Harmful Algae

Volume 54, April 2016, Pages 160-173
Harmful Algae

Review
Challenges for mapping cyanotoxin patterns from remote sensing of cyanobacteria

https://doi.org/10.1016/j.hal.2016.01.005Get rights and content

Abstract

Using satellite imagery to quantify the spatial patterns of cyanobacterial toxins has several challenges. These challenges include the need for surrogate pigments – since cyanotoxins cannot be directly detected by remote sensing, the variability in the relationship between the pigments and cyanotoxins – especially microcystins (MC), and the lack of standardization of the various measurement methods. A dual-model strategy can provide an approach to address these challenges. One model uses either chlorophyll-a (Chl-a) or phycocyanin (PC) collected in situ as a surrogate to estimate the MC concentration. The other uses a remote sensing algorithm to estimate the concentration of the surrogate pigment. Where blooms are mixtures of cyanobacteria and eukaryotic algae, PC should be the preferred surrogate to Chl-a. Where cyanobacteria dominate, Chl-a is a better surrogate than PC for remote sensing. Phycocyanin is less sensitive to detection by optical remote sensing, it is less frequently measured, PC laboratory methods are still not standardized, and PC has greater intracellular variability. Either pigment should not be presumed to have a fixed relationship with MC for any water body. The MC-pigment relationship can be valid over weeks, but have considerable intra- and inter-annual variability due to changes in the amount of MC produced relative to cyanobacterial biomass. To detect pigments by satellite, three classes of algorithms (analytic, semi-analytic, and derivative) have been used. Analytical and semi-analytical algorithms are more sensitive but less robust than derivatives because they depend on accurate atmospheric correction; as a result derivatives are more commonly used. Derivatives can estimate Chl-a concentration, and research suggests they can detect and possibly quantify PC. Derivative algorithms, however, need to be standardized in order to evaluate the reproducibility of parameterizations between lakes. A strategy for producing useful estimates of microcystins from cyanobacterial biomass is described, provided cyanotoxin variability is addressed.

Introduction

Toxic cyanobacterial blooms are increasingly a public health and water management concern. These blooms occur globally, causing animal deaths, human health risks, expenses to public water suppliers and a nuisance to environmental and recreational management communities (Chorus and Bartram, 1999, Ibelings et al., 2014). The variety of cyanotoxins produced by cyanobacteria includes hepatotoxins (e.g. microcystins, cylindrospermopsin), and neurotoxins (e.g. anatoxins and saxitoxins). Microcystins (MC) are the most common, and are produced by the ubiquitous cyanobacterium, Microcystis aeruginosa, as well as several other cyanobacterial genera (O’Neil et al., 2012, Pearson et al., 2016). Identifying the presence and distribution of cyanotoxins in lakes and reservoirs will benefit water suppliers and public health managers.
Numerous studies have evaluated or applied remote sensing as a means of mapping, evaluating, or monitoring cyanobacterial blooms. These studies have proposed a variety of algorithms for detecting or quantifying cyanobacterial blooms with different sensors (see reviews by Kutser, 2009, Matthews, 2011, Sass et al., 2007). Satellites have been used for retrospective evaluation of cyanobacterial blooms (Binding et al., 2013, Kahru and Elmgren, 2014, Matthews et al., 2012, Moradi, 2014, Palmer et al., 2015a, Palmer et al., 2015b; Palmer et al., 2015b; Stumpf et al., 2012, Wynne et al., 2010), and as part of monitoring programs (Gómez et al., 2011, Wynne et al., 2013b, Zhang and Duan, 2008). Not all strains of Microcystis and other cyanotoxin-producing taxa produce cyanotoxins (O’Neil et al., 2012), therefore maps of bloom intensity cannot be assumed to translate to valid maps of cyanotoxin distribution.
Of the several cyanotoxins, MCs are the most understood and the most commonly monitored. Microcystins are not pigments and do not absorb visible or near-infrared (NIR) light (Al-Ammar et al., 2013), so remote sensing cannot directly detect them. Mapping MCs with remote sensing will then require a dual-model strategy: (1) characterizing a relationship between the cyanotoxin and a surrogate pigment that can be reliably detected by satellite and (2) establishing the relationship between satellite observations and the surrogate. This concept has been used in other remotely sensed applications for aquatic environments. One of the simplest examples involves estimating salinity in coastal waters using colored dissolved organic matter (CDOM) (D'sa et al., 2002, Salisbury et al., 2011). Satellite-derived optical properties are also used as input to more sophisticated models; for example, those for primary productivity use chlorophyll-a (Chl-a) and light attenuation estimates as inputs (Behrenfeld et al., 2006). A dual-model approach for MCs was presented by Shi et al. (2015) for Lake Taihu, China. They used a satellite model to estimate Chl-a, and then used a few weeks of field data to establish a fixed relationship between MC and Chl-a. They recognized that this relationship would be unlikely to apply to other lakes because of known variations in the relationship between MC and Chl-a (Ha et al., 2011, Shi et al., 2015); although they proposed that this relationship should apply routinely for Lake Taihu. Understanding the relationships between MCs and pigments will be necessary to the cyanotoxin modeling strategy.
The dual-model strategy is both necessary and advantageous to cyanotoxin mapping. From an analytical perspective, models that directly relate satellite observations to the MC concentration will over-fit the data, because the models typically have insufficient observations relative to the number of parameters, and because they inherently force a fit to the training set (Babyak, 2004). As a result, a single model is suitable only for the specific conditions for which it was developed. With the dual-model approach, parameterization, evaluation, and adjustment (or retuning) become simpler and more robust. Evaluating and retuning the cyanotoxin-surrogate model can be performed without any remotely sensed data, allowing updates through a season; similarly, the surrogate-satellite model can be evaluated and adjusted, if necessary, without the need for cyanotoxin data. Additionally, the uncertainties can be partitioned between models, such that sources of error or confusion can be identified.
Applying a dual-model strategy for cyanotoxins involves several challenges. These include the need for surrogate pigments, the variability in the relationship between cyanotoxins and pigments, and the lack of standardization of the various laboratory and remote sensing measurement methods. This paper will consider these issues in attempting to estimate MCs from satellite data by examining: (1) the issues involved with choosing surrogate pigment, (2) the characteristics of the MC-surrogate relationship, (3) the strengths and limitations of the classes of remote sensing models used for cyanobacterial blooms, and (4) the strategies and considerations for determining whether, and how, to implement the dual models in cyanotoxin mapping. For remote sensing, the emphasis will be on the medium resolution imaging spectrometer (MERIS) sensor and bands, as this sensor has proved the most useful for studies of cyanobacteria; and its replacement, the Ocean Colour Land Imager (OLCI) will be launched in early 2016.
The World Health Organization (WHO, Chorus and Bartram, 1999) provided a widely used set of recommended action levels for risk associated with MC exposure (see also Chorus and Fastner, 2001, Falconer and Humpage, 2005). The WHO provisional guidelines use MC concentration of >10 μg L−1 in recreational waters to indicate additional monitoring is needed to ensure public health protection; many states in the U.S. have developed guidance around the 10 μg L−1 threshold (Graham et al., 2009). When source-water supplies exceed the WHO finished-drinking water guideline of 1 μg L−1, managers often initiate additional monitoring of source water or begin additional treatment of drinking water. Recognizing that chlorophyll is a biomass indicator and more frequently monitored than toxins, Chorus and Bartram (1999) also provided guidance using chlorophyll-a, suggesting that >10 μg L−1 of chlorophyll-a may indicate toxin concentrations that cause mild effects (e.g. contact dermatitis and gastrointestinal upset) during recreational contact. The essential assumption is that high biomass indicates a risk of high cyanotoxin concentration when cyanobacteria dominate the phytoplankton biomass. From a recreational perspective, health departments often provide warnings about avoiding discolored water and cyanobacterial scums (surface accumulations) through posted advisories, web-based sites, and social media (Chorus, 2005, Graham et al., 2009, Ohio EPA, 2015).

Access through your organization

Check access to the full text by signing in through your organization.

Access through your organization

Section snippets

Surrogate identification

A surrogate has to be an indicator of cyanobacterial biomass and reliably quantifiable from satellite data. Chl-a is a common metric for algal biomass and is used as a reference for cyanobacterial blooms (Chorus and Bartram, 1999). Phycocyanins (PC) are recognized as indicators of cyanobacterial presence in eutrophic systems (De Marsac, 2003, Kirk, 1994), although they are also found in rhodophytes and some crytophytes (Wehr and Sheath, 2003). (Another phycobiliprotein, phycoerythrin, is a

Microcystin–surrogate relationships

MCs are not always produced by cyanobacterial blooms, and even when produced, intracellular content may vary several orders of magnitude (Fastner et al., 2001, Ha et al., 2011) in response to various environmental factors. These cyanotoxins consist of a complex of amino acids, which require nitrogen. Consistent with this requirement, cellular production of MC appears to be greatest when nitrogen is abundant (Gobler et al., 2016, Monchamp et al., 2014, Orr and Jones, 1998). The amount of MC also

Modeling pigments from satellite

The second part of the dual model involves estimating the pigments from remotely sensed data. There are several classes of algorithms that can be used to estimate cyanobacterial abundance or biomass from Chl-a or PC. These classes can be described as follows: analytical, semi-analytical, and second derivative (or spectral shape). Analytical approaches (also called inversion algorithms) solve simplified forms of the radiative transfer equation to extract the spectral absorption of the various

Choice of surrogates

The choice of Chl-a or PC as the surrogate depends on several factors, particularly the trade-off between sensitivity and specificity. The balance lies with sensitivity for Chl-a against specificity for PC. For a lake with a spatially-variable mixed bloom of cyanobacteria and eukaryotic algae, PC should be the best surrogate (assuming there are limited cryptophytes or rhodophytes) as Chl-a is non-specific for cyanobacteria. For blooms with biomass dominated by cyanobacteria, Chl-a offers

Analytical uncertainties

Within a region, measurement of any constituent can vary considerably when different analytical methods and approaches are used, especially in the absence of certified reference materials, lack of standardized methods for sampling and processing, and differential response to analytical matrix effects that adversely impact the measurement of analyte concentration whether in the field, laboratory, or by remote sensing. When using multiple measurements, care must be exercised to adequately deal

Parameterization of the dual model

Parameterization of the toxin-surrogate model can be solved in one of several ways, all with different confidence levels. If samples are rare, for example one or two per season, toxin maps should be avoided. They would give the impression of a knowledge that does not exist. Kudela et al. (2015) demonstrated this point effectively in Pinto Lake, California. They found shifts between non-toxic Aphanizomenon, and toxic Microcystis, with large variations in the relationship between MC and pigments.

Conclusions

Cyanotoxins cannot be directly measured with remote sensing. If the variability between cyanotoxins and a surrogate pigment is addressed, a dual-model strategy for remote sensing of these compounds is appropriate. A relationship between MC and either Chl-a or PC can remain constant for days to weeks within a lake. Over longer time intervals, a fixed parameterization may lead to large errors in estimated MC concentrations, therefore the parameterization should be validated every few weeks and

Acknowledgements

This work was partially funded by the NASA Public Health and Water Quality Program (NNH08ZDA001N) under contract NNH09AL53I, the NASA Ocean Biology and Biochemistry Programs under proposal 14-SMDUNSOL14-0001, the U.S. Geological Survey's Toxic Substances Hydrology Program, and the Environmental Protection Agency's Great Lake Research Initiative. This is NOAA GLERL publication number 1804. Steve Ruberg and the captains and crew of NOAA GLERL's research vessels provided invaluable support. Any

References (105)

  • R.S. Lunetta et al.

    Evaluation of cyanobacteria cell count detection derived from MERIS imagery across the eastern USA

    Remote Sens. Environ.

    (2015)
  • M.W. Matthews

    Eutrophication and cyanobacterial blooms in South African inland waters: 10years of MERIS observations

    Remote Sens. Environ.

    (2014)
  • M.W. Matthews et al.

    An algorithm for detecting trophic status (chlorophyll-a), cyanobacterial-dominance, surface scums and floating vegetation in inland and coastal waters

    Remote Sens. Environ.

    (2012)
  • M.W. Matthews et al.

    Improved algorithm for routine monitoring of cyanobacteria and eutrophication in inland and near-coastal waters

    Remote Sens. Environ.

    (2015)
  • S. Mishra et al.

    Quantifying cyanobacterial phycocyanin concentration in turbid productive waters: a quasi-analytical approach

    Remote Sens. Environ.

    (2013)
  • M. Moradi

    Comparison of the efficacy of MODIS and MERIS data for detecting cyanobacterial blooms in the southern Caspian Sea

    Mar. Pollut. Bull.

    (2014)
  • W.J. Moses et al.

    Operational MERIS-based NIR-red algorithms for estimating chlorophyll-a concentrations in coastal waters—the Azov Sea case study

    Remote Sens. Environ.

    (2012)
  • J. O’Neil et al.

    The rise of harmful cyanobacteria blooms: the potential roles of eutrophication and climate change

    Harmful Algae

    (2012)
  • S.C. Palmer et al.

    Validation of Envisat MERIS algorithms for chlorophyll retrieval in a large, turbid and optically-complex shallow lake

    Remote Sens. Environ.

    (2015)
  • S. Palmer et al.

    Satellite remote sensing of phytoplankton phenology in Lake Balaton using 10 years of MERIS observations

    Remote Sens. Environ.

    (2015)
  • L.A. Pearson et al.

    The genetics, biosynthesis and regulation of toxic specialised metabolites in cyanobacteria

    Harmful Algae

    (2016)
  • L. Qi et al.

    A novel MERIS algorithm to derive cyanobacterial phycocyanin pigment concentrations in a eutrophic lake: theoretical basis and practical considerations

    Remote Sens. Environ.

    (2014)
  • W. Reuter et al.

    New trends in photobiology: adaptation of the photosynthetic apparatus of cyanobacteria to light and CO2

    J. Photochem. Photobiol, B: Biol.

    (1993)
  • A. Ruiz-Verdú et al.

    An evaluation of algorithms for the remote sensing of cyanobacterial biomass

    Remote Sens. Environ.

    (2008)
  • G. Sass et al.

    Understanding variation in trophic status of lakes on the Boreal Plain: a 20 year retrospective using Landsat TM imagery

    Remote Sens. Environ.

    (2007)
  • J. Seppälä et al.

    Ship-of-opportunity based phycocyanin fluorescence monitoring of the filamentous cyanobacteria bloom dynamics in the Baltic Sea

    Estuarine Coastal Shelf Sci.

    (2007)
  • S.G. Simis et al.

    Influence of phytoplankton pigment composition on remote sensing of cyanobacterial biomass

    Remote Sens. Environ.

    (2007)
  • R.P. Stumpf et al.

    Satellite detection of bloom and pigment distributions in estuaries

    Remote Sens. Environ.

    (1988)
  • C.C. Trees et al.

    Errors associated with the standard fluorimetric determination of chlorophylls and phaeopigments

    Mar. Chem.

    (1985)
  • R. Al-Ammar et al.

    Optical detection of microcystin produced by cyanobacteria

    J. Phys.: Conf. Ser.

    (2013)
  • M.A. Babyak

    What you see may not be what you get: a brief, nontechnical introduction to overfitting in regression-type models

    Psychosom. Med.

    (2004)
  • M.J. Behrenfeld et al.

    Climate-driven trends in contemporary ocean productivity

    Nature

    (2006)
  • C.E. Binding et al.

    Time series analysis of algal blooms in Lake of the Woods using the MERIS maximum chlorophyll index

    J. Plankton Res.

    (2011)
  • A. Bricaud et al.

    Variability in the chlorophyll-specific absorption coefficients of natural phytoplankton: analysis and parameterization

    J. Geophys. Res.

    (1995)
  • E. Briand et al.

    Evidence of the cost of the production of microcystins by Microcystis aeruginosa under differing light and nitrate environmental conditions

    PLoS ONE

    (2012)
  • R.E. Carlson

    A trophic state index for lakes

    Limnol. Oceanogr.

    (1977)
  • W.W. Carmichael et al.

    Using an enzyme linked immunosorbent assay (ELISA) and a protein phosphatase inhibition assay (PPIA) for the detection of microcystins and nodularins

    Nat. Toxins

    (1999)
  • J.D. Chaffin et al.

    Role of suspended sediments and mixing in reducing photoinhibition in the bloom-forming cyanobacterium Microcystis

    J. Water Resour. Prot.

    (2012)
  • E.I. Chorus et al.

    Toxic Cyanobacteria in Water: A Guide to Their Public Health Consequences, Monitoring and Management

    (1999)
  • I. Chorus et al.

    Recreational exposure to cyanotoxins

    Cyanotoxins, Occurrence, Causes, Consequences

    (2001)
  • E. D'sa et al.

    Estimation of colored dissolved organic matter and salinity fields in case 2 waters using SeaWiFS: examples from Florida Bay and Florida Shelf

    J. Earth Syst. Sci.

    (2002)
  • T.W. Davis et al.

    Effects of nitrogenous compounds and phosphorus on the growth of toxic and non-toxic strains of Microcystis during cyanobacterial blooms

    Aquat. Microb. Ecol.

    (2010)
  • T.W. Davis et al.

    Effects of increasing nitrogen and phosphorus concentrations on the growth and toxicity of Planktothrix blooms in Sandusky Bay

    Lake Erie. Environ. Sci. Technol.

    (2015)
  • T.W. Davis et al.

    Phylogenies of microcystin producing cyanobacteria in the lower Laurentian Great Lakes suggest extensive genetic connectivity

    PLoS ONE

    (2015)
  • N.T. De Marsac

    Phycobiliproteins and phycobilisomes: the early observations

    Photosynth. Res.

    (2003)
  • J.A. Downing et al.

    Predicting cyanobacteria dominance in lakes

    Can. J. Fish. Aquat. Sci.

    (2001)
  • H. Duan et al.

    Are algal blooms occurring later in Lake Taihu? Climate local effects outcompete mitigation prevention

    J. Plankton Res.

    (2014)
  • I.R. Falconer et al.

    Health risk assessment of cyanobacterial (blue-green algal) toxins in drinking water

    Int. J. Environ. Res. Public Health

    (2005)
  • J. Fastner et al.

    Microcystins and hepatocyte toxicity

    Cyanotoxins, Occurrence, Causes, Consequences

    (2001)

    • Cited by (167)

    • ACIX-Aqua: A global assessment of atmospheric correction methods for Landsat-8 and Sentinel-2 over lakes, rivers, and coastal waters

      2021, Remote Sensing of Environment

    • NASA's surface biology and geology designated observable: A perspective on surface imaging algorithms

      2021, Remote Sensing of Environment

    • Remote sensing of cyanobacterial blooms in inland waters: present knowledge and future challenges

      2019, Science Bulletin

    • Satellite monitoring of cyanobacterial harmful algal bloom frequency in recreational waters and drinking water sources

      2017, Ecological Indicators

    • Water Resources in Africa under Global Change: Monitoring Surface Waters from Space

      2023, Surveys in Geophysics

    • Global mapping reveals increase in lacustrine algal blooms over the past decade

      2022, Nature Geoscience
    View all citing articles on Scopus
    View full text
    Published by Elsevier B.V.