View all publications

Section 4. Sampling

4.1 Health and safety

When sampling cyanobacteria in lakes or rivers, consideration needs to be given to protecting the sampler. Samplers should wear gloves and rubber waders (rather than neoprene) or gumboots to reduce the risk of skin contact. If sampling when there is excessive foam present and windy conditions, a dust/surgical face mask should be worn. When wading into swift-flowing rivers and streams, standard water-quality sampling procedures (held by most regional councils) should be observed to identify hazards and reduce the risk of being swept downstream.

4.2   Biosecurity

The procedures detailed in this section involve entering water bodies that may contain the introduced invasive diatom didymo (Didymosphenia geminata). Therefore, all equipment and clothing should be decontaminated when leaving a water body where there is any chance of didymo being present. Decontamination protocols for didymo can be downloaded from

4.3   Planktonic cyanobacteria

The design of monitoring programmes for planktonic cyanobacteria is challenging due to factors such as:

  • their ability to grow in open waters
  • the ability of some species to regulate their buoyancy
  • their ability to form scums that may be shifted and concentrated by wind
  • the interactions of buoyant cells with the surface drift currents created by wind
  • the ability of some species to produce toxins that may be contained in their cells or dissolved in water.

Due to these factors, monitoring programmes for planktonic cyanobacteria should be tailored to the characteristics of each water body. They also need to be flexible to take account of changes in the risk posed by rapid changes in the cyanobacterial populations with time and location, which should be recorded along with the sample depth and type. Collection of historical information on blooms and growth conditions, and the identification of patterns of cyanobacterial growth, can be used to help focus the monitoring programme on critical periods and locations in the water body of interest. The aims of the sampling protocols outlined below are to enable an assessment of health hazards caused by planktonic cyanobacteria and their toxins in recreational use waters. Detailed protocols for sampling drinking water are provided by the Ministry of Health (2005a), and protocols for sampling for ecological and other studies are provided by Pridmore (1987) and Codd et al, (1999).

4.3.1    Site selection

The heterogeneous (mixed) and dynamic nature of many cyanobacterial populations can make selecting a sampling site difficult. A flexible response when choosing the sampling sites may, at times, be more appropriate than following a rigid programme. Alternatively, fixed sites can be sampled within a broader monitoring programme to provide linear time series, supplemented by sampling of sites currently harbouring cyanobacterial scums.

The selection of sampling sites is a key factor in collecting representative samples. The following should be considered.

A.         Use of the site for contact recreation

  • Sampling sites should include shoreline areas frequented by recreational users, perhaps with a focus on public bathing sites.
  • Make use of local logistical resources, and consider accessibility and safety factors.

B.        Risk of a site having cyanobacterial blooms/mats

  • The history, if available, of cyanobacterial population development and the occurrence of toxins in the water body is useful. This information may indicate sites most likely to harbour scums/mats.
  • Specific incidents, such as animal deaths or human illness, may provide indications of ‘high risk’ sites.
  • Morphometric and hydrophysical characteristics of the water body (eg, exposure to wind or thermal stratification) may help identify sites that are prone to scum accumulation.
  • Prevailing weather conditions, particularly wind direction, can lead to scum accumulation along certain shorelines.

4.3.2    Sample collection

An entry-point or near-shore sample should consist of a composite sample comprising five 50 cm depth-integrated column (hosepipe) sub-samples collected relatively randomly along an approximately 20–30 m transect (parallel to the lake shore) and mixed into a single container (eg, a bucket). From this, a composite sample is taken for the cell counts and/or toxin analysis.

The rationale for this sample type is that:

  • the 50 cm integrated column or tube covers the surface zone that recreational users are most likely to be exposed to
  • the sampling of this shallow 0–50 cm zone also covers the accumulation of buoyant cyanobacteria near the surface under calm conditions
  • the recommendation for five pooled sub-samples accounts for spatial variability within a single site.

The volume of the composite sample required will vary. When sampling eutrophic lakes, 100 mL is usually sufficient for cyanobacterial identification and 500 mL for cyanotoxin analysis. In oligotrophic lakes, two 500 mL samples are required: one for identification and one for cyanotoxins (see Sections 4.5.1 and 4.5.2).

Integrated samples can be collected using a rigid or flexible plastic hosepipe with an inner diameter of at least 2.5 cm; a rigid polyvinylchloride (PVC) or acrylic plastic pipe is more practical than a flexible pipe.

Where wading or boat access is not available, the alternative is to collect a pooled surface-grab (ie, dipped bucket samples). Additional individual, non-composite samples should also be collected where scums or obvious discoloured water are encountered. These individual ‘grab’ samples represent the maximum hazard at the time of inspection and may assist in the overall health risk assessment.

It is advisable to collect samples in the morning because cyanobacterial blooms are usually at their densest at the surface in the early morning. For comparative purposes, the sampling time should be consistent between sampling trips, where practical.

The frequency of samples collected at any one location is dictated by the alert-level framework (see Section 3.2).

4.3.3    Field data records

It is important to record all relevant details about the sampling site, sampling methods and prevailing conditions. The following should be noted, where possible:

  • weather conditions at the time of sampling and 24 hours prior to sampling (including wind direction and strength)
  • water transparency (use a Secchi disc if available)
  • any discolouration of the water or signs of blooms or mats
  • water temperature
  • dissolved oxygen.

Integration of sampling with a more comprehensive water-quality sampling programme will help to develop an understanding of the causal factors promoting cyanobacterial growth for each specific water body (see Section 4.6).

Interpretation of the significance of a particular cyanobacterial cell concentration in relation to others may require an examination of the field sheet to verify the type of sample collected (ie, surface, depth or integrated depth) or the place or time of collection. An example of a typical field sheet is provided in Appendix 12.

4.4   Benthic cyanobacteria

The method described below is intended for use in rivers where cyanobacterial mats are likely to occur and is recommended as a quick, easy and reproducible way of keeping a record of benthic cyanobacterial coverage. These records are designed to help assess the risk posed by cyanobacteria in rivers under recreational use. Routine sampling is recommended under low coverage (< 20%) if there is any doubt about the identity of observed algal mats.

At established sites it should be possible to complete the survey procedure in 15–20 minutes (ie, completion of the survey form, Appendix 13). The greatest investment in time will occur during site selection and collecting background information on the site.

4.4.1    Site selection and collecting background site information

Refer to Section 4.3.1 for key factors that should be considered when selecting sampling sites. Cyanobacterial mats tend to proliferate initially in riffles, then runs, so priority should be given to examining these habitat types.

On the first visit to the site choose a 40–60 m reach where a survey can be undertaken on a regular basis. Where possible, collect the following background information for each site:

  • reach length and river width (measure or estimate – photos are useful)
  • substrate composition (ie, bed substrate type – cobbles, gravels, sand-silt)
  • water velocity
  • amount of shade at each survey reach
  • bank vegetation (descriptive – this could be captured photographically)
  • hydrology (eg, time since last flood, 2x and 3x median flow, see Section 3.7).

Integration within a broader programme of water-quality monitoring may be useful.

4.4.2    Site surveying and sample collection

The following equipment is required to undertake a benthic cyanobacterial assessment:

  • Underwater viewer (Figure 1): this is constructed from clear perspex or bought (eg, a Nuova Rade viewer, available from These viewers allow a clear view of the stream bed with no interference from surface turbulence and reflection. They also enable a more-or-less standard area of the stream bed to be defined at each survey point (equivalent to a quadrat in terrestrial ecology). Photographs can be taken through these viewers for improved documentation of mat coverage.
  • Clipboard, pencils and monitoring forms (see Appendix 13): forms should preferably be printed on waterproof paper.
  • Sampling containers and permanent marker or equivalent (for labelling).

Figure 1: Using an underwater viewer

Figure 1: Using an underwater viewer

Photos: S Wood, Cawthron Institute  

Two photographs showing how the underwater viewer is used to get a clear view of the stream bed.

Ideally the survey should be undertaken in teams of two: one observer and one scribe. However, some tips are provided for one-person surveys below.

All monitoring should be undertaken under similar flow conditions (eg, at no more than median flow). This ensures the surveys always cover the permanently wetted channel. Surveys in very low flows are acceptable, but higher flows should be avoided due to associated safety issues and reduced water clarity.

4.4.3    Monitoring procedure

  1. After arriving at a survey area, spend approximately five minutes looking along a 30−60 m section of river bed for the presence of cyanobacterial mats. Ensure this section includes some riffles and runs. Mark out four transects in the selected area by placing marker rocks along the water’s edge, approximately 10–15 m apart.
  2. Complete the first section of the monitoring form (Appendix 13) with site, date, time, etc, and note the general presence/absence of cyanobacterial mat and the presence of any detached mat along the shoreline.
  3. Assemble the underwater viewer and, starting at the downstream end, wade into the stream at right angles to the water’s edge. Go out to a depth of approximately 0.6 m (see Figure 2 and Figure 3) A standard maximum depth of 0.6 m should be used at all sites, where possible. In shallow rivers the transects may span the entire width. Wading into fast-flowing water can be dangerous and extreme care should be taken.

Figure 2: Schematic of layout of transects (numbered in red) and survey areas (red circles, numbered in black) at a site (not to scale)

Figure 2: Schematic of layout of transects (numbered in red) and survey areas (red circles, numbered in black) at a site (not to scale)

Diagram showing aerial view of the monitoring layout in a river section. The river section is between 40 to 60 metres long. Four transects go from the water edge to a maximum depth of 0.6 metres (perpendicular to the water flow). The transects are monitored from the most downstream upstream. Each transect has five survey areas.

Figure 3: Schematic of transect cross-section showing arrangement of sampling points (not to scale)

Figure 3: Schematic of transect cross-section showing arrangement of sampling points (not to scale)

Diagram showing cross-section view of one transect in a monitored river section. The transect includes five viewing areas equally spaced along the transect to a depth of 0.6 metres (ie, divide transect length by five to get distance between each view). The viewing area at a depth of 0.6 metres is monitored first, monitoring the other four viewing areas as you move along the transect towards the water's edge.

  1. Record the maximum distance and depth in the boxes at the top of the column for transect 1.
  2. Hold the underwater viewer about 20 cm under the water, more or less on the transect line. The area of view should not be one that has just been walked over. Holding the viewer steady and as vertical as possible, estimate to the nearest 5 per cent the proportion of the area you see that is occupied by the cyanobacterial mat. Some examples are shown in Figure 4. Cyanobacterial mats are usually dark black, dark brown or dark green in colour, leathery, and have an earthy, musty odour. Refer to Appendix 7 for a photographic guide to benthic algae/cyanobacteria. Coverage should only be recorded if mats are greater than 1 mm thick, although it is useful to record the presence of thin mats.
  3. If there is any doubt about the identity of mat cover (ie, whether it is cyanobacteria) at any sampling point, take a sample for microscopic identification. Samples should be collected by scraping an egg-sized clump of mat into a sampling pottle. Samples for microscopy should be preserved with Lugol’s iodine (see Appendix 14), whereas samples for cyanotoxin analysis SHOULDNOT be preserved (see Section 4.3.2). Toxin content can vary markedly between rocks within a site (Wood, Heath et al, in press), so where possible take 10 samples from separate rocks and pool these for toxin analysis. To enable the amount of toxin within an area to be estimated, the sample should be taken from a known area on the rock (eg, use the top of a sampling pottle to mark out the sampling area) by scraping all periphyton from that defined area into a sampling pottle. Label the sampling container with the site name, area sampled and transect number.
  4. Record the percentage cover in the appropriate boxes for each transect. Ideally, be consistent with the order of survey points on each transect (eg, point 1 is always the deepest into the water and 5 is always closest to the waters’ edge, see Figures 2 and 3). Indicate at which (if any) sites samples were taken. Record any notes regarding other algal cover (eg, green filaments overgrowing cyanobacterial mats).
  5. Space the points evenly along the transect to a depth of 0.1–0.15 m nearest to the water’s edge, although this depth will vary according to the type of river. For example, if the river bank is incised (channelised), the closest survey point will be deeper.

Figure 4: Examples of different levels of cyanobacterial cover viewed through an underwater viewer

Figure 4: Examples of different levels of cyanobacterial cover viewed through an underwater viewer

Six photographs showing examples of cyanobacterial cover of 5, 15, 20, 40, 65 and 80% as it would appear if seen through an underwater viewer.

  1. Move upstream to transects 2, 3 and 4, and repeat steps 5 to 9 to complete the survey at this site.
  2. Calculate the average percentage cover per transect and then the average percentage cover per site. Average percentage cover results for each site should be interpreted via the alert-level framework (Section 3.5) and the appropriate actions taken.

The frequency of samples collected at any one location is dictated by the alert-level framework (Section 3.5).

For health and safety reasons it is usually advisable to work in teams of two or more. However, there may be occasions when only one person is available. In this case, a single person must handle the equipment for both observing and recording. Here are some tips to make this easier.

  • Always use data sheets copied onto waterproof paper (eg, Rite-in-the-Rain paper).
  • Reduce the size of the data sheet to A5 to create a smaller, more manageable clipboard.
  • Tether the viewer securely to your waist so that both your hands are free for writing.
  • Secure the clipboard and pencil to your waders so that, if dropped the items can be retrieved without damage.
  • Tie a small towel to the wader shoulder strap so that it is possible to dry your hands before writing.
  • For estimating water depth, mark a scale on the side of the waders in, for example, 5 cm intervals.

4.5   Sample storage and transport

The following are standard protocols for sample preservation, storage and transport. Analytical laboratories may have specific requirements and it is strongly recommended that you contact the relevant laboratory (see Appendix 8) well before sample collection.

4.5.1    Cyanobacterial identification and enumeration

Sub-samples should be preserved as soon as possible after collection by the addition of 1 per cent acid Lugol’s iodine preservative (Appendix 14). Lugol’s iodine is added drop by drop until the sample is the colour of beer or weak tea (approximately 4 drops per 100 mL in water). Dense samples (eg, scum material or benthic mats) will absorb Lugol’s and may require additional Lugol’s if long-term storage is required.

Samples should be stored in the dark. Some plastic bottles (polyethylene) tend to absorb iodine very quickly into the plastic, so care should be taken with any samples requiring longer term storage. It is useful to retain a portion of sample in a live (unpreserved) state, as cyanobacteria are often easier to identify in this way. Live samples degrade quickly, however, and a small amount of material should be collected and covered with water. Ensure there is plenty of air space above the sample and refrigerate. Examine as soon as possible after collection. Each bottle should be labelled clearly with the site name and location, approximate depth, date, sample type (integrated or grab), sampler’s name, and indication of whether Lugol’s has been added.

4.5.2    Cyanobacterial toxins

Samples for toxin analysis should be stored in glass bottles, where possible, because plastics may absorb cyanotoxins. The volume of sample required depends on the type of analysis. For planktonic samples, at least 500 mL of water should be collected. Benthic samples should be collected as described in Section 4.4.3 (point 6).

Cyanotoxins are readily degraded, both photochemically (in light) and microbially. Samples should be transported in dark, cold conditions and kept refrigerated prior to analysis. Where samples for toxin analysis won’t reach the analytical laboratory within 24 hours, samples can be stored frozen. However, note that freezing releases cyanotoxins from the cells and so only the total amount of toxins in a sample can be determined.

4.6   Susceptibility of a water body to a cyanobacterial bloom or benthic mat event

In some regions it may not be practical to monitor phytoplankton abundance in all water bodies where there is recreational use. The decision on which water bodies to monitor should be based on a combination of:

  • the amount of recreational use
  • pre-existing knowledge of the characteristics of the water body
  • any prior monitoring data on cyanobacterial events.

Aerial and satellite photography can be used to derive information on water clarity and phytoplankton biomass of larger surface waters, which then provides a basis for comparative assessments between different water bodies.

A decision support tree can also be useful to help assess the likelihood of a cyanobacterial bloom occurring in a water body. Figure 5 shows such an example, which has been extended to include the risk of formation of a benthic cyanobacterial mat as well as the occurrence of a cyanobacterial bloom. The logic proposed for cyanobacterial blooms is similar to that given by Oliver and Ganf (2000), but it has been adapted on the basis of New Zealand observations and simplified, as follows.

  • A prerequisite for blooms in Oliver and Ganf’s model is an available phosphorus concentration in excess of 10 mg/m3. Blooms of cyanobacteria have been observed in many New Zealand lakes when available phosphorus concentrations are close to detection limits, and in Lake Taupo, for example, when total phosphorus concentrations are below 10 mg/m3.
  • Some variables in Oliver and Ganf’s model that may not be easy to measure (eg, grazing, turbulent water velocity and cell floating velocity) are not included in the model presented in Figure 5.
  • A model to assist with the prediction of benthic mats of cyanobacteria is included for both lakes and rivers.
  • The inclusion in the decision tree for bloom development of stratified lakes in which bottom waters become anoxic does not reflect a direct process or causal linkage to bloom development. It is simply an observation that this category of lakes appears to be susceptible to cyanobacterial blooms.

Because of the complexity of interactions among nutrients, wind and lake size, probabilistic functions have been added to the decision support tree for a subset of water bodies (see Figure 6). These functions are intended to represent the fact that simple ‘yes’ or ‘no’ decisions are not always possible and that knowledge of the causal factors of cyanobacterial blooms is also imperfect. The subset of water bodies for which these probabilistic functions applies includes deeper lakes that undergo seasonal stratification but whose bottom waters do not become anoxic.

Feedback requested on Figures 5 and 6

Note that the decision support tree presented in Figure 5 is intended for testing and feedback, to provide a basis for iterative improvement and development on the basis of regional and local information. Your feedback would be appreciated.

The probability charts shown in Figure 6 are a first attempt to apply numerical values of probability for the occurrence of a cyanobacterial bloom. The inclusion of the variables lake area, total phosphorus concentration, dissolved inorganic nitrogen concentration (NO3-N + NO2-N + NH4-N) and wind speed reflect the fact that there is a gradation in the response of cyanobacterial biomass to these variables rather than an abrupt transition denoted by ‘yes’ or ‘no’. The probability charts given in Figure 6 provide, like the decision support tree in Figure 5, an opportunity for testing and iterative refinement and development as information is accumulated.

The individual probabilities (with values from zero to 1) for bloom occurrence are denoted as follows:

P(lake area)     = 0.0816 x Ln(A) + 0.41, where A is lake area in km2, and Ln is the natural logarithm
P(total phosphorus)  = 0.17 x Ln(TP), where TP is total phosphorus concentration (mg/m3)
P(IN)    = -0.0965 x Ln(IN) +0.88, where IN is the total dissolved inorganic nitrogen concentration (mg/m3)
P(U)    = 0.0032 x U4 – 0.037 x U3 +0.1084 x U2 – 0.1818 x U + 1.00, where U is wind speed (m/s) averaged over a period of six hours.

The basis of these functions is that increasing lake area leads to an increased likelihood that blooms will be ‘magnified’ at the water surface. The function relating to total phosphorus concentrations is intended to reflect the fact that an increased supply of phosphorus will increase cyanobacterial biomass as this group is generally a poor competitor under conditions of strong phosphorus limitation. High levels of inorganically bound phosphorus are more likely to occur under other options given in the flow chart of Figure 5 (eg, shallow, turbid lakes). A decreasing probability of cyanobacterial blooms with increasing inorganic nitrogen is intended to reflect the predominance of nitrogen-fixing cyanobacteria (especially Anabaena) as inorganic nitrogen becomes strongly limiting. The general trend of increasing probability of blooms with increasing nutrient concentrations is already reflected in the P(TP) function. Finally, the probability of a cyanobacterial bloom increases when wind speed decreases. A duration of six hours was chosen for this function, but there will inevitably be some interaction of duration and lake size: large lakes will have greater inertia and therefore respond more slowly than small lakes to changes in wind speed. This is not reflected in the current model.

Figure 5: Decision tree summarising the major environmental variables important in the development of cyanobacterial blooms and benthic mats, and the selection of specific genera

Figure 5: Decision tree summarising the major environmental variables important in the development of cyanobacterial blooms and benthic mats, and the selection of specific genera

Question 1 - Is the water running ie, fluvial stream or river

Yes - Go to question 2
No (water is quiescent ie, lake, reservoir or pond) - Go to question 6

Question 2 - Is the water temperature greater than 15ºC?

Yes - Go to question 3
No - Low probability of dense benthic cyanobacterial mats

Question 3 - Is the streambed substrate stable?

Yes - Go to question 4
No - Low probability of dense benthic cyanobacterial mats

Question 5 - Is the river or streambed well illuminated?

Yes - Potential for blooms of benthic cyanobacteria: primarily Phormidium
No - Low probability of dense benthic cyanobacterial mats

Question 6 - Is the water temperature greater than 10ºC for at least 1 month?

Yes - Go to question 7
No - Low probability of bloom

Question 7 - Is the bottom sediments illuminated, hard substrate over considerable area of lake and water clear?

Yes - Potential for blooms of benthic cyanobacteria: primarily Phormidium
No - Go to question 8

Question 8 - Is the flushing rate low (ie, less than 10 day turnover)?

Yes - Go to question 9
No - Low probability of bloom upstream inoculum high (amber mode or above)

Question 9 - Is the lake deep and seasonally stratified?

Yes - Go to question 10
No (lake is shallow and generally well mixed or perhaps polymictic) - Go to question 12

Question 10 - Is the mixing depth less than 3 times depth of visibility (measured with secchi disk)?

Yes - Low probability of surface bloom but potential for metalimnetic accumulations of cyanobacteria. These populations may occasionally switch to surface blooms for brief periods.
No - Go to question 11

Question 11 - Is the bottom waters seasonally anoxic (devoid of oxygen)?

Yes - High probability of blooms of Anabaena and Aphanizomenon (especially if inorganic N near detection limits) and Microcystis (especially when ammonium concentrations elevated over nitrate)
No - See probability charts (figure 6)

Question 12 - Is the water highly turbid (high levels of resuspended sediment)?

Yes - High probability of dispersed populations (Cylindrospermopsis - North Island, Anabaena, possibly Microcystis) when mixed, or surface blooms (Anabaena, possibly Microcystis) when calm/light winds and clarity improves (secchi disk > 15 cm)

To determine an overall weighted probability, the individual functions are weighted as follows:

P(weighted) = 0.2 x P(lake area) + 0.4 x P(TP) + 0.15 x P(IN) +0.25 x P(U).

This weighted probability function P (weighted) can then be interpreted according to the surveillance level (green mode), alert level (amber mode) and action level (red mode), as shown in Figure 6. Excel spreadsheets of the different functions (P(A), P(TP), P(IN), P(U) and P(weighted)) can readily be created or made available if required, and feedback from applications will allow this model to be refined and adapted to a wide range of conditions.

Figure 6: Probability charts demonstrating the individual probability contributions of lake area, total phosphorus, total dissolved inorganic nitrogen and 6-hour average wind speed to the weighted probability, assessed as P(weighted) = 0.2 x P(lake area) + 0.4 x P(TP) + 0.15 x P(IN) + 0.25 x P(U), and interpreted in terms of surveillance, alert and action levels with a colour bar

Figure 6: Probability charts demonstrating the individual probability contributions of lake area, total phosphorus, total dissolved inorganic nitrogen and 6-hour average wind speed to the weighted probability, assessed as P(weighted) = 0.2 x P(lake area) + 0.4 x P(TP) + 0.15 x P(IN) + 0.25 x P(U), and interpreted in terms of surveillance, alert and action levels with a colour bar

Probability charts showing that the probability (p) increases as lake area (A) increases, total phosphorus (TP) increases, dissolved inorganic nitrogen (IN) decreases and average wind speed (U) decreases. The formulae for each of the probability charts are: Lake area, A (km2): p = 0.0816Ln(A) + 0.41 Total phosphorus, TP (mg m-3): p = 0.17Ln(TP) Dissolved inorganic nitrogen, IN (mg m-3): p = -0.0965Ln(IN) + 0.88 Average wind speed, U (m s-1): p = 0.0032U4 - 0.037U3 + 0.1084U2 - 0.1818U + 1.0 The colour bar shows that as the weighted probability increases from zero to one it moves it from surveillance level (green) through alert level (amber) to action level (red).