Monitoring is critical element in cyanotoxin risk management. The goals of a monitoring program to support risk management are three-fold: to measure cyanobacteria concentrations in source and final drinking water, to measure the concentrations of cyanotoxins in source and final drinking water and to measure source water constituents and conditions that promote or inhibit cyanobacterial growth. Accurate and precise data in these three areas, collected on a regular basis and carefully tracked over time, will help water supply managers to achieve the greatest reduction of risk.

The design of an effective long term monitoring program requires that water supply managers ask, and answer, the following questions:
(1) What analytes do I sample for and how do I measure them?
(2) Where do I sample for these analytes?
(3) How often do I sample for these analytes?
(4) How much replication do I build into a sampling event?

Monitoring can be defined as including two components, sampling of the water body and analysis of the samples. Together they provide the information for early warning and tracking the development of cyanobacterial blooms [1]. An overview of recommendations for design of a monitoring and sampling program for cyanobacteria is given later in this section (see Table 3-2).

When choosing an organisation to sample and/or analyse cyanobacterial samples it is recommended that the testing laboratory selected is accredited to carry out these particular analyses by a national laboratory accreditation authority. For example in Australia the National Association of Testing Authorities (NATA) accredits and recognises facilities that are competent in specific types of testing, measurement, inspection and calibration. Not all accredited laboratories use the same methods for testing and this is not important provided the individual methods are accredited. It may however, make it difficult to compare results when samples are analysed by more than one laboratory. Where an accredited laboratory is not available it is important to ensure the analyses are undertaken according to the highest standards, and inter-laboratory testing has shown the validity of testing procedures.

One of the simplest and most important forms of monitoring of a water body is regular visual inspection for water discolouration or surface scums of cyanobacteria. This can be a secondary form of surveillance for higher classes of monitoring, or if few other resources are available, the principal form of surveillance used for remote sites or non-specialised field personnel. However some cyanobacteria, for example Cylindrospermopsis, do not form scums and a slight green discolouration of the water may be indicative of dangerously high cell numbers. In situations where non-bloom forming cyanobacteria are present it is essential that samples are collected for analysis to determine the abundance of cyanobacteria in the water body.

When bloom-forming cyanobacteria are present, a qualitative assessment through visual inspection can be a useful indicator of water quality and the relative hazard posed by the presence of cyanobacteria. The frequency of visual inspections may vary depending on seasonal and weather conditions. If visual inspection is the only monitoring being carried out, the position and extent of scum formation should be recorded on a dedicated report sheet.

The first visual indication of cyanobacteria may be the presence of small green particles in the water that may be more obvious by holding a jar of the water up to the light. Scum formation will not normally be observed until open water concentrations of cyanobacteria exceed 5,000-10,000 cells/mL, but exceptions are possible. Blooms or scums are usually most apparent early in the morning following calm days or nights, but as cell concentrations increase, or during prolonged periods of calm weather, scums may persist at the surface for days or weeks. Scum accumulations will normally be observed at the downwind end of a reservoir, lake or river reach and also in sheltered back waters, embayments and river bends.

In general, a healthy cyanobacterial scum will appear like bright green or olive green paint on the surface of the water. Scums only look blue in colour when some or all of the cells are dying. As the cells die, they release their contents, including all their pigments, into the surrounding water. Cyanobacteria have three main pigment types: chlorophyll, phycobiliproteins, and carotenoids. In healthy cells, the green chlorophyll colour normally masks the other pigments, although these other pigments may give blooms a more yellow-green or olive-green colour in some cases. When the cells die, the chlorophyll is rapidly bleached by sunlight, while the blue phycobiliprotein pigment (called phycocyanin) persists. Figure 3-1 shows some examples of cyanobacteria in concentrations that will cause a water quality problem for water suppliers.

Cyanobacterial scums should not be confused with scums or mats of filamentous green algae, which appear like hair or spider web material when a gloved hand is passed through the water. There are blooms of other phytoplankton that look very similar to cyanobacterial scums, but these cannot be readily distinguished without a microscope. Scums or mats of filamentous green algae are more common in slow flowing, shallow streams and irrigation channels and drains.

Figure 3-2 shows some examples of green algae similar in appearance to cyanobacteria. The major point of visual differentiation is the bright green colouring of the green algae, compared with a more olive- or blue- green for cyanobacteria.

Benthic cyanobacteria are usually submerged, and are difficult to monitor. Visual inspection is a very important way to identify an issue with benthic cyanobacteria as they will often break free of the surfaces to which they are attached, and float to the surface. Figure 3-3 shows some examples of attached benthic cyanobacteria and detached floating mats that may cause water quality issues.

Another tell-tale sign of cyanobacterial blooms is their odour. Some cyanobacteria produce a distinctive earthy/musty odour that can often be smelt at some distance before the bloom/scum can be seen. Therefore it is useful to conduct ‘odour surveillance’ in conjunction with any visual inspection program.

For an example of a recording sheet for a visual inspection, click here

The development of an appropriate sampling strategy will depend upon the primary objective of the monitoring program. The objective will be determined by the immediate use of the water, which in turn determines the level of confidence required in the monitoring results. For example if the water is being used directly to supply consumers, i.e. is in service, then you will want a very high degree of confidence in the monitoring result for any potential hazards from the occurrence of cyanobacteria. However if the reservoir is not directly in service or is a bulk water storage, then you may have less need for a high degree of confidence in the results. This objective-based approach can be used to design a program based upon the level of sampling effort which translates to resource needs and cost for the program.

For most purposes, the aim should be to obtain samples that are representative of the water body as a whole, or the part of a water body that is in use (e.g. near the water treatment plant offtake). Once the aim of the monitoring program is established the required level of sampling effort described as high, moderate or low, is determined by combinations of the following components:

These components, which are given in Table 3-2 are discussed in more detail below.

Cyanobacteria tend to be extremely patchy in distribution, both vertically and horizontally within the water body. Vertical patchiness results from the development of a stratified water column in warm calm weather, allowing buoyant cyanobacteria to maintain their position at the surface for extended periods. Horizontal patchiness is common for most phytoplankton, but can be particularly pronounced in cyanobacteria due to the effect of prevailing winds, which cause accumulation downwind along shorelines of reservoirs or bends in river reaches.

Depth integrated sampling in open water provides, in general, a better representation of the ‘true’ or average cyanobacterial population in a water body and is therefore the preferred option. Open water and mid-stream sampling is normally undertaken from a boat, but can also be achieved in some circumstances from a bridge over a river, or from an open water structure such as a reservoir offtake platform. For drinking water supplies, sampling the appropriate depth next to, or from, the water offtake tower is recommended. Due to the resources required for open water sampling (i.e. boat and two people), it is often reserved for high priority public health surveillance.

If open water sampling is not possible, the second option for monitoring drinking water supplies is to sample from reservoir/lake shorelines or riverbanks. Such samples may not be representative of the ‘true’ cyanobacterial population due to the bias in spatial distribution discussed above and the limited choice of suitable locations. In choosing a location for sampling the likely effects of the prevailing winds and water currents should be taken into account.

Benthic cyanobacteria are also known to cause problems associated with water quality so sampling of the sediments and attached growth, and therefore a different approach to sampling, may be required.

The methods used for sample collection will depend on whether the sites require access by boat, shore or platform and will include integrated water column (hosepipe) sampling, discrete depth (grab) sampling, grab sampling from an extension pole, sediment sampling by grab or corer for benthic cyanobacteria and sampling from a pipeline. Different methods are used to collect samples for cyanobacterial identification, for toxin analysis or for assessing benthic cyanobacteria. In addition different techniques may be used to collect these samples from a boat, from depth, from the shoreline or a pipeline.

It is important to be aware of the safety issues involved in sampling for cyanobacteria, whether from the shore or a boat. Samplers should be fully trained and aware of all aspects of sampling including:

Once training has occurred, hazards or risks involved with field sampling must be identified and documented on a site- and sampling- specific basis.

In some instances it may be necessary to collect benthic samples for identification of cyanobacteria, particularly if high levels of taste and odour compounds are detected but few, or no, cyanobacteria are present in water samples. In most cases benthic samples are not collected routinely and are generally for qualitative analysis only. The most convenient way to sample benthic cyanobacteria is from any mats that have become detached from the substrate and are floating on the surface. In the absence of floating mats a representative assessment of numbers and distribution of benthic cyanobacteria is difficult. Samples should be collected from a number of transects throughout or around the perimeter of a reservoir. Particular attention should be paid to shallow protected bays and any areas where benthic mats have been observed in the past. Samples at varying depths may be required down to approximately 5 metres, although this will depend upon light attenuation in the water body. Samples can be collected using a benthic sampler such as an ‘Eckman’ grab or a rigid plastic corer (e.g. PVC or polycarbonate pipe). A transect in a shallow, protected bay should be chosen to sample. Duplicate samples of sediment at varying depths are collected either by grab or hosepipe and emptied into a container with a fitted lid. If large quantities of sediment are collected, a subsample can be taken and stored in a smaller specimen jar. Visual observations of the sediment surface can also provide very useful information on the distribution of benthic cyanobacteria. More detailed surveys can be conducted using underwater cameras or divers. This requires access to relatively sophisticated expertise and resources.

Benthic cyanobacteria may also be found attached to dam walls or offtake structures. Cyanobacteria attached to these structures can be scraped off, most easily when water levels drop.

The preferred method for sampling a reservoir or river is by boat, which should always be stationary while sampling proceeds. The sampling stations, or locations, in a reservoir should preferably be chosen randomly within several defined sectors, representing the entire water body. For boat sampling the use of permanent moorings with marker buoys placed in each of the sectors is the most practical approach and makes open water sampling easier, especially in windy weather. Having permanent sampling sites also gives consistency which enables the comparison of results at each site over a given time frame. If it is not possible to place permanent marker buoys in a water body, a global positioning system (GPS) should be used to ensure the consistency of sampling points over time. One way to introduce randomness when boat sampling is to move sampling station moorings within sectors on a yearly basis. For monitoring rivers, randomness of sampling sites is less critical due to instream flow.

Go to Level 2 for more information on open water sampling
Integrated water column samples
Discrete depth samples

Sampling from a bank or shoreline is comparatively simple, but introduces a risk of excessive bias of samples from patchy shoreline accumulations. A ‘pole-type’ sampler can be used, where the bottle is placed in a cradle at the end of an extendable pole of 1.5-2 metres length. This procedure is depicted in Figure 3-4. Alternatively, a spear sampler as described in [2] is a useful sampling device for collecting an integrated depth water sample when standing on the bank or shoreline. It is also important to note that in using either the pole or spear sampler, scum accumulations near to the shoreline will not be sampled. A separate dip sample of any accumulations may be needed for toxin analysis.

Qualitative toxin analysis is done by mouse bioassay and is usually carried out either when more sophisticated techniques are unavailable, or the identity of the toxin is initially unknown. These samples are generally collected from dense accumulations of scum along shorelines and riverbanks if these are present. Alternatively, cells may be concentrated by either trailing a phytoplankton net (25-50μm nylon mesh) from a boat or from the shoreline, or by collecting a large volume of water that can be concentrated in the laboratory. Figure 3-5 shows sampling from a shoreline with a net-tow sampler to concentrate the cyanobacteria.

The volume of sample required depends upon the concentration of scum or cyanobacteria collected. Up to 2 litres of sample may be required if cyanobacterial concentrations are low, or if species present are small enough to pass through a phytoplankton net and samples therefore need concentration by other means such as filtration or centrifugation.

This test should be used as a screening tool only. If a mouse bioassay proves positive, quantitative methods are then required to determine the type of toxin, and concentrations present.

Quantitative toxin analysis is performed using a variety of methods suited to the type of sample and toxin present (see following sections). Samples are collected in the same manner as those taken for phytoplankton identification and enumeration and the volume of sample required is dependent upon the type of analysis to be used. In general, at least 500 mL of water should be collected.

For monitoring trends in cyanobacterial abundance, an indication is required of the ‘true’ cyanobacterial population, representing the entire water body. This can be achieved by collecting a suite of discrete samples from different sampling sites, which are counted separately and then may be averaged. As an alternative to undertaking separate counts on samples collected at several sites, samples may be pooled or composited. These samples are collected at three or more individual sites and pooled into one container. The sub-sample for counting is then taken from the container after its contents have been thoroughly mixed. If composite samples are made, the individual samples must be of equal volume to prevent bias. An alternative to pooling samples in the field is to send discrete samples to a laboratory, where they can be sub-sampled, pooled and analysed. Using this process, a portion of the original discrete sample can be retained for further analyses if required. The trade off from compositing is a decrease in statistical power for subsequent data analysis against a three-fold or greater reduction in counting costs.

The number of sampling sites in a water body is chosen to determine the spatial variability of the cyanobacterial population and will also be influenced by time and cost considerations. It is recommended that a minimum of three sites be used when cyanobacterial counts exceed 2,000 cells mL^-1 for both open water sampling and shoreline sampling, or sampling should be undertaken according to the appropriate cyanobacteria incident management plan (see Chapter 6). For lakes and reservoirs the sampling stations should be at least 100 m apart (where possible), while for rivers replicate samples should represent different ‘parcels’ of water. When sampling from a boat, replicate samples should preferably be taken at the downstream end first to avoid re-sampling the same ‘parcel’ of water.

The appropriate frequency of sampling will be dictated by a number of factors including the category of use, the current alert level status (see Chapter 6), the cost of monitoring, the season and the growth rate of the cyanobacteria. Apart from cost, the underlying consideration in operations monitoring is the possible health consequences of missing an early diagnosis of a problem. Cyanobacterial growth rates are generally related to seasonal conditions and previous studies have shown that cyanobacteria in the field can exhibit growth rates from 0.1-0.4 d-1 (equivalent to population doubling times of nearly a week to less than two days respectively). These estimated growth rates can be used to construct a set of theoretical ‘growth curves’ for a population of cyanobacteria starting from an initial count of either 100 or 1,000 cells mL^-1 (Table 3-1). Historical data should be used as an indicator of likely rates of increase in cyanobacterial numbers.

Initial Concentration (Cells/mL)	Growth Rate -Population doubling time (days)	Cyanobacteria Concentration
		at 3 days	at 7 days	at 14 days	at 28 days
100	6.93 (μ=0.1) - slow		200	400	1500
100	1.73 (μ=0.4) - fast		800	6400
1000	6.93 - slow		2000	4000	>15000
1000	1.73 - fast	3500	16000	>250000

Based on this assessment, it is recommended that sampling for high risk/high security supplies (i.e. drinking supplies) should occur on at least a weekly basis and probably twice-weekly when cyanobacterial count of > 2,000 cells mL^-1 is reached. It is important to understand that frequency of sampling is determined by the need to detect real changes in population numbers and significant upward trends in growth, data collected will inform changes to treatment plant operations, and the application of cyanobacteria management plans, discussed in Chapter 6.

For supplies where the public health risk is deemed to be low (i.e. low cell counts in non-supply reservoirs), fortnightly sampling may be adequate, but caution is advised given the rate at which the cyanobacterial population may increase.

The timing of sampling for buoyant cyanobacteria can be important during calm, stratified periods especially if depth integrated samples are not collected. Buoyant cyanobacteria tend to accumulate near or at the water surface overnight, which can result in an over-estimation of cell concentration in surface samples collected early in the morning or an under-estimate in those collected at depth at the same time. Temporary surface scums may be observed early in the morning, but they tend to disperse as winds increase and may even be mixed back into the water column during the day. Thus, a sample that is less biased by scum formation is, on average, more likely to be obtained later in the day. If the option exists, it is preferable to delay sampling to later in the day, but whatever time is chosen it is best to adhere to the same sampling times for each location on each sampling occasion if possible.

At some point, analytical results from a monitoring program may be compared with a fixed standard, set internally by a drinking water provider, or externally by a regulatory agency. Because crossing a regulatory threshold often involves significant consequences, it is critical that water providers understand the degree of statistical uncertainty that is associated with an analytical result. Collecting single samples has the lowest short term cost. However it is impossible to characterize the uncertainty associated with a given sampling event. Moving to duplicate sampling allows characterization of the uncertainty. Triplicate sampling in turn permits a more precise estimate of the confidence interval surrounding the “true” value of the analyte of interest. As a result, it is recommended that, budgets permitting, some degree of replication be practiced in the sampling of critical analytes. A popular compromise is to collect replicate samples at some fraction, such as 30%, of all sampling events. With careful record keeping, it will be possible to develop a feeling for the statistical uncertainty associated with the sampling and analysis of a given analyte.

Purpose of monitoring	Confidence required from results	Water body type	Sampling effort required	Access required for sampling	Sample type (method)1	Number of samples2	Frequency of sampling3
Public health surveillance of drinking supplies: in direct service	Very High	Reservoirs & lakes	High	Supply offtake and Open water by boat	Discrete sample at offtake depth and Integrated depth	Both offtake location and multiple open water sites	Weekly or 2x-weekly
		Rivers and weir pools		Mid-stream by boat; from bridge or weir	Integrated depth
Public health surveillance of drinking supplies: bulk water storage / not in service	High	Reservoirs & lakes	Moderate	Supply offtake location and/or Open water by boat	Discrete sample at offtake depth and/or integrated depth	Multiple sites	Weekly or 2x-weekly
		Rivers and weir pools		Mid-stream by boat; from bridge or weir	Integrated depth
Public health surveillance of recreational water bodies & non-potable domestic supplies	Moderate	Reservoirs & lakes	Low	Shoreline	Surface Sample	Limited number of sites	Weekly or fortnightly
		Rivers and weir pools		River bank	Surface Sample

Integrated water column samples are also called ‘hosepipe’ samples and are recommended for open water sampling, where a representative sample of the water column over depth is desirable. The samples should be collected using a flexible hose pipe or rigid plastic pipe (Figure 3‑1 (L2)). A rigid pipe can be fitted with a one way valve, which simplifies the operation of withdrawing the pipe and sample from the water. The depth that the sample pipe is dipped should reflect the approximate depth to which cyanobacterial cells are likely to be mixed. When the stratification status is uncertain, a temperature probe, if available, may be used to determine the depth of any thermocline present. If this equipment is not available, a 5 metre long flexible pipe is recommended, but a 2 metre long pipe may be more appropriate in shallower water bodies (those that are less than 3 metres deep). The inner diameter of the pipe should be at least 2.5 cm and flexible pipes are probably more practical than rigid pipes for pipe lengths greater than two metres. The recommended method of obtaining a ‘hosepipe’ sample is shown in Figure 3‑1 (L2).

Water sampling for public health surveillance is often required at the raw water abstraction point for reticulation to a drinking water treatment plant. For this purpose discrete depth samples or ‘grab’ samples are often collected with a sampling bottle apparatus (e.g. ‘Van Dorn’ or ‘Niskin’ samplers), that can be triggered to be filled at a specific depth below the surface corresponding to the offtake depth (Figure 3‑2(L2)). The rationale for this is to determine the total load of cyanobacteria (and their toxins) to the water treatment plant. In addition, the degree of cell lysis and toxin release through the reticulation system can be measured from an accurate assessment of intact cells at the offtake point. This is important information for determining the appropriate strategy for cell and toxin removal in the treatment plant. When choosing a sampling site near the water abstraction point in a reservoir the size of the offtake and the abstraction pumping rate should be considered. If pumping rates are high, vortices may occur around the offtake or abstraction valves which indicate that surface water is being drawn down into the offtake. If this situation is present in the reservoir, a number of samples at depths ranging from the surface to the offtake depth should be taken to determine the total load of cyanobacteria cells and toxins entering the water treatment plant. The method for collecting a water sample at depth is depicted in

Figure 3-2(L2) Using a depth sampler or closing bottle to collect a grab sample at a discrete depth

Duplicate samples are collected from the effluent of a drinking water treatment plant. They are analysed by LC-MS for microcystin-LR (MC-LR), with the following results:[1]

Sample 1: 1.12 µg L^-1
Sample 2: 1.27 µg L^-1

The estimated sample mean m is:

The observed mean of 1.20 µg L^-1 is an estimate of the true MC-LR concentration in the effluent. At first inspection, the effluent MC-LR concentration appears to exceed the World Health Organization provisional MC-LR standard of 1.0 µg L^-1. However, given the observed variability in these two observations, how confident can water supply managers be about their estimate? In order to quantify the level of uncertainty, the following information is needed:

The sample standard deviation, F_n-1 = 0.106 µg L^-1

The number of observations in the sample, n = 2

The degrees of freedom, d.f. = n – 1 = 1

Student’s t statistic for 95% confidence, d.f. = 1, t_{(1 – α = 0.95, d.f. = 1)} = 6.31

These values are used to calculate a one-sided 95% confidence interval and establish a lower confidence level (LCL). A one-sided confidence interval was chosen because the primary question is whether or not the true MC-LR concentration exceeds a regulatory threshold.

The calculated LCL of 0.73 is less than the 1.0 µg/L WHO provisional standard. Based on this data, a decision is made to resample the treatment plant effluent in triplicate, with the following results:

Sample 1:              1.11 µg L^-1

Sample 2:              1.21 µg L^-1

Sample 3:              1.27 µg L^-1

From this raw data, are calculated the following:

m = 1.20 µgL^-1

F_n-1 = 0.0808 µg L^-1

n = 3

d.f. = n – 1 = 2

t_{(1 – α = 0.95, d.f. = 2)} = 2.92

The second round of sampling, with the same mean as the original sampling event, yielded an LCL of 1.06 µg L^-1 at the 95% level of confidence, which is greater than the 1.0 µg L^-1 WHO standard. The standard deviation decreased by 24% versus the first round of sampling. More importantly, increasing the number of samples from 2 to 3 increased the degrees of freedom from 1 to 2. Increasing the degrees of freedom by 1 caused the critical t statistic to drop by more than one half. The combination of increased sample size and slightly lower standard deviation led to the calculation of a smaller one-sided confidence interval.

Myponga Reservoir is a moderate-sized drinking water reservoir that has regular growth of the nuisance cyanobacterium Anabaena circinalis each summer. The reservoir is used directly for drinking water supply after water treatment with conventional treatment plant incorporating dissolved air flotation (DAF), and the capacity to dose with powdered activated carbon (PAC) for taste, odour and toxin control.

Myponga Reservoir (S 35° 21' 14", E 138° 25' 49") is located 70 km south of Adelaide in South Australia. The reservoir has a capacity of 26,800 ML at a full supply level of 211.7 m AHD (Australian Height Datum), an average depth of 15 m, a maximum depth of 36 m and a surface area of 2.8 km². The mean retention time based on abstraction is approximately 3 years. Water is removed from the reservoir via an offtake valve located on the dam wall at 195.2 m AHD.

Samples are collected weekly in winter and twice-weekly in the summer growth season for identification and counting of phytoplankton from up to 10 separate locations. Sampling is concentrated at the offtake site where 4 separate samples are collected: a 0-5m integrated surface sample (Location 1221) and three discrete depth samples at 10, 20 & 30m (Locations 1222, 1223 & 1230). Spatial variability is assessed by collecting integrated column samples (0-5m) at 6 locations (Locations 1224-1229) spaced throughout the reservoir. The winter sampling frequency is weekly for 6 months from April - September which then increases to twice-weekly from October - March inclusive. The sampling program in winter incorporates a process of collecting and ‘pooling’ samples from the 6 reservoir locations which are then processed for a single cell count. If cyanobacteria are recorded in this pooled sample above a certain threshold (200 cells mL^-1) the individual sites will be re-assessed individually. Note that this pooling is only used in winter and all locations are sampled and counted individually in summer.

The precision (counting error) can be calculated from the total number of units counted (n) using the simple formula derived by Laslett et al. [13]:

Counting error (± %)  =  100Ö (2/n)                                                  (1)

Equation 1 accounts for the variability of cells in a counting unit and assumes that the number of cells in a colony or trichome is always counted. Although it would be unusual for an analyst to count all cells in all trichomes or colonies as this equation assumes, it is still recommended as the benchmark method for enumeration of filamentous cyanobacteria. This is due to the fact that it takes into account the sometimes large variability in trichome/colony size when calculating counting error.

Higher precision will require a higher analytical effort and generally a higher cost. The relationship between counting error and counting effort is shown in Figure 3‑4(L2). There is a very strong ‘law of limited returns’ applying for increased effort beyond about 50 colonies or trichomes counted.

At the completion of a count, the counting precision should be calculated using Equation 1. This figure could be reported in the following format for each taxon and/or for total cyanobacterial abundance:

                                XXXX cells mL^-1    (minimum counting error  =  ±  YY %)

Due to the fact that some of the counting errors may be very large, it is important to accompany any reporting of errors with some clarification and interpretation of these errors.  

Enzyme linked immune substrate assays (ELISA) can be used for the analyses of several cyanobacterial toxins. The method is based on the coating of well plates or test tubes with toxin antibodies. The antibodies are molecules with a shape that specifically matches the structure of a toxin molecule. This specificity allows the antibodies to bind to, and immobilize, toxin molecules in solution. The number of antibody molecules that can be coated onto a given surface area is controllable and repeatable. Toxin molecules in solution will bind to coated antibodies when an unknown sample is added to a well or tube. Following the sample addition step, a solution containing a known concentration of enzyme-conjugated toxin molecules is added to the well or tube. Toxin molecules in this solution will occupy any binding sites left unoccupied after addition of the unknown sample. The enzyme in turn catalyzes a reaction that yields a color change that is inversely proportional to the concentration of toxin in the unknown sample.

ELISA assays for microcystins, nodularins, saxitoxins and cylindrospermopsins are commercially available as kits containing all of the necessary reagents. The advantage of these assays is that they are relatively inexpensive, simple and rapid to run, and the samples do not require pre-concentration. The total turnaround time is less than half a day. The assays can be performed in any laboratory equipped with multi-well pipetters and a spectrophotometer. The major disadvantage of the assays is that they cannot distinguish between toxin variants. This can complicate risk management decisions because of inter-variant toxicity differences. As a result, ELISA assays are ideal screening tools. They can be incorporated into a suite of routine analyses used to pinpoint the initial stages of a bloom event, and to determine when it is necessary to begin more expensive and time-consuming analyses capable of resolving toxin variants.

The mode of toxicity of microcystins involves the inhibition of certain enzymes, the protein (serine/threonine) phosphatases, responsible for the dephosphorylation of intracellular phosphoproteins. The phosphatase PP2A is the most susceptible to inhibition by microcystin toxins. The basis of the PP2A inhibition assay is the measurement of phosphate release from a suitable substrate in the presence of a phosphatase enzyme preparation and an inhibitor such as microcystin. The most commonly used PP2A assay utilises p-nitrophenyl phosphate (pNPP) as substrate. In the presence of PP2A p-nitrophenol is released from the pNPP and can be measured photometrically. In the presence of an inhibitor such as microcystin, the release of the p-nitrophenol is reduced, and the difference between the sample and the control (in the absence of an inhibitor) can be calibrated to microcystin-LR concentration [[22]]. The assay does not discriminate between the different microcystin variants, and as mLR shows the greatest inhibition effect the result is usually described in terms of mLR toxicity equivalent concentration [[23]].

Liquid chromatography (LC) is the technique used to separate mixtures of dissolved compounds (including toxins) prior to some form of instrumental detection. The sample for analysis is injected into a column packed with solid media. Different components of the mixture pass through the column at different rates due to differences in their partitioning behaviour between the liquid and solid phases. Using appropriate combinations of media, mobile solvent, temperature, and flow rate, unknown compounds will exit the column as discrete slugs whose presence in the effluent stream can be detected with ultraviolet (LC-UV), photodiode array, where a UV scan is taken for each peak, rather than an absorbance measurement at one wavelength only (LC-PDA), fluorescence (LC-FD) and mass spectrometric (LC-MS) detectors. When UV and FD response is plotted versus time, the separated compounds will show up as discrete peaks. Identification and quantification of the unknown is accomplished by comparing the timing and size (or scan) of an unknown peak exiting the column with the timing and size of peaks from calibration standards.

Microcystins and cylindrospermopsins can be analyzed by LC-UV. Ultraviolet detectors are more economical than MS and require less skill to maintain and operate. However, UV detectors are not capable of distinguishing among co-eluting toxin variants, or among toxins co-eluting with background organic material.

Saxitoxins can be analysed by LC-FD. Fluorescence detectors are less expensive than MS detectors and have the advantage of greater sensitivity. The primary disadvantage of fluorescence-based detection methods for cyanobacterial toxins is that they require additional reagents added to the LC column effluent. These compounds react with eluted saxitoxins to form a fluorescent end product. This post-column derivatisation procedure adds an additional level of complexity and cost to the analysis.

Microcystins, cylindrospermopsins, saxitoxins and anatoxins can all be analysed by LC-MS. In this technique, a portion of the flow exiting the chromatography column is routed through an MS detector, which generates mass spectra. A mass spectrum shows the relative distribution of components in a sample by their mass to charge ratios. Because cyanotoxin molecular weights are known with great precision, MS detectors allow the analyst to resolve co-eluting toxin variants with small molecular weight differences. MS detectors are also very sensitive, allowing analysts to achieve lower detection limits. The disadvantage of LC-MS systems is that they are complex, expensive, and the interpretation of their results requires a high level of experience.

Sample concentration and clean-up is a critical step in toxin analysis by instrumental techniques. The toxin of interest is often present at such low concentrations that an unprocessed sample may not generate a quantifiable signal when injected into an analytical instrument. Sample concentration involves passage of a known volume of raw water through a solid adsorbent material to which the toxin preferentially partitions. The adsorbent material is then treated with much smaller volumes of a second solvent, such as methanol, in order to remobilise the toxin into the liquid phase. Ideally, all of the toxin originally partitioned on to the solid phase will desorb into a volume of solvent that is orders of magnitude smaller than the original, thus yielding a concentrated sample. The ratio of the original sample volume to the second solvent volume is the concentration factor. This liquid phase, containing the desorbed toxin, is then subjected to further analysis.

The specifics of the concentration procedure described above vary somewhat from toxin to toxin. Variations in the procedure include the type of sorbent and the eluent solvent. The decision to use a procedural variation is driven not only by the type of toxin, but also by the concentration and nature of the background organic material present in the original water sample. As a result, it may take some time to optimise the concentration procedure when toxin analyses are initiated on samples from a new source. This will be the case whether analyses are performed in house or by a contract laboratory. The uncertainty is exacerbated by the fact that standardised analytical procedures do not yet exist for many of the cyanobacterial toxins in many countries. As a result, the most prudent course of action for water supply managers may be to negotiate the desired quality control criteria (internal standard recovery, surrogate recovery, duplicate reproducibility, etc.) and allow the laboratory to choose the method that best meets the contractual requirements.

In a raw water sample, total toxin concentration is measured after all the cyanobacteria in the sample have been lysed to release the toxin into the dissolved state. The most appropriate technique for liberating intracellular toxin is freeze/thawing in the presence of a solvent appropriate for the particular toxin. If dissolved toxin concentration is also required, two samples should be taken, one treated as for total toxin analysis (above). The other should be gravity filtered through glass fibre filter, to avoid damage to the cells, and the filtrate analysed for toxin concentration. The difference between the total toxin concentration and the filtrate, or dissolved toxin concentration, is the intracellular concentration.

·       Use the following formula for the determination of chlorophyll-a:

Where:

                             A₆₆₅                    =              Absorbance at 665nm before acidification

                             A₇₅₀                    =              Absorbance at 750nm before acidification

                             A_665a                   =              Absorbance at 665nm after acidification

                             A_750a                   =              Absorbance at 750nm after acidification

                             28.66                  =              Constant (taking into account: ethanol with its specific absorption coefficient and path length of the cuvette)

                             Ve                      =              Volume of ethanol used for extraction in mL (usually 10mL)

                             Vm                     =              Volume of sample filtered in mL

                             x                        =              Multiplication

·         The chlorophyll-a values are “rounded off” as follows:

0 < Result < 1	Report to 2 decimal places
1 < Result < 10	Report to 1 decimal place
10 < Result	Report to the nearest integer

Inorganic phosphate has a very low toxic potential, however phosphate can interfere with flocculation processes and also stimulate algal growth. Phosphate is known to be the primary limiting nutrient in most cases where cyanobacteria blooms occur, especially with the occurrence of nitrogen fixing cyanobacteria species.

This method covers the determination of orthophosphate, also called Soluble Reactive Phosphate, in water samples using Flow Injection Analysis followed by photometric detection. The method is based on the reactions that are ion specific. The results are expressed in mg L^-1 P. This method is fit for the purpose of, and is suitable for the determination of orthophosphate in drinking, ground, catchment and surface waters, from the range of 0 - 10mg L^-1.

The orthophosphate ions react with ammonium molybdate and antimony potassium tartrate under acidic conditions to form a phospho-molybdic complex. This complex is reduced with ascorbic acid to form a blue complex which absorbs light at 880 nm.

The absorbance is proportional to the concentration of orthophosphate in the sample.

Orthophosphate forms a blue colour in this test. Polyphosphates and organic phosphorus compounds do not react. The sulphuric acid in the molybdate reagent does not have enough contact time with polyphosphates to hydrolyse them.

Two important sources of interferences have been identified in this method:

It should be noted that alternate volumes of reagents, standards and Q.C. verification standards may be prepared if desired. The concentrations however shall conform to those specified. This shall not constitute a method deviation.

(NH₄)₆ Mo₇ O₂₄.4H₂O 40 g

Preparation: Dissolve 40g ammonium molybdate to ± 900mL of water. Dilute to a final volume of 1000mL and mix. The solution is stable for 1 month.

Dissolve 3.0g potassium antimony tartrate in ± 800mL water. Dilute to a final volume of 1000mL and mix. The solution is stable for 1 month.

Dilute 500mL water with 35mL concentrated sulphuric acid (H₂SO₄) (± 96% acidity). Cool the solution. Then add 213mL ammonium molybdate ([NH₄]₆ Mo₇O₂₄.4H₂O) solution and 72mL potassium antimony tartrate (K₂Sb₂NO₃) solution. Dilute to a final volume of 1000mL and mix. This solution must be degassed with argon gas for ± 15 minutes. This solution is stable for one month.

Dissolve 60.0g ascorbic acid (C₆H₈O₆) in ± 700mL water and then add 1.0g sodium dodecyl sulphate. Dilute to a final volume of 1000mL and mix. The solution is stable for one month.

Dissolve 65.0g sodium hydroxide (NaOH) and 6.0 g titriplex (C₁₄H₂₂N₂O₈.H2O) in ± 800mL water. Dilute to a final volume of 1000m and mix. The solution is stable for 2 weeks.

Dry potassium dihydrogen phosphate (KH₂PO₄). Dissolve 4,3940 ± 0.0005g of the dried potassium dihydrogen phosphate (KH2PO4) in ± 800mL water. Make up to a final volume of 1000mL and mix. Store solution in a one litre amber bottle. The solution is stable for 1 year.

Dilute 100mL orthophosphate calibration stock solution to a final volume of 1000mL with reagent water to make up a working stock of 100mg/L. Prepare monthly.

Use A-Grade bulb pipettes and prepare these standards weekly from working stock solution.

Standards	Volume (mL)	Concentration ppm (mg L-1)
1	100 mL in 1000 mL	10 ppm
2	75 mL in 1000 mL	7.5 ppm
3	50 mL in 1000 mL	5 ppm
4	20 mL in 1000 mL	2.0 ppm
5	10 mL in 1000 mL	1.0 ppm
6	5 mL in 1000 mL	0.5 ppm
7	1 mL in 1000 mL	0.1 ppm
8 Blank (Milli-Q water)	0 mL	0 ppm

CALCULATION AND EXPRESSION OF RESULTS:

A method developed in 1865 by an astrophysicist Fr. Pietro Angelo Secchi involves a device called a Secchi disk (Figure 3-6(L3)), to measure the transparency of open waters. It is a quick and inexpensive way to determine transparency of the water.

The Secchi disk with its alternate black and white quadrants are slowly lowered into the water. The depth at which the pattern on the disk is no longer visible is taken as a measure of the Secchi disk depth of that specific water body or part of the water body.

Secchi measurements may be subjective to the operator and more precise measurements should be done with a turbidimeter.

A wide variety of field instruments exist for the determination of temperature in the field (e.g. YSI 556 MPS (Multi Probe System) - Oxygen, Conductivity, pH and Temperature). Since field instruments are usually not as sensitive as bench-top instruments it is very important to calibrate/verify the instruments before going out to the field and on return to the laboratory. Verification is done with a calibrated mercury-filled Celsius thermometer. This thermometer should be calibrated at least once a year by a SANAS / ILAC (or equivalent) accredited testing laboratory. The data is recorded on an applicable form. If the verification data does not fall within the specified limits (±1.0ºC), the analysis should be repeated. If the results are still out of specification, maintenance on the probe should be carried out and the analysis repeated. If the results persist to be out of specification the laboratory supervisor should be informed and he/she should take the necessary action (e.g. to have the instrument serviced and or calibrated by a suitable supplier).

Notes:
1) Temperature readings should be taken to the closest integer, unless specified differently.
2) When calibrated thermometers are used in a refrigerator, it should be kept in a screw-cap tube filled with glycerol.

Please note that this procedure applies to the YSI 556 MPS (Multi Probe System - Oxygen, Conductivity, pH and Temperature), but that the overall quality control and other aspects can be applied to any other applicable instruments.

A wide variety of field instruments exist for the determination of pH in the field (e.g. YSI 556 MPS (Multi Probe System)). Since field instruments are usually not as sensitive as bench-top instruments it is very important to calibrate/verify the instruments before going out to the field and on return to the laboratory.

Calibration is done with pH Buffer solutions from appropriate suppliers. Verification is done with two a verification buffers, namely potassium tetroxalate dihydrate - KH3C4O8.2H2O - and calcium hydroxide - Ca(OH)2. The data is recorded on an applicable form. If the verification data does not fall within the specified limits (±0.2pH units), the analysis should be repeated. If the results are still out of specification, maintenance on the probe should be carried out and the analysis repeated. If the results persist to be out of specification the laboratory supervisor should be informed and he/she should take the necessary action (e.g. to have the instrument serviced by a suitable supplier).

Although this method is suitable for all aqueous samples (source water, sewage, factory effluent and drinking water), at a pH above 10, high sodium ion concentrations may cause interference. The estimated error measurement for samples is 0.2pH units. The analytical range is 0.5 to 13.5pH units. This range has been validated to be true in Rand Water’s hydrobiology laboratory, but validations for any method should be repeated in each laboratory since environmental factors may influence validations.

Notes: 1) pH readings should be taken to two significant digits, unless specified differently.

The basic principle of electrometric pH measurement is the determination of the activity of the hydrogen ions by potentiometric measurement using a standard glass electrode. The electromotive force (emf) produced in the glass electrode system varies linearly with pH. This linear relationship is described by plotting the measured emf against the different pH buffers. Sample pH is determined by extrapolation.

Please note that this procedure applies to the YSI - MPS (Multi Probe System) Oxygen, Conductivity, pH and Temperature System, but that the overall quality control and other aspects can be applied to any other applicable instruments.

Note: It is preferable that pH readings be taken on site at the sampling locality, since pH changes are inevitable when samples are enclosed in a smaller container. If pH readings cannot be taken on site, sample bottles should be filled to the brim and capped tightly after sampling, and transported in a cooler bag with ice bricks. The pH readings should be done within 8 hours of sampling.

Dissolved oxygen (DO) levels in natural and wastewaters depend on the physical, chemical and biochemical activities in the water body. The analysis for DO is a key test in water pollution and waste treatment process control. Two basic methods for DO analysis are generally available: the Winkler or iodometric method (and its modifications) and the electrometric method using membrane electrodes. The iodometric method is a titrimetric procedure based on the oxidising property of DO while the membrane electrode procedure is based on the rate of diffusion of molecular oxygen across a membrane.

The most common method used in the field is the electrometric method using membrane electrodes. This method is suitable for the analysis of Dissolved Oxygen (DO) of surface water, ground water (e.g. borehole water), drinking water, industrial effluent, sewage water and other fluid substances.

The calibration of the DO meter must be performed every time it is used (e.g. in the morning before it goes out to the field). A dirty DO probe is a common source of incorrect calibration and/or erratic results.

Erratic results may be due to:

Dissolved oxygen ranges from 0 to 500% air saturation or 0 to 50mg L^-1. Dissolved oxygen is very sensitive to air pressure and therefore the specific height above sea level is necessary for correct calibration.

Oxygen-sensitive membrane electrodes of the polarographic or galvanic types are composed of two solid metal electrodes in contact with supporting electrolyte separated from the test solution by a selective membrane. The basic difference between the galvanic and the polarographic systems is that in the former the electrode reaction is spontaneous, while in the latter an external source of applied voltage is needed to polarize the indicator electrode.

Different instruments exist for the determination of dissolved oxygen e.g.

Note: This procedure applies to the YSI MPS (Multi Probe System - Oxygen, Conductivity, pH and Temperature), but the overall quality control and other aspects can be applied to any other applicable instruments.

With the handheld systems, the measurement of dissolved oxygen for field determinations is done on site. The samples cannot be stored for later analysis using the handheld systems.

Clarity of water is important in producing products destined for human consumption and manufacturing uses. The clarity of a natural body of water is a major determinant of the condition and primary productivity of that system. Turbidity in water is caused by suspended matter, such as clay, silt, finely divided organic and inorganic matter, soluble coloured organic compounds, plankton and other microscopic organisms. Correlation of turbidity also affects the light scattering properties of the suspension.

The method is suitable for all water i.e. potable water, source water, as well as sewage waters and industrial effluents.

The method is based on a comparison of the intensity of light scattered by the sample under defined conditions with the intensity of light scattered by a standard reference suspension under the same conditions. The higher the intensity of scattered light, the higher the turbidity.

Turbidity can be determined for any water sample that is free of debris and rapidly settling coarse sediment. Dirty glassware, the presence of air bubbles as well as vibration that disturb the surface visibility of the samples will give false results. Temperature fluctuations of the sample may alter suspended particle characteristics, which may also interfere with the readings.

(Note that this is the procedure for the 2100 AN turbidimeter, but can be adapted to suit most turbidimeters):

(Calibration procedure must be performed once a month. The new values for the adjusted Gelex standards and the date of calibration must be recorded. Verification checks must be performed daily before proceeding with the measurement procedure. The checks must be conducted with the secondary Gelex standards that were adjusted with the most recent calibration procedure.)