For Observing Harmful Algal Blooms

Home , Algal bloom, Nanophytoplankton, Particle (ecology)

HARMFUL ALGAL BLOOMS

BY MARCEL BABIN, JOHN J. CULLEN, COLLIN S. ROESLER, PERCY L. DONAGHAY, GREGORY J. DOUCETTE, MATI KAHRU, MARLON R. LEWIS, CHRISTOPHER A. SCHOLIN, New MICHAEL E. SIERACKI, AND HEIDI M. SOSIK Approaches and Technologies for Observing Harmful Algal Blooms

Th is article has been published in Oceanography, Volume 18, Number 2, a quarterly journal of Th e Oceanography Society. Copyright 2005 by Th e Oceanography Society. All rights reserved. Reproduction of any portion of this article by photo- 210 Oceanography Vol.18, No.2, June 2005 copy machine, reposting, or other means without prior authorization of Th e Oceanography Society is strictly prohibited. Send all correspondence to: [email protected] or Th e Oceanography Society, PO Box 1931, Rockville, MD 20849-1931, USA. Harmful algal blooms (HABs) represent a diverse of phytoplankton. Also, in situ analysis of nutri- range of phenomena that universally share only two ent concentrations is now a reality, complementing Hcharacteristics: they produce effects on ecosystems established methods for measuring temperature, or food resources that humans perceive as harmful, salinity, and oxygen concentrations with submers- and their progression is fundamentally a process ible sensors. of population dynamics under oceanographic con- Many of the new approaches and technologies trol. Because of the complexity, scales, and transient are unfamiliar to potential users, therefore, they are nature of HABs, their monitoring and prediction not immediately useful to them. Hence, we orga- requires rapid, intensive, extensive, and sustained nized a “Workshop on Real-Time Coastal Observ- observations at sea. These requirements cannot ing Systems for Ecosystem Dynamics and Harmful be met with traditional approaches that depend Algal Blooms” in Villefranche-sur-Mer (France), on ships for sampling and laboratories for chemi- held 11-21 June 2003, to review the new technolo- cal or biological analyses. Fortunately, new sensing gies and platforms appropriate for autonomous technologies that operate autonomously in situ will and in situ observation of HABs. We provided the allow, in the near future, the development of com- participants with both the theory relevant to un- prehensive observation strategies for timely detec- derstanding the basic principles of real-time ob- tion of HABs. In turn, developments in modeling servation and modeling tools, and tutorials to ex- will support prediction of these phenomena, based plain the use of these tools. The proceedings of this directly on real-time measurements. workshop (“Habwatch”) appear in two forms. First, Enabling ocean-observation technologies that all invited lectures and contributed talks, and most have emerged during the last decade refl ect advanc- posters, are available on the workshop web site es in optical imaging, molecular biology, chemistry, (www.obs-vlfr.fr/habwatch); oral talks are available acoustics, phytoplankton physiology, and marine as slide shows with voice. Second, chapters based on optics. Now, sensors are increasingly deployed from invited lectures will be published in a peer-reviewed various platforms such as in situ profi lers, autono- volume of the UNESCO series Monographs on mous underwater vehicles (AUVs), and moorings Oceanographic Methodology (Babin et al., in press). to derive a broad range of quantitative and qualita- Here, we present an overview of the major ob- tive information about the pelagic environment. servational technologies described during the For example, optical sensing, used for decades to Habwatch workshop, referring the reader to de- measure photosynthetically available radiation and tailed reviews on specifi c topics, which form the to estimate concentrations of suspended particles chapters of the Habwatch proceedings. We con- from turbidity or phytoplankton from chlorophyll clude by suggesting some priorities for the devel- fl uorescence, has been greatly enhanced to provide opment of observational tools within the frame- much better measures of the light fi eld and sea- work of the Global Ecology and Oceanography of water constituents, plus indicators of the species Harmful Algal Blooms (GEOHAB) program (see composition, cell size, and physiological properties www.geohab.info/).

Oceanography Vol.18, No.2, June 2005 211 AVAILABLE AND EMERGING 1998). This instrument measures light among HAB species. Thus, unique ge- OBSERVATIONAL TECHNOLOGIES scattering, and fl uorescence from chlo- netic signatures (DNA, RNA) targeted by Many technologies show promise for rophyll and phycoerythrin, on individual molecular probes have been increasingly real-time observation of HABs and the particles larger than 5 µm. Its CCD cam- used for detection of microorganisms at oceanographic processes that in large era is triggered by, for instance, the chlo- the species level, including the discrimi- part determine their dynamics. Some rophyll a fl uorescence signal; digital sil- nation of potentially toxic from non- enhance existing capabilities while others houette images are stored on disk. Data toxic species of a genus (e.g., Pseudo- can provide information in situ that until analysis allows enumeration of targeted nitzschia spp.). However, simply the pres- recently could be obtained only through morphotypes with semi-automatic shape ence of a toxigenic species provides no hands-on analysis of samples. recognition. This particle analyzer is es- information on the associated toxicity, as pecially good in the microplankton size toxin levels can vary widely as a function Detection of Phytoplankton at the range (20 to 200+µm) where morpholo- of environmental effects on the organ- Group or Species Level gy is an important identifi cation feature. ism’s physiology. Ideally, the potential im- The harmful effects of HABs are general- Benchtop, portable, and submersible pacts of a toxic HAB should be assessed ly attributable to single species. It is thus versions are available. Jaffe (in press) re- necessary to discriminate phytoplankton views other optical imagers that could be Marcel Babin ([email protected]) is at the group or species level for the de- used in HAB research and monitoring. Research Scientist, Laboratoire d’Océan- tection, early warning, and oceanograph- Sensitive to cells smaller than Flow- ographie de Villefranche, CNRS/UPMC, ic characterization of HABs. This is tra- CAM’s limit, fl ow cytometers can be Villefranche-sur-Mer Cedex, France. John J. ditionally done by microscopic examina- used to discriminate phytoplankton Cullen is Professor, Department of Ocean- tion or analysis for toxins, both of which groups based on pigment fl uorescence ography, Dalhousie University, Halifax, are laborious and generally performed in and light scattering as an index of size. Nova Scotia, Canada. Collin S. Roesler is the laboratory; these approaches are not Automated submersible fl ow cytometers Research Scientist, Bigelow Laboratory for suitable for real-time observation. There have been developed and deployed suc- Ocean Sciences, West Boothbay Harbor, are, however, new and emerging tech- cessfully to obtain signifi cant results on ME, USA. Percy L. Donaghay is Research nologies derived from those traditional phytoplankton population dynamics Scientist, Graduate School of Oceanogra- methods that allow autonomous and/or (Figure 2). With sample preparation, phy, University of Rhode Island, Narra- gansett, RI, USA. Gregory J. Doucette is Research Scientist, NOAA/National Ocean Service, Charleston, SC, USA. Mati Kahru Many technologies show promise for is Research Scientist, Scripps Institution of real-time observation of HABs and the Oceanography, UCSD, La Jolla, CA, USA. oceanographic processes that in large Marlon R. Lewis is Professor, Department of Oceanography, Dalhousie University, part determine their dynamics. Halifax, Nova Scotia, Canada. Christopher A. Scholin is Associate Scientist, Monterey Bay Aquarium Research Institute, Moss near real-time detection of phytoplank- fl ow cytometers can identify cells labeled Landing, CA, USA. Michael E. Sieracki is ton at the species and group levels. One with specifi c markers (e.g., for proteins, Research Scientist, Bigelow Laboratory for such commercial technology is Flow- bulk DNA, or gene probes). Ocean Sciences, West Boothbay Harbor, ME, CAM, a particle analyzer and imager In many cases, phenotypic attributes USA. Heidi M. Sosik is Associate Scientist, that combines aspects of microscopy and of cells (e.g., morphology, pigment com- Biology Department, Woods Hole Oceano- fl ow cytometry (Figure 1) (Sieracki et al., position) are insuffi cient to discriminate graphic Institution, Woods Hole, MA, USA.

212 Oceanography Vol.18, No.2, June 2005 A B

Figure 1. Schematic diagrams and results from the FlowCAM imaging-in-fl ow system (Fluid Imaging Technologies, USA; www.fl uidimaging.com). Sample fl uid is pulled through the fl ow cell at rates up to 4 mL/min. (A) When a cell enters the fi eld of view, fl uorescence is excited by a fan of laser light. Epifl uorescence optics are used that em- ploy dichroic mirrors (DM) to split the light signals. Diff erent objectives (OBJ) can be used for diff erent magnifi cations (e.g., 4X to 20X). Th e fl uorescence signal is detected by photomultiplier tubes (PMT) and triggers a light emitting diode (LED) to fl ash and backlight the fl ow cell. Th e camera image is digitized by the frame grabber. Th e cell is located within the image and measured, and a subimage is stored on disk in real time. Particles can also be detected by light scatter (detector not shown for simplicity). (B) Details of the fi eld of view of the fl ow cell. (C) Th ese images are linked to the size and fl uorescence values and results can be browsed using the interactive scattergram. Th e dot plot shows fl uorescence versus size and populations can be investigated by se- lecting points and the corresponding images are displayed.

Oceanography Vol.18, No.2, June 2005 213 x 105 106 6 Synechococcus C

A B -1 Picoeukaryotes CryptophytesCryptophytes 4 Synechococcus 4 2 10 Cells ml

0 x 105 Nano- 3 D 2 Nano- Nanoeukaryotes 10 -1 eukaryotes 2 Cryptophytes (x 50) Red Flourescence eukaryotes

1 Picoeukaryotes Cells ml 100 0 100 102 104 106 01 Aug 15 Aug 01 Sep 15 Sep Side Scattering

Figure 2. FlowCytobot (A), an automated submersible fl ow cytometer designed and built at the Woods Hole Oceanographic Institution (Olson et al., 2003), has been successfully deployed at coastal cabled observatory sites on the U.S. East Coast. FlowCytobot makes nearly continuous measurements of pico/nanophytoplankton abundance, cell scattering, and cell fl uorescence characteristics (B, typical data from 1 hour of sampling; classifi cation of cells considers all measured properties including orange fl uorescence, critical for discrimination of Synechococcus and cryptophytes). A two-month deployment at the Long-Term Environ- mental Observatory off New Jersey in 2001 reveals a high level of temporal variability in abundance of Synechococcus, cryptophytes and other pico- and nanophytoplankton (C, D), some of which can be explained on the basis of changes in population growth rates inferred from FlowCytobot-based observations of diel changes in cell size distributions (Sosik et al., 2003). An autonomous fl ow cytometer is now commercially available in portable, fl oating, and submersible versions (CytoBuoy b.v., Th e Netherlands; http://www.cytobuoy.com).

with integrated detection of organisms proaches such as molecular imprinting bidometers on a variety of platforms and toxins using in situ sensors in real- and in situ mass spectrometry also show can continuously detect phytoplankton or near-real time. Toward this end, the promise for autonomous detection of or particles on the vertical and horizon- Environmental Sample Processor (ESP) toxins (Scholin et al., in press). tal scales of the physical and chemical has been developed (Figures 3 and 4). processes that infl uence HAB dynamics. A prototype of this autonomous, in situ Improved Detection of However, the optical sensors in wide use system has performed DNA probe array Phytoplankton Biomass and other measure bulk properties of the water analyses when deployed at sea (Scholin et Bulk Properties of Surface Waters and do not allow much discrimination al., in press). An ELISA-based (enzyme- The new technologies for detection of beyond measures of algal biomass or linked immunosorbent assay) toxin ar- phytoplankton at the group and species particle load. This shortcoming is largely ray deployable on this same autonomous levels are not presently suited for syn- compensated by the amount of informa- sampling platform can detect Pseudo- optic or highly resolved sampling on the tion generated over a broad spectrum of nitzschia’s toxin, domoic acid. The toxin scales of HABs. Fortunately, such spe- temporal and spatial scales. New optical array has undergone initial laboratory cialized measurements can be comple- sensors described at the Habwatch work- testing (Figure 4) and will soon undergo mented with synoptic remote sensing shop can provide much more informa- fi eld trials. Once operational, the next- and continuous, in situ deployments of tion on the same scales. They measure generation ESP will be capable of provid- other instruments to provide essential a range of optical properties that can be ing concurrent data on the abundance oceanographic context. For example, related directly to phytoplankton and of a HAB species and its toxin. Other ap- fl uorometers, transmissometers, or tur- other constituents of the water.

214 Oceanography Vol.18, No.2, June 2005 Figure 3. Use of the Environmental Sample Processor (ESP) to detect phytoplankton species in near real-time and to archive al- iquots of the same samples for fl uorescent in situ hybridization (FISH) and toxin analyses (for details see Scholin et al. in press; http://www.mbari.org/microbial/ESP). Th ese examples show how a given sample can be interrogated for HAB species and toxins using a variety of molecular analytical techniques. For probe arrays, the ESP concentrated phytoplankton, homogenized that material, and passed the homogenate over a custom oligonucleotide probe array to reveal ribosomal RNA (rRNA) molecules indica- tive of a variety of HAB species (e.g., Figure 4). An image of the resulting array was recorded using a CCD camera, providing a near real-time indication of presence and abundance of selected species. For FISH, a sample corresponding to that used to develop the array was collected, preserved, and stored onboard the ESP. Later, the ESP was programmed to process that archived sample with any one of a number of fl uorescently labelled probes. Once the hybridization process was completed, the sample fi lter was recov- ered and viewed using conventional epifl uorescence microscopy. Finally, the ESP was used to archive samples for determinations of either domoic acid or saxitoxin activity. In the examples presented here, images of the probe arrays are shown in the top row (frames 1 to 4) and corresponding micrographs showing results of FISH assays are shown in the bottom two rows (frames 1a to 4a). Locations of probes on the arrays for Pseudo-nitzschia australis and Alexandrium catenella, spotted in triplicate, are denoted “P. aus” (labels rRNA specifi c to P. australis) and “A. cat” (labels rRNA specifi c to A. catenella), respectively. Diff erent rRNA probes were used for FISH analysis depending on the sample: “Univ.” (targets rRNA from all organisms), as well as fl uorescent versions of “P. aus” and “A. cat” similar to those on the arrays. Domoic acid or saxitoxin activity associated with a sample was determined using a re- ceptor binding technique, and values obtained are printed at the bottom of the respective micrographs.

One group of commercially available, in seawater: the coefﬁ cients of absorp- organic matter (CDOM) have different in situ optical instruments uses carefully tion, attenuation, scattering, and back- optical properties, spectra of absorp- designed light sources and detectors to scattering, at an increasing number of tion and scattering can be deconvolved measure the inherent optical proper- wavelengths. Because phytoplankton, to retrieve estimates of their respective ties (IOPs) of the substances contained non-algal particles, and colored dissolved concentrations (Roesler and Boss, in

Oceanography Vol.18, No.2, June 2005 215 press; Sosik, in press). Sometimes, species tral instruments. They can be deployed tential for using IOPs in the detection composition can be retrieved. It has been on virtually all in situ platforms (e.g., and study of HABs is huge: fulfi llment shown by Kirkpatrick et al. (2000) that towed vehicles, moorings, and AUVs) of this potential will require further de- the toxic dinofl agellate Karenia brevis and provide data at various spatial velopment of theory and measurement can be detected by analyzing the shape of and temporal scales, depending on the technology, but more importantly the the phytoplankton absorption spectrum, platform (Chang and Dickey, in press; education of a broader community in measured in situ with a fl ow-through Griffi ths, in press). An example of high- relevant principles of bio-optics. spectrophotometer, in comparison with a frequency and high-resolution vertical A second group of optical instruments known absorption spectrum for the spe- profi les of the absorption coeffi cient includes passive sensors that measure ir- cies in question (Schofi eld et al., in press). is shown in Figure 5, from a study of radiance or radiance, and thereby the ap- Submersible IOP sensors now exist the dynamics of thin layers formed by parent optical properties (AOPs) of sea- in different formats, ranging from bulk Pseudo-nitzschia spp. in East Sound, WA water, namely the vertical diffuse attenu- hyperspectral to miniature multispec- and Monterey Bay, CA (USA). The po- ation coeffi cient (Kd) and water refl ectance (R; the ratio of upward radiance or irradiance to downward irradiance) at different wavelengths from the UV to the near-infrared. Because they depend on the sun for illumination rather than on internal optics, AOP sensors cannot Figure 4. A)-(C). Cus- measure spectral absorption or scatter tom DNA probe arrays printed on fi lters and directly, or at night. But the attenua- developed autono- tion coeffi cient is strongly a function of mously on board the absorption, and refl ectance depends on Environmental Sam- ple Processor. Images the ratio of backscatter to absorption, so show target organisms the constituents of the water can be re- and corresponding trieved from AOPs. AOP inversion mod- array image (Scholin, unpublished; images els discriminate phytoplankton from of cells kindly provided CDOM and non-algal particles, and may by P. Miller and Y. even identify pigment-based taxonomic Fukuyo). (D). Proto- type domoic acid (DA) groups if the sensors are hyperspectral toxin array printed on (Figures 6 and 7). Like the IOP decon- a fi lter and showing binding of DA-specifi c volutions, these models are based on a antibody to toxin-pro- forward model with assumptions about tein conjugate fol- the spectral shapes of IOPs for the dif- lowing conduct of a competitive ELISA (G. ferent constituents. But, the inversion Doucette, C. Mikulski, is more complex since it is necessary to and C. Scholin, unpub- fi rst derive IOPs from AOPs. This is an- lished). other reason why the use of new optical approaches for monitoring and research requires some knowledge of bio-optics. Radiometric sensors that measure AOPs are calibrated in an absolute sense against international standards; mea-

216 Oceanography Vol.18, No.2, June 2005 Figure 5. Th e Ocean Response Coastal Analysis System (ORCAS) autonomous bottom-up profi ler used to collect a week-long time series of vertical profi les in northeastern Monterey Bay, CA. Th e profi ler design and picture are illustrated at the top of the fi gure along with 2 of the 187 sequential vertical profi les of absorption at 440 nm (the wavelength of peak absorption by chlorophyll a) (red circles) and density (black dots). Th is thin layer was dominated by Pseudo-nitzschia and persisted for 5 days. Th e profi ler uses a positively buoyant sensor package and a small underwater winch to profi le fi ne scale physical, chemical, and optical structure from the bottom up. Th ese profi lers, co-developed with WET Labs (www.wetlabs.com), are fully self-contained with onboard microprocessors, controllers, batteries, and radio communication systems. Th ey can characterize vertical structure on the scale of cm using sensors for spectral absorption, spectral attenuation, spectral scatter, chlorophyll a fl uorescence, optical backscatter, and mechanically stimulated bioluminescence.

Oceanography Vol.18, No.2, June 2005 217 0.04 0.06 Diatom surements of irradiance or radiance tak- 0.03 0.010 0.04 en at one location or time can be quan- 0.02 titatively compared to measurements 0.005 0.02 0.01 taken elsewhere or through long time series. AOP sensors are therefore especially 0 0.000 0 appropriate for extensive and sustained 0.04 0.04 Dinoﬂagellate monitoring, and for applications with 0.010 0.03 0.03 diverse participants, such as the Global

0.02 0.02 Ocean Observing System (GOOS). Ad- 0.005 0.01 0.01 ditionally, quantities such Kd and R can be derived without absolute calibration, 0 0.000 0 providing the opportunity for develop- 0.04 0.8 Dinophysis ment of low-cost applications. 0.010 0.03 0.6 AOP sensors can be placed on many

0.02 0.4 different platforms, although their de- 0.005 0.01 0.2 ployment needs some care regarding measurement geometry (Morel and Lew- 0 0.000 0 is, in press). One of the most spectacular 0.04 0.6 Mesodinium approaches for AOP measurement is 0.010 0.03 ocean color remote sensing from space. 0.4 0.02 There are now several operational sensors 0.005 0.2 0.01 in fl ight providing global and recurrent coverage, with spatial and spectral reso- 0 0.000 0 lution that has progressively improved 0.04 0.4 with newer sensors. Ocean color remote 0.010 Chlorophyte 0.03 0.3 sensing has been successfully used, in a 0.02 0.005 0.2 limited number of cases, to detect HABs 0.01 0.1 (Ruddick et al., in press; see the examples 0 0.000 in Figure 8). Current limitations of this 400 600 800 0 13 20 27 Wavelength (nm) technique are related to atmospheric cor- rections and interpretation of the signal Figure 6. Example of taxonomic identifi cation from a hyperspectral surface refl ectance time series collected during an extensive red tide bloom off the west coast of South Africa in the in coastal waters where interference from Benguela upwelling zone in March 2001. Left panel: Sample refl ectance spectra of pre-HAB CDOM and suspended sediment can conditions (black curve, left axis) and near-monospecifi c conditions (colored curve, right axis) observed over the 17-day time series. Right panel: Time series of phytoplankton absorp- confound conventional algorithms. tion coeffi cients (676 nm, m-1) for fi ve separable taxonomic groups derived from inversion of The optical signal of chlorophyll a hyperspectral refl ectance (solid symbols, Roesler and Boss, 2003) and computed from mi- in vivo fl uorescence, stimulated by sun- croscopic cell counts, cell size, and cellular absorption effi ciency (Roesler et al., 2003). X-axis is date in March 2001, correlation coeffi cients between refl ectance- and microscopy-derived light, is another powerful tool for detect- estimates are from top to bottom: 0.63, 0.90, 0.62, 0.84, and 0.93, respectively. ing phytoplankton in seawater despite the large variability of the fl uorescence quantum yield always observed in the natural environment (Babin, in press).

218 Oceanography Vol.18, No.2, June 2005 Figure 7. Detection of bloom dynamics with measurements of light attenuation. (A) A moored chain of upward-looking irradiance sensors (490 nm K-chain) was used to estimate the attenuation of irradiance as a function of depth in a coastal embayment (Bedford Basin). An intense, but ephemeral, sub-surface bloom of dinoﬂ agellates was observed; the surface manifestation was much weaker. Moored K-chains are thus useful for detecting intense blooms, even below the surface (M. Lewis and J. Cullen). (B) Th e Marine Environ- mental Prediction System-Bay (www.cmep.ca/ bay) has three moorings in Lunenburg Bay, Nova Scotia with salinity-temperature chains, current meters, and meteorological and optical sensors, including spectral K-chains; a data assimilation model is being developed to incorporate these data and other local observations into a real-time, coupled atmosphere-ocean simulation of the bay (photo by P. Kuhn). (C) Nearly continuous measurements of hyperspectral ocean color were analyzed by C. Brown and Y. Huot (unpubl.) with an inverse model, generating estimates of phytoplankton absorption (black dots) corrected for the substantial contribution of CDOM and other constituents of the water. Th e black line is a lo- cally weighted least squares regression to indicate trends. Blue symbols show direct measurements of phytoplankton absorption (ﬁ lter pad method, corrected for detritus) and open red symbols are determinations of extracted chlorophyll. Large blooms were absent and absorption by phytoplankton at 490 nm was much less than that by CDOM, so subsurface gradients of phytoplankton were masked by CDOM.

Oceanography Vol.18, No.2, June 2005 219 Figure 8. A harmful phytoplankton bloom dominated by the dinofl a g e l l a t e Gymnodinium sanguineum in Paracas Bay, Peru in April 2004 caused estimated economic damage of US$28.5 million. While standard ocean color products of SeaWiFS and MODIS satellites were of little use in this case due to insuffi cient resolution and problems in atmospheric correction and radiance inversion, MODIS medium-resolution bands provided valuable information with empirical processing algorithms (Kahru et al., 2004). Th e image, repub- lished from Eos, Trans. AGU. 85: 465-472, shows the application of two empirical products in monitoring of the devastating bloom in Paracas Bay. Th e left column shows that the true-color (red-green- blue) images using, respectively, MODIS bands 1 (red), 4 (green), and 3 (blue) can clearly identify the distribution of the bloom in the bay by its conspicuous bright color. Th e right column shows the turbidity index, a semi-quantitative measure of the amount of par- ticulate material in the near-surface water. Darker areas show higher turbidity. Julian day is shown for the true-color images, and the corresponding date (month/day/year) is shown for the turbidity images. While turbidity is not specifi c to algal blooms, it is a quantitative estimate of the intensity of the bloom once the existence of the bloom is detected by the true-color images. During the rise and fall of the bloom in the bay, turbidity was inversely correlated with oxygen concentration. Oxygen depletion caused most of the damage to the benthic communities. Th e top panel (A) shows the bloom in the increasing phase, panel (B) shows the maximum extent of the bloom, panel (D) shows the decreasing phase of the bloom, and the bottom panel (E) shows the normal conditions after the bloom.

More traditionally, fl uorometers with internal light sources are used to detect phytoplankton. New applications have been developed. In situ spectrofl uorometry provides a direct measurement of the shape of the phytoplankton absorption spectrum and allows discrimination of phycobilin-containing phytoplankton groups (mostly cyanobacteria and cryptophytes) from others. A more sophis- ticated physiological analysis of chlorophyll a fl uorescence from a special class of fl uorometer provides quantitative information on the status of the photo- synthetic apparatus (e.g., Kolber et al., 1998), and can be used to estimate pri-

220 Oceanography Vol.18, No.2, June 2005 mary production (Kolber and Falkowski, cally extend their sensitivity down into should be observed (Chang and Dickey, 1993), vastly increasing the information the nanomolar range where nutrient in press) and installation of the appro- potential from in situ fl uorometry. concentrations may affect competition priate sensors on platforms that can ob- between species. serve these scales. There is now a wide Nutrient Sensors array of platforms appropriate for the In many cases, nutrients play a central BIOFOULING observation of HABs at various scales, role in controlling population and com- Technology has its limits, and the ocean for instance, airplanes, satellites, moor- munity dynamics of HABs. This is, of often defi nes them. All in situ sensors are ings, bottom-up profi lers, towed vehicles, course, especially true for eutrophica- sensitive to biofouling, which can range powered AUVs, gliders, various kinds tion, but vertical nutrient structures may from a thin organic fi lm to heavy incrus- of fl oats, and ships of opportunity (see play important roles in the development tation with a community of microbes, Chang and Dickey, in press; Griffi ths, in of HABs that occur as vertically migrat- plants, and animals. Depending on the press). Some of the cheapest and most ing populations or subsurface thin layers. application and environment, sensors accessible platforms are ships of oppor- Submersible chemical analyzers using can be rendered useless by fouling in a tunity, including ferries. Their use for wet chemistry have recently been com- few days, or they may operate for many sustained HAB monitoring has proven mercialized. They can measure nutri- months without signifi cant problems. to be valuable, for instance, in the Baltic ent concentrations continuously in situ Any detecting surface or transducer can Sea (see www.itameriportaali.fi ). None- and thus can provide data on vertical be compromised by fouling, but optical theless, to describe the different relevant or horizontal nutrient structure. Com- sensors are especially sensitive because scales of HABs and the processes that mercial versions of these systems can be the validity of their calibration relies infl uence them, it is necessary to develop deployed from ships for vertical profi ling upon the cleanliness of the windows. an observation program combining the or on towed bodies and AUVs for spa- Heavy fouling alters the chemical envi- use of multiple platforms in strategic de- tial mapping (reviewed by Hanson and ronment near a sensor, so measurements ployments suited for the HAB phenom- Moore, 2001). The submersible chemi- of nutrients or oxygen are not immune. ena of interest. cal analyzers can measure nitrate, ni- Many remedies have been developed and trite, phosphate, silicate, and iron. Issues some are quite successful (see Lehaître et STRATEGIES FOR OBSERVING limiting the application of these in HAB al., in press; Chang and Dickey, in press), DIFFERENT TYPES OF HAB research include ease of use by non- but the intense biofouling in rich coastal The scales of variability for HABs are chemists, storage of reagents, calibra- environments still prohibits long-term among the most important factors to tion, length of deployment, and power deployments without maintenance. consider when designing an observation consumption. An alternative new type of strategy (Chang and Dickey, in press). commercial nutrient sensor uses absorp- PLATFORMS FOR INSTRUMENT The important spatial scales relate not tion in the UV to measure nitrate con- SYSTEMS only to the distributions of target species centration (Johnson and Coletti, 2002). Having the tools for observing oceano- in horizontal and vertical dimensions Although this nitrate sensor is easier graphic phenomena is one thing, but (for which taxonomic identifi cation and to use and calibrate, and has excellent using them effectively is another. As we physiological characterization would long-term stability, it cannot be used to have shown, a new generation of sen- be important), but also to the physi- measure other types of nutrients. More sors is now available for autonomous cal, chemical, and community processes research is needed to adapt autonomous deployment in ocean and estuarine that affect the population dynamics of nutrient analyzers to a variety of in situ observing systems targeted at charac- the HAB species, including the interac- platforms and to resolve the above issues terization of HABs and the factors that tions between behavior (sinking, fl oat- that currently limit their application. infl uence them. Effective use of sensors ing, swimming) and circulation (Franks, More work is also needed to dramati- requires clear defi nition of the scales that 1997, and in press). The physical pro-

Oceanography Vol.18, No.2, June 2005 221 cesses that can affect the spatial-scale rapidly, it will never be possible to mea- Hypotheses about bloom dynamics focus characteristics of HABs include turbu- sure everything continuously and syn- on the processes of initiation, including lence, waves, wind-driven vertical mix- optically. Observation strategies must transport of populations from offshore, ing, convection, light distribution, tides, be designed to make the most of limited and interactions of populations with thermohaline vertical gradients, fronts, resources. As proposed by Cullen (in surface circulation during the progres- eddies, bathymetry-related circulation, press), a coarse classifi cation of HABs sion of a bloom. climate oscillations, and climate change. can be useful as an initial guide to iden- Many environmental properties must The full spectrum of physical processes tify relevant scales and appropriate ob- be measured for effective early warning, covers scales of micrometers to thou- servation strategies for local or regional monitoring, and prediction of blooms sands of kilometers, but some of these observation programs. The classifi cation progressing along a coast. For early warn- processes are more signifi cant for given is summarized here, with suggestions for ing, species- or group-level observation HAB species and these must be identi- observation strategies. technologies are generally necessary (Fig- fi ed when designing a targeted observa- ures 1 to 4, Figure 6). Once the species tion strategy. The same is true for chemi- Widespread HABs Dominated by forming a bloom is known, and when cal (e.g., nutrient concentration and One Species: Extensive, Progressive conditions permit, remote sensing from ratios, allelopathy) and community (e.g., Coastal Blooms satellites and aircraft can provide key grazing) processes. There are several examples of nearly information on distributions and trans- A minimum temporal resolution of monospecifi c and often toxic, extensive port of biomass (Ruddick et al., in press; sampling corresponds to each of the blooms that appear in coastal waters Stumpf et al., 2003), especially when above-mentioned physical, chemical, and and progress along the shoreline. Some supplemented by observation networks community processes. In addition, physi- examples include blooms of Karenia that include direct sampling, for instance, ological mechanisms such as growth, brevis in the Gulf of Mexico, Alexan- for microscopic examination (Figure 7) (Johnsen et al., 1997). Even if surface distributions of developed blooms are resolved with remote sensing, early stages Technical advances are rapidly transforming and subsurface distributions signatures oceanography, but technology alone cannot solve must be described by other means. In pressing environmental challenges such as the need particular, vertical distributions of phytoplankton should be well resolved because to detect and predict harmful algal blooms. the interaction of swimming, sinking, or fl oating with frontal features, aggrega- tion of seed populations in subsurface diel vertical migration, circadian clocks drium catenella off the western coast of layers near the pycnocline, and changes and the cell cycle, acclimation processes, South Africa in 2002, Karenia mikimo- of behavior in mixed waters landward of and life-cycle events such as encystment toi (formerly Gyrodinium aureloum or a front (possibly associated with nutri- and excystment, can strongly affect the Gymnodinium mikimotoi) in northern tion) all may be important in initiation, dynamics of a HAB. Consequently, the European shelf waters, the toxic bloom maintenance, and transport of extensive, spectrum of temporal scales relevant to of Chrysochromulina polylepis in Scan- progressive, coastal blooms (Donaghay HABs ranges from a second or so to hun- dinavian waters in 1988, blooms of Het- and Osborn, 1997; Cowles, 2003). Con- dreds of years; however, as it is the case erosigma in the Strait of Georgia and sequently, for early warning and moni- for spatial scales, some of them are more adjacent waters in western Canada, and toring, observation systems must resolve signifi cant for given HAB phenomena. the dramatic bloom of Karenia digitata vertical distributions of phytoplankton in Although technology is advancing in Hong Kong waters during April 1998. relationship to temperature, salinity, and

222 Oceanography Vol.18, No.2, June 2005 currents, and they must have the means fects include nitrogen enrichment as well that must be observed to understand the to detect target species (and, optimally as toxicity upon landfall. Blooms of Pha- local scale of interest; Intergovernmen- their toxins) in situ. Because nutrient eocystis in the North Sea are infl uenced tal Oceanographic Commission [IOC], availability can infl uence toxicity and by nutrient loading from rivers; they can 2003) some locations experience recur- depletion of nutrients can terminate a deliver prodigious quantities of noxious rent, though not necessarily predictable, bloom, the nutrient regime should also foams to beaches, with signifi cant eco- HABs, while other nearby locations may be assessed as part of monitoring and nomic impact. Finally, China’s Bohai is be spared. Even though localized HABs modeling strategies. strongly affected by widespread HABs, are likely related to larger-scale forcings, Progressive coastal blooms move sometimes clearly detectable from space. local conditions must have a strong in- with coastal currents and can appear or For extensive blooms in open waters, fl uence on their occurrence and impacts disappear on the time scale of days. Ef- long records that can characterize fun- and thus merit direct focus in the devel- fective monitoring thus requires nearly damental changes in both the physico- opment of observation and prediction continuous measurements, and mitiga- chemical environment and the ecological systems for monitoring and manage- tion responses (such as the movement system, including the frequency, dura- ment. A few of many examples include: of aquaculture cages) require communi- tion, and extent of blooms, are needed • Blooms of Heterosigma akashiwo or cations in near-real-time. Strategies for for observation and prediction. Predic- Alexandrium tamarense in Hiroshima management, such as controls on coastal tions could include long-term trends in Bay, which can be related to patterns nutrient loading or site selection for bloom frequency and yearly projections of eutrophication and local hydrog- aquaculture, depend on sustained ob- of probabilities. Except for properties raphy interacting with cyst dynamics, servations over many years to determine like N:P ratios and deep water salinity growth, and behavior of the algae; the relationships among environmental and oxygen, periodic surveys are inad- • PSP toxicity in oceanic bays (rias) variability (e.g., climate change), human equate for developing and testing predic- of northwest Spain, where Gymno- infl uences (e.g., nutrient loading), bloom tive models because transient and patchy dinium catenatum is transported from occurrences, and their impacts. events cannot be resolved. The strategy elsewhere but exerts its effects on lo- of continuous sampling from ferries and cal mussel farms due to interactions Widespread HABs: Extensive remote sensing, supplemented with re- among longshore transport, estuarine Blooms in Open Waters search cruises, appears to be on the right circulation under the infl uence of For a second class of HABs, our inter- track. Although this does not reach the winds, and swimming behavior of the est is focused on harmful or potentially ideal of continuous and synoptic ob- dinofl agellates; and harmful blooms that occur in open wa- servations, data obtained through these • Brown tides of the pelagophyte Aureo- ters in semi-enclosed seas or near coasts approaches can be used to describe the coccus anophagefferens in U.S. mid-At- where they can infl uence coastal ecosys- variability of phytoplankton with un- lantic coastal waters, which are recur- tems and be affected by terrestrial inputs precedented temporal and spatial resolu- rent and persistent, but not predict- of freshwater and nutrients. The Baltic, tion, so the occurrence of HABs can be able—explanatory hypotheses invoke North Sea, and Bohai are exemplary. In related directly to environmental forcing, preferences for organic nitrogen and the Baltic Sea, summer blooms of nitro- including climate change and nutrient other nutrients that could be elevated gen-fi xing cyanobacteria are common. loading from terrestrial sources. when estuarine fl ushing is reduced, The hepatotoxic Nodularia spumigena and also top-down control as infl u- forms conspicuous HABs in open wa- Widespread HABs Dominated by enced by suppressed grazing. ters; during the latter stages of a bloom, One Species: Localized Blooms Description and prediction of localized fi laments form highly visible aggregates When they occur, HABs cause local prob- blooms require assessment of their ex- at the surface that can be detected from lems, regardless of regional extent. With- tent and duration in relationship to local space (Kahru et al., 1994). Harmful ef- in regions (defi ned as the next larger scale conditions, quantifi cation of exchanges

Oceanography Vol.18, No.2, June 2005 223 with adjacent waters, and enough ob- sonable to expect that many toxic spe- As a result of these issues, the task of servations of nearby systems to explain cies (and other phytoplankton species) observing and predicting the dynamics why the HABs occur in one location and will be found in thin layers of stratifi ed of toxic HABs starts with the approaches not another. coastal waters. Highly resolved vertical described for blooms on the appropri- profi les, for example, with special sam- ate scales as described above, but should Blooms Strongly Influenced by plers and moored, towed, or autono- include several more components: de- Buoyancy or Swimming Behavior mous underway profi ling systems (e.g., tection and physiological characteriza- Some of the most dramatic photographs Figure 5), are required to describe the tion at the species level; measurement of blooms depict strong discoloration distributions of subsurface blooms. Be- of toxin; assessment of toxic effects; and of water near frontal features in coastal cause buoyancy and swimming behav- description of how toxins reach various waters. These phenomena can have sig- iors of phytoplankton are strongly in- target species (Doucette et al., in press). nifi cant impacts, for example, when they fl uenced by nutrition, the association of Toxic effects on competitors, grazers, or impinge on aquaculture sites or decay in subsurface layers with nutrient gradients predators that feed back on population restricted inlets, causing anoxia. Dense is quite likely, though rarely explored on dynamics should also be explored. aggregations of phytoplankton, such as this scale of thin layers. Fine-scale de- Examples of toxic species illustrate those at fronts, surface scums, concen- termination of nutrient concentrations, the diffi culties that are encountered trated subsurface layers, and transient as well as temperature, salinity, and cur- when trying to detect and predict toxic surface accumulations due to diel verti- rents, is thus needed to resolve causes HABs. The toxic species of the diatom cal migration, are all associated with in- and dynamics of blooms infl uenced by genus Pseudo-nitzschia cannot be distin- teractions between vertical movements buoyancy or swimming behavior. guished on the basis of gross morphol- of phytoplankton and discontinuities in ogy or pigmentation. Its presence can the water column (cf. Franks, 1997 and Toxic HABs be inferred through the use of specifi c in press). Consequently, detection and Toxic HABs merit special consideration probes or analysis of samples for domoic description of these blooms require ef- for several reasons: they can have harm- acid; however, an uncoupling of organ- fective sampling of phytoplankton and ful effects even if the species is not domi- isms and toxin is not unusual given the physical-chemical properties on the nant; effective detection at the species wide potential fl uctuations in cellular scales of the biological-physical inter- level and discrimination from other spe- toxicity (Scholin et al., in press). Spe- action, and modeling to describe the cies is often required; and the production cies such as Alexandrium fundyense in consequences of these interactions in of toxin is under physiological control the Gulf of Maine can cause signifi cant three dimensions. and can vary among strains within a spe- toxicity in shellfi sh even when present Subsurface layers illustrate the chal- cies and with environmental conditions at low concentrations, representing only lenges of observation and modeling. during the course of a bloom. Toxicity a fraction of the phytoplankton assem- Many phytoplankton taxa, including di- must therefore be detected in concert blage. In these cases, routine detection is nofl agellates, the prymnesiophyte Chrys- with distributions of the toxic species a major problem. In general, observation ochromulina polylepis, and diatoms of the (Figure 4) and, if possible, assessment of and prediction of algal blooms is a chal- genus Pseudo-nitzschia, can form subsur- their physiological state. The effects of lenge that requires a multidisciplinary face thin layers, thereby evading detection toxic HABs depend on the toxin, the tar- approach to detect phytoplankton and to with conventional sampling (Rines et al., gets, and how the toxin gets to the tar- describe physical-biological interactions. 2002; Holliday et al., 2003). Considering gets. Pathways and effi ciencies of transfer With the inclusion of toxic effects as a that thin layers are commonly found if between trophic compartments, as well factor, the problem becomes even more appropriate sampling is conducted, and as exchanges between pelagic and ben- multidisciplinary, complicated and chal- that specialized sampling and analysis thic communities, must be understood lenging (Scholin et al., in press). have not been widely employed, it is rea- and assessed.

224 Oceanography Vol.18, No.2, June 2005 PREDICTION OF HABS at the Habwatch workshop and is re- According to the GEOHAB Science Plan, Prediction is the stated objective of most ported in the proceedings. In the context “GEOHAB will foster the development of plans for real-time coastal observation of our discussion here, it is important new observation technologies and mod- systems. It can be deﬁ ned as the estima- to recognize that all predictions depend els to support fundamental research on tion of properties that are not observed on observations, and the observations HABs, improve monitoring, and develop directly with known certainty (IOC, should provide not only the input to the predictive capabilities. Because capabilities in coastal observation and modeling are advancing rapidly on many fronts, coordination and cooperation among Integration is the answer: molecular biology and the different elements of the GEOHAB program are essential. The intention is bio-optics with ocean observation technology; to ensure rapid and effective integration physiological ecology with oceanography and of knowledge, technical capabilities, and data across disciplines and regions.” The numerical modeling; real-time observing systems Habwatch workshop contributed strong- with ongoing efforts in monitoring and research. ly to the latter, and based on the proceedings we offer two recommendations as guidance for the GEOHAB program.

2003). This broad defi nition includes models, but also the data for validating Encourage Further Advances In hindcasts, nowcasts, and forecasts. The the predictions and estimating error. Species-Specific Detection latter two are key products of real-time Consequently, the scales of observations Many of the new, powerful, and excit- systems, but their development and eval- and models should match. ing tools for observation of HABs have uation depends on the former. Nowcasts widespread oceanographic applications serve as the best possible assessment of SUMMARY AND GUIDANCE and merit strong, broad-based support. current conditions, useful in early warn- FOR THE GEOHAB PROGRAM Some of the most promising approaches ing. Also, as a time series, nowcasts pro- The scientifi c goal of the GEOHAB for early warning and for studying HAB vide a record of environmental change program is to “improve prediction of species are those employing molecu- that is richer than a compilation of di- HABs by determining the ecological and lar-based methods. Prototypes of rRNA rect observations alone; this is the future oceanographic mechanisms underlying probe arrays for in situ and autonomous of coastal monitoring (Cullen, in press). their population dynamics, integrating deployment have been tested and several Forecasting of HABs, however, is clearly biological, chemical, and physical stud- new toxin detection technologies may an ultimate goal. ies supported by enhanced observation be amenable to in situ platforms. GEO- All prediction depends on models, and modeling.” The specifi c objectives of HAB should encourage development of which include conceptual descriptions GEOHAB for observation are: a broad range of tools for observations of ecological relationships, statistically • To develop capabilities to observe of HABs, with a priority on the develop- based empirical models, and a range of HAB organisms in situ, including ment of species-specifi c sensors. numerical models of varying complexity. their properties and the processes that For many, the Holy Grail is the coupled, infl uence them; Promote The Development of physical-biological, ocean-atmosphere, • To develop and evaluate systems for Integrated Observation Systems data assimilative model of coastal dy- long-term monitoring of HAB species; Populations of HAB species develop in namics including HABs. A broad range • To develop capabilities for real-time different environments (e.g., bays, fjords, of modeling approaches was addressed observation and prediction of HABs. open ocean), and at various temporal

Oceanography Vol.18, No.2, June 2005 225 and spatial scales (e.g., patches, large CONCLUSION entifi c Commission on Oceanic Research blooms, thin layers). Some form high- Technical advances are rapidly trans- (SCOR), Centre Nationale de la Recher- biomass blooms while others do not. forming oceanography, but technology che Scientique (CNRS), Institut Français Observations serve many purposes in alone cannot solve pressing environmen- de Recherche pour l’Exploration de la the study, monitoring and prediction of tal challenges such as the need to detect MER (Ifremer), National Oceanic and harmful algal blooms: (1) early warn- and predict harmful algal blooms. Inte- Atmospheric Commission (NOAA). ing; (2) description of population and gration is the answer: molecular biology The Habwatch workshop was endorsed by the following programs or organi- zations: GEOHAB, Global Ocean Ob- serving System (GOOS), International Workshops such as Habwatch and programs such Council for the exploration of the Sea (ICES), and Programme National sur as GEOHAB foster the interdisciplinary and l’Environnement Côtier (PNEC). We are international interactions that will result in new especially thankful to ACRI-ST (www. capabilities to observe and predict HABs. acri-st.fr) for their major contribution to the organization of the Habwatch workshop, and to Lewis Conference Services International, Inc. for logistical support. ecosystem dynamics; (3) model develop- and bio-optics with ocean observation Comments of an anonymous reviewer ment, including parameterization and technology; physiological ecology with were particularly helpful. This is Woods validation; (4) determination of model oceanography and numerical modeling; Hole Oceanographic Institution contri- initial and boundary conditions; (5) data real-time observing systems with ongo- bution 11335. assimilation in predictive models; and ing efforts in monitoring and research. Disclaimer: The National Ocean (6) time-series analysis. A growing arse- Workshops such as Habwatch and pro- Service (NOS) does not approve, rec- nal of sensors can provide observations grams such as GEOHAB foster the inter- ommend, or endorse any product or to serve these purposes, but many sen- disciplinary and international interac- material mentioned in this publication. sors operate only in specifi c conditions tions that will result in new capabilities No reference shall be made to NOS, or (e.g., clear skies for ocean color remote to observe and predict HABs. Continued to this publication furnished by NOS, sensing) and with different spatial and commitment to communication and col- in any advertising or sales promotion temporal resolutions. So, a combination laboration across disciplines and sectors which would indicate or imply that NOS of platforms and sensors is necessary to will ensure rapid progress. approves, recommends, or endorses any observe and predict HABs. Observation product or material mentioned herein strategies must be developed for each ACKNOWLEDGEMENTS or which has as its purpose any intent HAB phenomenon using knowledge of The authors gratefully acknowledge the to cause directly or indirectly the adver- oceanography and bloom dynamics to following agencies for their fi nancial tised product to be used or purchased design a system making the best use of support of the Habwatch workshop: because of NOS publication. available platforms and sensors to serve European Commission (EC), National identifi ed scientifi c and management Scientifi c Foundation (NSF), European REFERENCES goals. We therefore recommend that, Space Agency (ESA), Offi ce of Naval Re- Babin, M. In press. Phytoplankton fl uorescence: Theory, current literature and in situ measure- within the framework of GEOHAB, high search (ONR), International Foreign Of- ment. In: Real-Time Coastal Observing Systems priority be given to research on integrat- fi ce of ONR, Centre National d’Etudes for Ecosystem Dynamics and Harmful Algal ed observation strategies tailored to the Spatiales (CNES), Intergovernmental Bloom, M. Babin, C.S. Roesler and J.J. Cullen, eds. UNESCO Publishing, Paris, France. specifi c phenomena being studied. Oceanographic Commission (IOC), Sci-

226 Oceanography Vol.18, No.2, June 2005 Chang, G.C. and T.D. Dickey. In press. Interdisci- Kahru and W. Brown, eds. Landes Bioscience C.S. Roesler and J.J. Cullen, eds. UNESCO Pub- plinary sampling strategies for detection and Publishing, Georgetown, TX, USA. lishing, Paris, France. characterization of harmful algal blooms. In: Johnson, K.S. and L.J. Coletti. 2002. In situ ultra- Roesler, C.S., S.M. Etheridge, and G.C. Pitcher. Real-Time Coastal Observing Systems for Eco- violet spectrophotomethry for high resolution 2003. Application of an ocean color algal taxa system Dynamics and Harmful Algal Bloom, M. and long-term monitoring of nitrate, bromide detection model to red tides in the Southern Babin, C.S. Roesler and J.J. Cullen, eds. UNES- and bisulfi de in the ocean. Deep-Sea Research I Benguela. In: Proceedings of the Xth Internation- CO Publishing, Paris, France. 49:1291-1305. al Conference on Harmful Algae, K.A. Steidinger, Cowles, T.J. 2003. Planktonic layers: Physical and Kahru, M., U. Horstmann, and O. Rud. 1994. Satel- J.H. Lansdberg, C.R. Tomas and G.A. Vargo, eds. biological interactions on the small scale, Hand- lite detection of increased cyanobacteria blooms Florida Fish and Wildlife Conservation Com- book of Scaling Methods in Aquatic Ecology: in the Baltic Sea: Natural fl uctuation or ecosys- mission and Intergovernmental Oceanographic Measurements, Analysis, Simulation. CRC Press. tem change? Ambio 23:469-472. Commission of UNESCO, St. Petersburg, FL, Cullen, J.J. In press. Observation and prediction Kahru, M., B. G. Mitchell, A. Diaz, and M. Miura. USA. of harmful algal blooms. In: Real-Time Coastal 2004. MODIS Detects a Devastating Algal Ruddick, K., G. Lacroix, Y. Park, V. Rousseau, V. De Observing Systems for Ecosystem Dynamics and Bloom in Paracas Bay, Peru. EOS Transactions of Cauwer, W. Debruyn, and S. Sterckx. In press. Harmful Algal Bloom, M. Babin, C.S. Roesler the American Geophysical Union 85:465-472. Overview of ocean colour: Theoretical back- and J.J. Cullen, eds. UNESCO Publishing, Paris, Kirkpatrick, G.J., D.F. Millie, M.A. Moline, and O. ground, sensors and applicability for the detec- France. Schofi eld. 2000. Optical discrimination of a tion and monitoring of harmful algae blooms Donaghay, P.L. and T.R. Osborn. 1997. Toward a phytoplankton species in natural mixed popula- (capabilities and limitations). In: Real-Time theory of biological-physical control of harmful tions. Limnology and Oceanography 45:467-471. Coastal Observing Systems for Ecosystem Dy- algal bloom population dynamics and impacts. Kolber, Z. and P.G. Falkowski. 1993. Use of active namics and Harmful Algal Bloom, M. Babin, C.S. Limnology and Oceanography 42:1283-1296. fl uorescence to estimate phytoplankton photo- Roesler and J.J. Cullen, eds. UNESCO Publish- Doucette, G.J., J.T. Turner, C.L. Powell, B.A. Ke- synthesis in situ. Limnology and Oceanography ing, Paris, France. afer, and D.M. Anderson. In press. ECOHAB- 38:1646-1665. Schofi eld, O., J. Bosch, S. Glenn, G. Kirkpatrick, J. Gulf of Maine. Trophic accumulation of PSP Kolber, Z., O. Prasil, and P.G. Falkowski. 1998. Mea- Kerfoot, M. Moline, M. Oliver, and P. Bissett. toxins in zooplankton during Alexandrium surements of variable chlorophyll fl uorescence In press. Harmful algal blooms in a dynamic fundyense blooms in Casco Bay, Gulf of Maine, using fast repetition rate techniques: Defi ning environment: How can optics help the fi eld-go- April–June, 1998. I. Toxin levels in A. fundyense methodology and experimental protocols. Bio- ing and sample-poor biologist? In: Real-Time and zooplankton size fractions. Deep-Sea Re- chimica et Biophysica Acta 1367:88-106. Coastal Observing Systems for Ecosystem Dy- search II. Lehaître, M. and C. Compère. In press. Biofouling namics and Harmful Algal Bloom, M. Babin, C.S. Franks, P.J.S. 1997. Spatial patterns in dense algal and underwater measurements. In: Real-Time Roesler and J.J. Cullen, eds. UNESCO Publish- blooms. Limnology and Oceanography 42:1297- Coastal Observing Systems for Ecosystem Dy- ing, Paris, France. 1305. namics and Harmful Algal Bloom, M. Babin, C.S. Scholin, C.A., G.J. Doucette, and A.D. Cembella. In Griffi ths, G. In press. Glider and autonomous un- Roesler and J.J. Cullen, eds. UNESCO Publish- press. Prospects for developing automated sys- derwater vehicle observing systems. In: Real- ing, Paris, France. tems for in situ detection of harmful algae and Time Coastal Observing Systems for Ecosystem Morel, A. and M.R. Lewis. In press. Introduction to their toxins. In: Real-Time Coastal Observing Dynamics and Harmful Algal Bloom, M. Babin, optics of water bodies, phytoplankton and other Systems for Ecosystem Dynamics and Harmful C.S. Roesler and J.J. Cullen, eds. UNESCO Pub- marine constituents, and to measurements of Algal Bloom, M. Babin, C.S. Roesler and J.J. Cul- lishing, Paris, France. optical properties. In: Real-Time Coastal Ob- len, eds. UNESCO Publishing, Paris, France. Hanson, A.K.J. and C. Moore. 2001. Real-time nu- serving Systems for Ecosystem Dynamics and Sieracki, C.K., M.E. Sieracki, and C.S. Yentsch. trient surveys in coastal waters. Sea Technology Harmful Algal Bloom, M. Babin, C.S. Roesler 1998. An imaging-in-fl ow system for automated 42:10-14. and J.J. Cullen, eds. UNESCO Publishing, Paris, analysis of marine microplankton. Marine Ecol- Holliday, D.V., P.L. Donaghay, C.F. Greenlaw, D.E. France. ogy Progress Series 168:285-296. McGeehee, M.M. McManus, J.M. Sullivan, and Olson, R.J., A.A. Shalapyonok, and H.M. Sosik. Sosik, H.M. In press. Characterizing seawater con- J.L. Miksis. 2003. Advances in defi ning fi ne- and 2003. An automated submersible fl ow cytom- stituents from optical properties. In: Real-Time micro-scale pattern in marine plankton. Aquatic eter for pico- and nanophytoplankton: FlowCy- Coastal Observing Systems for Ecosystem Dy- Living Resources 16:131-136. tobot. Deep-Sea Research I 50:301-315. namics and Harmful Algal Bloom, M. Babin, C.S. IOC. 2003. The integrated, strategic desing plan for Rines, J.E.B., P.L. Donaghay, M.M. Dekshenieks, Roesler and J.J. Cullen, eds. UNESCO Publish- the coastal ocean observations module of the and J.M. Sullivan. 2002. Thin layers and camou- ing, Paris, France. global ocean observing system. UNESCO, Paris, fl age: Hidden Pseudo-nitzschia populations in a Sosik, H.M., R.J. Olson, M.G. Neubert, A.A. Shala- France. fjord in the San Juan Islands, Washington, USA. pyonok, and A.R. Solow. 2003. Growth rates of Jaffe, J.S. In press. Sensing plankton: Acoustics and Marine Ecology Progress Series 225:123-137. coastal phytoplankton from time-series mea- optical imaging. In: Real-Time Coastal Observ- Roesler, C.S. and E. Boss. 2003. Spectral beam at- surements with a submersible fl ow cytometer. ing Systems for Ecosystem Dynamics and Harm- tenuation coeffi cient retrieved from ocean color Limnology and Oceanography 48:1756-1765. ful Algal Bloom, M. Babin, C.S. Roesler and J.J. inversion. Geophysical Research Letters 30:1468- Stumpf, R.P., .M.E. Culver, P.A. Tester, M. Tomlin- Cullen, eds. UNESCO Publishing, Paris, France. 1472. son, G.J. Kirkpatrick, B.A. Pederson, E. Truby, V. Johnsen, G., Z. Volent, K. Tangen, and E. Sakshaug. Roesler, C.S. and E. Boss. In press. In situ measure- Ransibrahmanakul, and M. Soracco. 2003. Mo- 1997. Time series of harmful and benign phyto- ment of the inherent optical properties (IOPs) nitoring Karenia brevis blooms in the Gulf of plankton blooms in northwest European waters and potential for harmful algal bloom detection Mexico using satellite ocean color imagery and using the Seawatch buoy system. Pp. 115-143 in and coastal ecosystem observations. In: Real- other data. Harmful Algae 2:147-160. Monitoring algal blooms: New technologies for Time Coastal Observing Systems for Ecosystem detecting large-scale environmental change, M. Dynamics and Harmful Algal Bloom, M. Babin,

Oceanography Vol.18, No.2, June 2005 227