshwater bodies in the UK to determine the presence of cyanobacterial blooms [
14,
29,
30,
31,
32]. Blooms were identified in a significant number with many of these linked to eutrophication. As such, in England the Environment Agency (EA), and in Scotland the Scottish Environmental Protection Agency (SEPA) both operate programmes whereby water samples are taken from locations in response to the occurrence of visual blooms of freshwater algae. Neither organisation currently operates a routine monitoring programme to establish the frequency and intensity of cyanobacterial blooms [
14,
33]. Water samples are taken and processed using light microscopy to determine the presence and density of potentially harmful cyanobacterial species. In England as well as Scotland, water samples containing cyanobacterial species are enumerated to quantify the number of cyanobacterial cells/L. Samples containing over 20,000 cells/mL, equating to the relatively low probability of adverse health effects limit [
34], are designated as exceeding the monitoring threshold, and actions are generally taken by the relevant water body owner to prevent public exposure of both humans and animals to the affected water bodies [
14,
33]. Bloom samples showing cyanobacterial scum formation are automatically designated as exceeding the high probability of adverse health effects threshold, as they may increase cell density and bloom toxicity by up to a factor of 1000 [
33,
34].
Whilst the detection and enumeration of high cell densities of cyanobacterial genus enables water body owners to react to a potential risk, toxin production by blooms is not guaranteed.
Microcystis blooms may contain either toxin or non-toxin producing strains, and even toxic blooms may vary in the levels of toxin production [
34]. Worldwide, typically 40–70% of blooms are reported as being toxic [
13,
34]. Consequently, whilst any actions taken to close water bodies to the water body users in the event of a dense cyanobacterial bloom may be a sensible precaution, there is the potential for unnecessary closure if the blooms are non-toxic. Such closures are therefore likely to result in socio and economic impacts in at least 40% of cases on average. As such, there are notable benefits for the confirmation of bloom toxicity to enable more focused action to be taken.
Whilst visual examination of cyanobacterial blooms using light microscopy is a useful tool for detection and enumeration of potentially toxin-producing genera, this approach is not sufficient for detection of toxins [
35]. As such, toxin testing methods are necessary to investigate the potential toxicity of cyanobacterial populations during bloom incidents, as well as providing an overall understanding of the occurrence of cyanotoxins to facilitate risk assessment and risk management strategies [
11]. Toxin detection methods available include bioassays, enzyme and antibody based assays and chemical analytical techniques, such as liquid chromatography [
11,
24,
36,
37,
38,
39], allowing for the quantification of toxicity or toxin levels in environmental samples.
Recently Cefas has developed and validated a rapid liquid chromatography with tandem mass spectrometry (LC-MS/MS) method for the detection of cyanotoxins, including several of the MCs, in water samples as well as concentrated algal material, allowing for direct quantification of cyanotoxin presence from bloom samples [
40]. This method has subsequently been accredited by the United Kingdom Standards Authority (UKAS) to ISO 17025 standard. Since the development of this new method and in partnership with the EA, samples found to contain cyanobacteria and therefore suspected of containing cyanotoxins, have been analysed for their toxin content at the Weymouth laboratory of Cefas to provide confirmation of toxin occurrence and aid the overall risk assessment and risk management process. This paper presents the findings from one year of this testing.
3. Discussion
For many years, government agencies in the UK have conducted microscopic detection and enumeration of cyanobacterial species in freshwater bodies containing visual algal or bacterial blooms. Water samples found to contain potentially toxin-producing species above a cell density threshold of 20,000 cells/mL or the presence of cyanobacterial scum have triggered action by water body owners to prevent or restrict public access to the water body, thereby reducing the health risk to both humans and animals. In the absence of toxicity data, however, such action may result in the unnecessary closure of lakes and ponds. Conversely, potentially low-density blooms of highly toxic cyanobacterial cells may contain toxins at high enough concentrations to cause health effects, without triggering the cyanobacteria cell density threshold. In addition, there is a scarcity of information describing the spatial and temporal occurrence of toxic blooms in the UK, as well as the concentrations of toxins typically encountered, so the impact on the general public remains unknown. This study aimed to generate data to describe the toxicity of cyanobacterial blooms throughout a year-long period in England, and highlight any patterns of occurrence which could aid the risk assessment and management process.
Overall, LC-MS/MS results showed more than half the water samples contained quantifiable concentrations of microcystins. These findings are not surprising given that these were reactively-sampled, based on visual reports of blooms by water body owners. This proportion therefore concurs with previous reports of 40–70% of global blooms being toxic [
1,
34] and the UK work conducted in the 1980s reporting 68% of blooms containing toxins [
13]. Microcystin concentrations reported here are well above average values reported in the majority of previous studies. Codd et al. found concentrations of microcystins reaching a maximum of 131 μg/L in water from the UK following analysis by HPLC [
21], with reports from Germany and Portugal showing maximum toxin concentrations of 36 μg/L and 37 μg/L respectively [
41] and water samples from Japan reaching 480, 1300, 15,600 and 19,500 μg/L in a number of different studies [
41,
42,
43]. A recent report describing the application of a LC-MS/MS method for cyanotoxins in natural waters across Europe including France, Italy, Ireland, Germany, described microcystin concentrations <3 μg/L [
44]. Some of the water samples from this study contained very high concentrations of microcystins, with a maximum above 40 mg/L. Very high concentrations of microcystins per litre of water have been reported up to 25 mg/L microcystin [
34,
45], whilst noting that these would be from scums or from very dense accumulations of cyanobacteria. The formation of highly toxic scum with the 1000-fold accumulation in cyanobacterial bloom and toxin concentrations have been well described, notably resulting in an increase in cyanotoxin risk [
34,
46]. Sivonen, K reported that whilst toxin concentrations vary hugely in water, in bloom conditions milligram amounts of microcystins have been reported. The water samples reported in this study, some of which included large amounts of cyanobacterial scum, were sampled reactively, and in many cases represent a worse-case scenario in terms of total toxin loading. As such, the high concentrations of toxins reported here for some samples must be interpreted with care. During this study, dry weight values of cyanobacterial biomass were not determined, so it is difficult to compare our values directly against those determined elsewhere. In relation to potential health effects, 29% and 13% of water samples were found to exceed the low and medium probability health alert thresholds for recreational water (2.0 and 20 μg/L), respectively. The results therefore demonstrate a significant risk to humans and animals accessing recreational waters, which could potentially increase in the future with predicted changes to climatic change, in particular, temperature and rainfall patterns [
2,
8,
13,
20].
The data obtained from this study have shown the potential for toxic cyanobacterial blooms to be formed throughout at least nine months of the year, with significant concentrations of microcystins being present as early as March and as late as November. The data show, however, the highest prevalence of bloom formation and toxin production occurred between the narrower window of July to October, representing one third of the year-long study. As such, the period of highest toxicity risk occurred in 2016 during the mid to late summer, extending into early autumn, although the potential for earlier bloom formation should not be discounted. This would fit with the well-established notion that cyanobacterial growth rates and thus bloom occurrence is more widespread during periods when light intensities and water temperatures are higher [
1,
2,
10,
11,
16,
34] and agrees with the findings of [
3] who noted the common occurrence of cyanobacterial blooms in late summer during a five year period between 2000 and 20005 in UK water reservoirs. Other previous studies have concluded that the summer months are dominated by green algae, with cyanobacteria succeeding in late summer, autumn and early winter [
47]. During August and Sept 2016, the weather was changeable throughout the country, but hot and humid weather was present intermittently, with the UK mean temperature being 2.0 °C above the 1981–2010 long-term average, making it the equal second warmest September on record since 1910 [
48]. In addition, there were periods of sustained rain, including heavy thunderstorms bringing localized flooding in some areas of the country, which are likely to have resulted in increased nutrient loading to some water bodies around this time. The higher number of toxic blooms obtained during these two months may therefore relate to both the unseasonably high temperatures as well as the increased levels of rainfall. Whilst microcystin production is generally thought to be fairly constant for any given strain of cyanobacteria [
49,
50], there are reports of MC variant proportions changing in response to temperature modifications [
16]. Consequently, the role of environmental factors in both bloom abundance and toxin production is not fully elucidated [
51]. Temperature may well influence not just bloom dynamics, but also influence the preferred production of the toxic fraction of any given cyanobacterial population [
1,
52,
53,
54]. Whilst there may be evidence for such a relationship for
Microcystis sp., there are contradictory data showing an inverse relationship between temperature and toxin production in
Planktothrix [
55]. However, any significant increase in water column temperature and nutrient concentrations could explain the formation of high-density blooms and consequently high toxin concentrations. Further screening work will be required in future years to establish inter-annual variations in bloom formation and toxin production, together with an assessment of environmental inputs before any formalised risk assessment can be performed.
LC-MS/MS data showed the occurrence of high toxicity samples across the entire country, with no indications of any geographical patterns which may link to temperature or any other meteorological parameters. Interestingly, sample 114, which contained extraordinarily high concentrations of microcystins, was taken from a relatively small artificial lake, approximately 100 m × 40 m in size, during September in NW England, an area of the country not traditionally associated with high air temperatures. However, during September, the NW of England experienced maximum and minimum temperatures 2.2 and 2.6 °C higher than the 1961–1990 average, together with higher than expected rainfall and sunshine hours [
48]. The relatively small water body would therefore have been subjected to prolonged above-average warming and freshwater/nutrient input, all of which may have potentially heightened the bloom intensity and associated toxin production. However, results indicated there was no statistical correlation between air temperature, rainfall or lake size and total microcystin concentrations, expressed either in terms of total toxins per litre of water or total toxins per bacterial cell. Overall, it is likely that highly specific localized factors such as nutrient concentrations and wind conditions will have influence on the growth and toxicity of cyanobacterial blooms. Further studies would be required in GB to explore these environmental parameters in greater detail.
It is recognised that cell densities enumerated by microscopy and toxin concentrations do not necessarily correlate due to the considerable variation in cell toxin content between cyanobacterial strains [
56,
57]. As such, the determination of cyanobacterial cell densities is not the best indicator of actual toxin exposure [
58]. Over half the samples taken during this study contained some form of visible algal scum, with 72% in total exceeding the cell density threshold. Whilst the inclusion of significant amount of scum is likely to have resulted in the very high microcystin concentrations reported in some water samples, out of these 99 samples only 18% were found to contain total microcystins above the WHO medium health alert guidance limit of 20 μg/L. As a consequence, sole reliance on microscopy identification and enumeration is likely to result in the closure or access restriction to more water bodies than necessary, potentially resulting in impacts on industry, event organisation and public recreation. Conversely, out of the 18 bloom samples containing total microcystins above the 20 μg/L guidance limit, 17 contained a visible algal scum with the other exceeding the cell density threshold. This therefore shows good evidence that the current monitoring regime provides effective protection for humans against recreational water exposure during periods of freshwater algal blooms. Together, these results indicate that whilst a monitoring programme involving microscopy alone appears suitable for protection of human health, the incorporation of toxicity screening using chemical detection methods will provide a more realistic assessment of the risks associated with recreational water bodies, reducing potential future socio-economic impacts relating to water body closures.
The most commonly identified cyanobacterial genera in the 117 samples studied were firstly
Microcystis, followed by
Anabaena,
Aphanizomenon,
Oscillatoria and
Gomphospaeria, with
Planktothrix and
Gloeotrichia also identified in a low number of samples. These findings therefore agree well with those reported from cyanobacterial blooms in the UK during the 1980s, where
Microcystis, Anabaena Aphanizomenon and
Oscillatoria were also the top four most commonly identified taxa [
11,
13]. Whilst approximately 40% of the samples were found to contain multiple species of cyanobacteria, the remainder of samples containing one genera enabled the first determination of toxin concentrations profiles in a range of natural cyanobacterial species occurring in UK freshwaters. Whilst the presence of significant levels of microcystins is sometimes linked primarily to the presence of
Microcystis spp., the data reported here show concentrations of toxins above the WHO low and medium probability thresholds in single genera bloom samples containing
Microcystis,
Gomphospaeria,
Anabaena,
Aphanizomenon and
Oscillatoria. Consequently, there is the potential for toxin production in a wide range of cyanobacteria found to inhabit UK freshwater ecosystems. Out of the 18 water samples with total MC concentrations above the 20 μg/L threshold, only six contained a single cyanobacterial genus, indicating the greatest risks to be present in bloom samples containing mixed taxa.
To date there have been very few descriptions of microcystin toxin profiles present in natural water samples from the UK. Previously, it has been reported that usually only one or two microcystin variants are dominant in any single strain [
45]. Lawton, L.A. et al. reported the presence of a number of different MC variants in two samples from England, taken from central and NE England during 1989 and 1992 respectively, with the former implicated in animal deaths at the time of sampling [
59].
M. aeruginosa was present in both samples and through a combination of LC-DAD and LC-MS were found to contain MC-LR as the dominant analogue, with lower relative proportions of MC-LY and the more hydrophobic MC-LW and MC-LF in addition to a demethylated analogue of MC-LR. In this study, all of these MC variants were identified with high average proportions of MC-RR and MC-YR being quantified across all samples. In addition, MC-LA, MC-WR, D-Asp3-MC-LR and MC-HtyR were quantified, together with occasional low levels of MC-HilR and D-Asp3-MC-RR. These results therefore increase the number of variants reported to date in natural UK cyanobacterial samples, potentially related to the increase in sensitivity of the current LC-MS/MS instrumentation and the larger number of toxins available as reference material standards.
LC-MS/MS results showed a wide variety in the relative proportions of microcystin analogues from sample to sample. Notably, toxin profiles differed greatly between cyanobacterial genera, in particular with
Aphanizomenon, and to a lesser extent
Oscillatoria, containing simpler profiles in comparison to
Microcystis and
Anabaena. Interestingly, both these genera were found to contain the highest mean proportion of the more hydrophobic MC-LF. Further afield in Europe, Pekar et al., 2016 quantified a large number of MC variants from natural lake waters containing
Microcystis spp. in Sweden, most commonly MC-LR, MC-RR, MC-YR, MC-D-Asp3-LR and D-Asp3-RR, all of which were found in
Microcystis samples from this study. Similarly, LC-LR, MC-RR and MC-YR were the three most abundant toxins present in two lakes from Greece, although other variants including MC-HilR, MC-WR, MC-LY and various demethyl analogues were also identified [
60]. Halinene, K. et al. [
61] also reported the LC-MS identification of six MC variants in a number of strains of
Anabaena sampled from the Baltic Sea. These included MC-LR, as well as MC-HtyR and four demethylated variants. Overall, however, in this study there seemed to be little correlation between cyanobacterial genus and specific toxin profiles, and with profiles reported previously from other geographical areas. Identified profile types were not found to correlate with either water body size or air temperatures. However, the level of rainfall measured at meteorological stations close to each sampling site were found to vary significantly between the three profile types. Specifically, profile 3 samples were found to be associated with areas with significantly lower rainfall, with higher levels of rainfall determined in sites dominated by profile cluster 2. As such, the data suggests that microcystin profiles dominated by MC-RR are associated with water samples occurring during periods of lower rainfall, with high rainfall relating to toxin profiles containing a wide variety of MC analogues, including MC-LR, RR, LA, LF, LW, WR, D-Asp3-LR and HtyR. An overall lack of any clear factors affecting the profile type fits with the published notion that toxin profiles produced by cyanobacteria are highly strain dependent [
62], with toxin production regulated at the genetic, cellular and population levels [
16]. However, the apparent enhanced production of MC-RR in water bodies associated with lower levels of rainfall in this study, and increased proportions of other analogues during high rainfall, is an observation that needs to be assessed systematically in future studies.