Environ. Sci. Technol. 2009, 43, 2627–2633
Using Remote Sensing to Aid the Assessment of Human Health Risks from Blooms of Potentially Toxic Cyanobacteria PETER D. HUNTER,* ANDREW N. TYLER, DAVID J. GILVEAR, AND NIGEL J. WILLBY School of Biological and Environmental Sciences, University of Stirling, Stirling FK9 4LA, Scotland
Received October 22, 2008. Revised manuscript received February 10, 2009. Accepted February 13, 2009.
Mass populations of toxic cyanobacteria in recreational waters can present a serious risk to human health. Intelligence on the abundance and distribution of cyanobacteria is therefore needed to aid risk assessment and management activities. In this paper, we use data from the Compact Airborne Spectrographic Imager-2 (CASI-2) to monitor seasonal change in the concentration of chlorophyll a (Chl a) and the cyanobacterial biomarker pigment C-phycocyanin (C-PC) in a series of shallow lakes in the UK. The World Health Organization guidance levels for cyanobacteria in recreational waters were subsequently used to build a decision tree classification model for cyanobacterial risk assessment which was driven using Chl a and C-PC products derived from the CASI-2 data. The results demonstrate that remote sensing can be used to acquire intelligence on the distribution and abundance of cyanobacteria in inland waterbodies. It is argued the use of remote sensing reconnaissance, in conjunction with in situ based monitoring approaches, would greatly aid the assessment of cyanobacterial risks in inland waters and improve our ability to protect human health.
Introduction Mass populations of cyanobacteria are an increasingly global phenomenon in nutrient-polluted inland waters. These mass populations, which can occur as blooms, scums, or biofilms, can present serious risks to animal and human health because many species are capable of producing a number of highly potent toxins (so-called cyanotoxins) (1, 2). The cyanotoxins recognized to date include the hepatotoxic microcystins (MC) and nodularins (NOD), the cytotoxin cylindrospermopsin (CYN), and neurotoxic agents such as the saxitoxins (STX) and anatoxin-a (ATXa). Moreover, all Gram-negative bacteria, including the cyanobacteria, contain the endotoxin lipopolysaccharide (LPS) (3) and it has also been shown the neurotoxic amino acid beta-methylamino alanine (BMAA) is also widely produced by cyanobacteria (4). The severity of the risks posed by toxic cyanobacteria is now evidenced by an accumulating number of documented incidences of toxicosis in animals and humans resulting from acute or chronic exposure (5-7). In recognition of these risks, the World Health Organization (WHO) has established provisional guidance levels (GLs) * Corresponding author phone: +44 (0)1786 447810; fax: +44 (0)1786 472133; e-mail:
[email protected]. 10.1021/es802977u CCC: $40.75
Published on Web 03/05/2009
2009 American Chemical Society
for cyanobacterial toxins in recreational waters (Supporting Information Table S1) (8). These advisory guidelines provide an estimate of the risk of adverse health outcomes if accidental or incidental recreational exposure was to occur and also offer recommendations on appropriate management interventions (from warnings to access restrictions) at each level of risk. In addition, as cyanotoxins are not routinely monitored in recreational waters, equivalent GLs for cyanobacterial cells are also provided to facilitate risk assessment activities within theframeworkofexistingwaterqualitymonitoringprogrammes. However, the recognition of cyanobacterial hazards in the field is problematic and is typically limited to crude visual assessments of gross water discoloration or the identification of floating scums and mats (5). More quantitative monitoring is reliant on the identification and enumeration of cells by microscopic examination or the analysis of diagnostic pigments analysis by high performance liquid chromatography in a laboratory. The use of flow cytometry and spectrofluorescence (with specificity for diagnostic cyanobacterial pigments), or genomics-based techniques (e.g., quantitative real-time PCR), may improve our ability to recognize cyanobacterial hazards in the field in the future. Nevertheless, such approaches are ill-suited for monitoring large numbers of waterbodies at regional or national scales and, as cyanobacterial blooms and scums are patchy and transient, such point-based-monitoring approaches may frequently underestimate actual cyanobacterial abundance (9, 10). It has been widely shown that phytoplankton blooms in inland waters can be monitored using remote sensing through the use of algorithms for the retrieval of chlorophyll a (Chl a) concentrations (9, 11, 12). However, as Chl a is common to almost all phytoplankton, its retrieval from remotely sensed data cannot be used to specifically determine the abundance of cyanobacteria where other groups of eukaryotic algal cooccur. In contrast to Chl a, C-phycocyanin (C-PC) is a pigment only found in high concentrations in freshwater cyanobacteria. The optical properties of C-PC-containing cyanobacteria are distinct from those of the eukaryotic algae because C-PC has a pronounced absorption feature near 620 nm (13, 14). This allows targeted algorithms to be developed for the retrieval of C-PC concentrations from remotely sensed data (15-18). In waters with low C-PC:Chl a ratios, it has been shown that absorption by Chl a (or other accessory pigments) near to 620 nm can affect the accuracy of C-PC retrieval, particularly when using semiempirical algorithms that cannot correct for such effects (17, 19). However, the algorithms published to date have generally been shown to perform more than satisfactorily at those concentrations where cyanobacteria may pose risks to human health. This suggests remote sensing may be able to make a significant contribution to cyanobacterial hazard identification and risk assessment. In this paper, we use data from an airborne spectrographic imager to monitor seasonal change in the concentration of Chl a and C-PC in a series of shallow lakes used heavily for recreational activities. The remotely sensed Chl a and C-PC products were subsequently used as inputs into a risk assessment model based upon the WHO GLs for recreational waterbodies and used to estimate the potential risk to human health. We argue that, in the absence of information on the presence or concentration of toxins, such an approach could be used to provide regulatory agencies with rapid intelligence on the abundance and distribution of cyanobacteria in recreational waters and to aid hazard identification and risk management activities for public health protection. VOL. 43, NO. 7, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2627
Materials and Methods Study Sites. The Norfolk Broads are a series of small (0.001-1.3 km2) and shallow lakes (mean depth June > August). This was most likely the result of the VOL. 43, NO. 7, 2009 / ENVIRONMENTAL SCIENCE & TECHNOLOGY
9
2629
disproportionate contribution of Chl a and other accessory pigments (such as Chl b) to absorption at the C-PC absorption maximum (620 nm) (17, 19). In similarity to the findings of refs 17-19, the highest errors occurred during the survey on 21 April 2005 when the C-PC:Chl a ratio was generally
