Cyanobacteria blooms can complicate the economical or recreational use of waters. Many of the bloom forming species are also potential producers of harmful cyanotoxins. The standard method for quantifying phytoplankton biomass, based on inverted microscopy, has high accuracy and is the only one producing biomass results on taxonomic level, but it requires specialized expertise and is time-consuming. Phycocyanin (PC) pigment concentration has been proven as a useful proxy for the concentration of cyanobacteria. Since 2006, we have studied practical solutions of in-situ monitoring of cyanobacteria using PC fluorescence probes. We have studied two eutrophic lakes, Lake Littoistenjärvi and Lake Kuralanjärvi in southwestern Finland using stationary monitoring stations equipped with PC probes. The fluorescence results were compared to independent water samples analyzed using standard methods. The PC fluorescence was positively correlated to cyanobacteria biomass in both lakes. Using site-specific post-calibrations of biomass, PC fluorescence can be used to estimate the absolute biomass of cyanobacteria. The monitoring techniques used in these studies are an applicable and relatively low-cost method to monitor cyanobacteria abundance. With nearly real-time data transfer possibilities, they can be used in management and early warning applications to minimize the harmful effects of cyanobacteria blooms.

Eutrophication is currently recognized as the major threat to good ecological state of Finnish freshwaters and the Baltic Sea. Large and intense phytoplankton blooms, most commonly associated to cyanobacteria, have been increasingly reported during the recent decades. Cyanobacteria blooms can complicate or even stop the economical or recreational use of waters. Many of the bloom forming cyanobacteria species are also potential producers of harmful cyanotoxins. Managing authorities need to actively monitor current bloom situation so that possible harmful effects to human and animal health can be minimized. The most commonly used standard method to quantify phytoplankton biomass is based on inverted microscopy (Utermöhl-method). With the traditional widely accepted standard methods the results are comparable, accurate and usually reliable. The microscopy method however needs specialized expertise and is time consuming and expensive. Phycocyanin (PC) is a photosynthetic and strongly fluorescent pigment which is found as major pigment in cyanobacteria and Cryptophyceae and as a trace pigment in Rhodophyceae (Wetzel 2001). It has an absorption peak at 620 nm and fluorescence emission peak at 647 nm (Bastien et al. 2011). In freshwaters cyanobacteria are the only organisms to produce significant quantities of PC (Wetzel 2001). PC concentration has been proven to indicate the actual concentration of cyanobacteria cells and it can be used as an indicator cyanobacteria biomass (Gregor et al. 2007). Since 2006, we have studied practical solutions for continuous on-line and in situ monitoring of cyanobacteria using PC fluorescence probes suitable for field use.

We have studied two eutrophic lakes, Lake Littoistenjärvi (60°27.2N 22°23.5E) during years 2006–2012 and Lake Kuralanjärvi (60°23.6N 21°52.2E) in 2008 in southwestern Finland, using stationary monitoring stations equipped with TriOS microflu-blue probes. The probe is fully submersible and compact sized. The source red LED light uses peak wavelength at 620 nm and the detector is photodiodes with internal interference filter measuring peak wavelength of 655 nm. According to manufacturer (trios.de) the probe should be sensitive and selective for PC fluorescence and insensitive to potential optical interferences like chlorophyll, turbidity and dissolved organics. The probe type has been on market since 2004 and has been used in several published studies (e.g Brient et al. 2008; Bastien et al. 2011). In our studies, the basic setup was a TriOS microflu-blue probe mounted to a 12 VDC battery powered monitoring station equipped with a datalogger and a GSM-modem. The measurement interval was set to log the PC fluorescence every hour. One fluorescence measurement was based on average of 1000 readings taken during a 10 second period. Fluorescence signal results were converted to wet weight biomass in the datalogger using manufacturer's fixed calibration. The station deployment periods have started in early summer (May-June) and ended in late autumn (October-November), before water freezes. The monitoring stations were deployed to floating platforms and the PC fluorescence probe was submersed in 0.5 meters water depth. During the years 2006–2012 the Lake Littoistenjärvi monitoring station collected 143–180 days and 3400–4300 observations annually. To assess the absolute cyanobacteria biomass, the fluorescence station results were compared to independent composite water samples analyzed using the standard inverted microscopy method. In Lake Littoistenjärvi, the samples were taken as part of a long-term monitoring program at ca. two week intervals during the open water period between April and October (11–13 samples per year). In Lake Kuralanjärvi, PC fluorescence data and water samples from two sampling stations were collected during a cyanobacteria removal experiment in May–October 2008.

The biomass results of the phytoplankton samples and the PC fluorescence results (12am and daily mean) from the automated measuring stations were compared using linear regression. In both lakes studied, the PC fluorescence measurement results corresponded well to the independent water sample microscopy results. Best relationships (Lake Littoistenjärvi R2 = 0.59, p < 0.001, N = 57 and Lake Kuralanjärvi R2 = 0.83, p < 0.001, N = 18) were observed between the daily mean reading from fluoroprobe station and the water sample cyanobacteria biomass (Figure 1). No relationships were established between the fluorescence results and the biomass of other phytoplankton groups. Because of cyanobacteria dominance in both lakes, the statistical relationship between PC fluorescence daily mean and total phytoplankton biomass was stronger in some cases, but fitted poorly to the early season results, when non-cyanobacteria species were more abundant and cyanobacteria biomass was low.
Figure 1

The relationships between PC fluorescence station raw cyanobacteria biomass and water sample biomass in Lake Littoistenjärvi 2006–2012 (N = 57) and Lake Kuralanjärvi 2008 (N = 18).

Figure 1

The relationships between PC fluorescence station raw cyanobacteria biomass and water sample biomass in Lake Littoistenjärvi 2006–2012 (N = 57) and Lake Kuralanjärvi 2008 (N = 18).

Close modal
Fluorescence station raw results were post-calibrated using the coefficients of regression curves in Figure 1. After the post-calibration the absolute biomass difference was small and no statistically significant difference (paired t-test) was found between the sample pairs (Littoistenjärvi t = 0.819, df = 56, p = 0.416 and Kuralanjärvi t = −0.212, df = 24, p = 0.834). The observed succession patterns were similar by both methods (Lake Littoistenjärvi example in Figure 2). However, the lake specific coefficients were different. This can probably be explained by different phytoplankton composition ,(e.g. varying pigment concentrations of different species) and the effect of sampling locations. According to manufacturers, (e.g. TriOS and YSI), calibrations with a primary standard (cyanobacteria cultures or extracted PC) are not practical for field use to obtain absolute biomass results and the calibrations should be done by comparing the fluorescence signal to water sample results from the deployment site. Observed reasons for erroneous estimation of cyanobacteria presence and abundance include high turbidity, the occurrence of other phytoplankton taxa containing PC pigmentation (mainly Cryptophyceae), presence of picocyanobacteria and extracellular PC presence in the water after the seasonal succession collapse (Bastien et al. 2011; McQuaid et al. 2011; Zamyadi et al. 2012). These reasons can cause elevated detection threshold and under/overestimates. Biofouling and other mechanical disturbance to optical sensors is one of the most critical technical issues when automated monitoring stations and instruments are deployed to field conditions. To ensure good data quality, effective anti-fouling and cleaning measures and sufficient maintenance interval should be applied during the deployment period. Automated data transfer makes remote controlling of error sources and technical malfunctions easier and faster. The technical performance of the TriOS probe and the datalogger station used was very reliable and there was only very minor data loss during the deployment periods.
Figure 2

Post-calibrated biomass (daily mean) from the PC fluorescence probe station and water sample cyanobacteria biomass from Lake Littoistenjärvi 2006–2012.

Figure 2

Post-calibrated biomass (daily mean) from the PC fluorescence probe station and water sample cyanobacteria biomass from Lake Littoistenjärvi 2006–2012.

Close modal

In stationary monitoring the spatial coverage and representativeness can be a problem. A fixed sampling depth and position can produce a bias to estimating the cyanobacteria biomass in a wider scale. In Lake Littoistenjärvi the comparability between the sampling locations (distance ca. 1 km) was good on relative scale but the absolute biomass values of different methods varied. Part of the difference can probably be explained by natural variation in regional and vertical distribution of cyanobacteria, as the distribution of phytoplankton in the water column is affected by meteorological factors and vertical migration of cells (Wetzel 2001). Temporal representativeness of in situ data is high. Anttila et al. (2012) concluded in a study conducted in Lake Vesijärvi, Finland that manual sampling, e.g. monthly can create errors in seasonal phytoplankton statistics and the sample intervals and the timing of sampling in monitoring programs should be designed using better knowledge of temporal variation. High temporal interval data produced by in situ monitoring can be used for this purpose and it also supports the evaluation of usefulness of the existing datasets.

Application value of in situ PC online monitoring systems is high. Using local site specific biomass calibrations, PC fluorescence can be used to estimate the absolute biomass of cyanobacteria. The fluorometer probe station techniques used in these studies are an applicable and relatively low-cost method to monitor seasonal succession and short-term changes in cyanobacteria abundance. The equipment for online in situ monitoring has developed more reliable and easy to use and also the cost of the device has decreased. With nearly real-time data transfer possibilities, they can be used in management and early warning applications. To rapidly minimize the negative effects of possibly harmful and toxic cyanobacteria blooms, automated monitoring stations measuring PC fluorescence should be considered as a cost-effective and valuable tool when up-to-date information is needed in sensitive locations, as in water intake facilities and public recreational sites.

Anttila
S.
Ketola
M.
Vakkilainen
K.
Kairesalo
T.
2012
Assessing temporal representativeness of water quality monitoring data
.
J.Environ.Monit.
14
,
589
595
.
Bastien
C.
Cardin
R.
Veilleux
E.
Deplois
C.
Warren
A.
Laurion
I.
2011
Performance evaluation of phycocyanin probes for the monitoring of cyanobacteria
.
J.Environ.Monit.
13
,
110
118
.
Brient
L.
Lengronne
M.
Bertrand
E.
Rolland
D.
Sipel
A.
Steinmann
D.
Baudin
M.
Le Rouzic
B.
Bormans
M.
2008
A phycocyanin probe as a tool for monitoring cyanobacteria in freshwater bodies
.
J.Environ.Monit.
10
,
248
255
.
Wetzel
R. G.
2001
Limnology: Lake and River Ecosystems
. 3rd edn,
Academic Press
,
San Diego
.