|SCOR Working Group 118: New Technologies for Observing Marine Life|
Terms of Reference
Working Group Members
Funding provided by
the SLOAN Foundation's
Census of Marine Life
2002 Working Group Meeting
(28-30 October 2002, Lima, Peru)
Reports and Activities
Mar del Plata
AGU/ASLO Ocean Sciences
Imaging cytometry using digital analysis of epifluorescence microscope images is ideal for analysing prokaryotes and heterotrophic protists from natural marine samples. Such imaging systems provide rapid determination of cell abundance and sizes for calculating size spectra and biomass. Flow cytometry is ideal for detecting and quantifying prokaryotes and pico- and nano-phytoplankton from natural samples. 'Allometric analysis' uses plots of side scatter and forward scatter; 'taxonomic analysis' uses plots of fluorescence and forward scatter. The FlowCAM instrument, which can be installed on a floating dock or in the flow system of a ship underway, images cells in flow using a chlorophyll fluorescence trigger. Cell sizes are measured directly from the images and the instrument is ideal for analysis of microplankton (>20µm), including phytoplankton and ciliates. New methods of deployment allow in-situ cytometry to be undertaken from a boat, a mooring or a submarine (e.g. AUTOSUB). SIPPER (Shadowed Image Particle Profiling & Evaluation Recorder), which produces shadow profiles of larger organisms (e.g. chaetognaths, copepods, euphausiids, pteropods, salps, siphonophores, fish larvae and many other organisms) can be deployed in an AUV or in a towed package for high-resolution in-situ measurements. A variety of SIPPER images can be found at http://cot.marine.usf.edu/multimedia.sap.
Size fractionated in-situ absorption spectroscopy can be used to produce spectral distribution curves from different size fractions (e.g. <5µm, 5-20µm, >20µm) of a bulk sample. Dominant species in each size fraction can be identified microscopically and species composition confirmed by the spectral characteristics of the relevant cell pigments. Fourth order derivative spectra and similarity indices can be used to differentiate between mixed assemblages of phytoplankton.
Taxonomic data can be derived by remote sensing of spectral reflectance, which can be expressed as a function of the backscattering to absorption ratio. These two properties are in turn parameterised by linear combinations of optically active components, which include water, particles, phytoplankton, coloured particulate and dissolved organic materials. Assuming spectral shapes for each component (eigenfunctions), magnitude (eigen values) can be estimated by non-linear regression. Observed spectral reflectance curves can be compared either with a single phytoplankton eigenfunction (standard model) or with a species-dependent reflectance inversion model based on six phytoplankton absorption eigenfunctions, whose spectral differences are due primarily to pigment composition and secondarily to relative pigment concentrations. The more advanced model can be used to derive the species compositions, which can be validated by direct sampling and identification.
Bench-top methods based on the optical properties of both single cells (e.g. flow-cytometry) and bulk cells (e.g. absorption spectroscopy) are now being packaged for in-situ analysis on moorings, hydrocasts and AUVs. In future, molecular techniques will be combined with single cell optical methods to provide genetic taxonomic data. Optical properties of bulk cells will be utilised by routine use of inversions of hyper-spectral remote sensing. One in situ flow cytometer is in commercial production, although not yet fully debugged. A silhouette flow camera and several remote-sensing instruments are also available.
POGO + IOC/CoML Worksop (Thailand)
Topics discussed at this workshop were: biodiversity & conservation; sustainable management of living resources ('responsible fisheries'); oceanic biota & global change; bio-invasion; ocean fertilisation; and threatened habitats (corals & seagrass). For each of the three related scientific issues (global change & the carbon cycle; constraints on primary production & remineralisation; biodiversity & ecological function) key variables and measurements were identified. For biodiversity & ecosystem function, highest priority was afforded to ocean colour, CPR and CTD. DNA probes were also accorded high importance and recommended for development to the operational level, together with functional groups (DNA), image analysis, molecular data banks, and microscopy. Appropriate sensors and platforms were: time series stations & oceanic observations; small AUVs; volunteer observing ships (VOS); Argo type autonomous floats (with a variety of sensors for fluorometry, oxygen, CTD, photosynthetic yield, nutricline); bioprobes (telemetry tags on mobile marine mammals); and research vessels. The workshop discussed OBIS and drew lessons about data management and the distribution of biological observations from GOOS, IOC, PICES and CoML. It also recommended various ways of building research capacity, which included links with graduate education in marine science and contact with local scientists and research cruises. In relation to bio-diversity, it was concluded that there was a need to embed biological observations in a physical context (CTD), to develop a number of emerging technologies to an operational level, and to conduct low cost surveys at large scales. In this context, the key emerging technologies were DNA probes, flow cytometers on buoys (automated); holographic cameras and small AUVs of the hovering type.
Elgar Desa's second report concerned the IOC/CoML workshop on marine biodiversity held at the Marine Biological Centre in Phuket in October 2001. The purpose of this workshop was to introduce and expand CoML activities in SE Asia and to introduce SCOR WG 118 to scientists in the WESTPAC region. Regional participation was from Singapore, Philippines, Thailand, Indonesia, Malaysia, China, Vietnam, Cambodia and Australia. The workshop addressed two questions: to what extent is advanced sampling/identification technology known and being used in SE Asia; and what are the major needs for new technology in the region? Discussion revealed that most researchers are aware of and use the following technologies: video cameras and microscopes; DNA probes (in the Philippines); electronic keys for taxonomy; data buoys; ROVs (limited use in Malaysia because of cable problems); GIS; satellite ocean colour; acoustics (mainly for fisheries investigations). Reactions to emerging technologies were varied. Some scientists felt there was no pressing need, others welcomed it but expressed the need for training, and others felt it was too costly; there were also reservations about ocean colour imagery because of cloud cover. Most regional participants agreed that their needs were: training to organise, clean up, catalogue and expand available databases; references on 'species'; computer aided taxonomists; image processing techniques; and appropriate technologies for shallow water ecosystems, including high-resolution digital cameras and expertise in DNA probe technology for species identification because of genetic diversity in the region. The output of the workshop is available as eight reports on CD (Country Reports, IOC Workshop on the Census of Marine Life in South East Asia, Phuket, Thailand, 10-12 October 2001) . It was concluded that, whilst there is considerable local awareness of the rich biodiversity in the region, there is a need to organise available information and to introduce low cost surveys to rapidly monitor and identify biodiversity over large scales.
ICES Annual Science Conference 2002
One interesting French presentation (L18) had compared results of close-up studies with ROVs and Landers equipped with video cameras and bait with data from fishing gear. Because of behavioural reactions of fish to noise and artificial lights, fish that were often caught in trawls were rarely seen on video. The need therefore was for cheaper, faster and less noisy platforms that provided a larger sampling volume and did not affect the behaviour of the organisms under investigation. For deep water observations, such as were needed for CoML's MAR-ECO project, it would be necessary to apply standard acoustic techniques in deep water using towed vehicles and AUVs, together with complementary sampling techniques involving fishing and video observations. Stomach sampling was also needed for diet analysis and here the challenge was to catch fish at depth to avoid regurgitation, a problem to which the Icelandic automatic tagging and fish-collecting machine might provide a solution, http://www.star-oddi.com/. Temporal resolution during snapshot surveys might be studied with instrument rigs that recorded acoustic measurements and environmental factors (as described in paper L10) and ships of opportunity. Because of sampling bias, catch results could not be taken at face value and interpretation required visual validation and an understanding of the processes between the application of the technology and the received results.
In discussion, it transpired that, although ICES was addressing ecosystem problems, it still had a fisheries bias and had not fully integrated the CoML approach. ROV technology was very noisy and, although it was improving, did not yet allow quantitative sampling. Quieter vehicles with laser cameras and acoustic imaging or visualisation would be very useful, particularly if the technique allowed species recognition.
SCOR Working Groups 119 and 115
The WG 115 meeting included a series of short presentations designed to define the breadth of the scientific questions, challenges and 'hot topics' that need to be addressed by biological oceanographers globally. These talks provided graphic illustrations of some of the issues relevant to future discussions of WG115 and WG 118. They identified a number of basic needs, which included: ensuring that monitoring surveys in different oceans produce comparable data; providing basic technology, research vessels and trained staff in countries with limited resources; preserving, curating and archiving samples for long periods; sampling over the correct scales of space and time to identify the unaliased and unbiased patterns required to understand cause and effect; developing techniques for rapid synoptic surveys over wide areas at reasonable cost to allow ecosystem management.
Whilst its Terms of Reference strongly encourage it consider the addition of unconventional technologies to existing plankton survey and sampling methods, WG 115's first meeting was largely concerned with the Continuous Plankton Recorder and its unique place in plankton monitoring. As a result, consideration of new technologies was limited to a discussion of optical and acoustic methods of plankton sampling and assessment at a generic level. Following presentations on these two subjects, WG 115 divided into four discussion-groups, one of which considered a standard package of additional measurements that should be made in conjunction with routine plankton surveys. The initial list of ancillary measurements identified and prioritised by this group was: latitude, longitude and time associated with each sample; sample depth(s); temperature; salinity; irradiance or PAR (photosynthetically active radiation); wind speed and direction; fluorescence; and flow. The group also suggested the collection (where possible) of data from multi-frequency acoustics, a variety of optical instruments (e.g. transmissometers & scatteromemeters), bioluminescence sensors and the Optical Plankton Camera (especially the new imaging version when commercially available).
With respect to future deliberations of WG 118, Van Holliday drew a number of conclusions about the use of new technology in plankton sampling, which can be summarised as follows. Current CPR programmes address only a small fraction of the issues that biological oceanographers must consider, if they are to understand the processes that determine the distribution of plankton in space and time at all scales from that of individual organisms to ocean basins. Although monitoring programmes such as the CPR can reveal shifts in species distribution, monitoring programmes cannot provide all of the information needed to develop predictive capability and provide sound management advice. For this it is necessary to conduct process studies with much greater spatio-temporal resolution than can be provided by conventional methods. Advanced technologies exist to overcome these problems and the challenge is how to get them into the hands of trained users. New and improved sensors are, however, still needed to examine the small-scale distributions that are now known to be ecologically critical. For example, whilst modern optics and acoustics have shown that some sub-meter scale vertical structures may contain 80-90% of the plankton biomass in the water column, no techniques exist to collect zooplankton from within them. Techniques for sampling phytoplankton in these structures are also limited. Some of these structures are known to harbour seed organisms for harmful algal blooms and current sampling methods are unlikely to detect these "seeds" before they bloom and become a health issue.
Avoidance, extrusion and sampling at critical scales (i.e. Shannon or Nyquist rate sampling) in 4-D space in a heterogeneous environment from a moving, heaving platform are all real and demanding problems. As a community, biological oceanographers need to detect and localise biological and physical structure in the water column with high-tech sensors in order to direct their limited sampling effort efficiently. Because of limited budgets, they also need to develop sensors that can be deployed on cruises and moorings whose primary purpose is oceanography or meteorology.
Training is a critical issue in some disciplines, where the disappearance of specialist skills (e.g. plankton taxonomy) may result in failure to progress. We need more scientists and engineers trained to collect, maintain, calibrate and interpret data from high-technology sensors and also more people to develop new technology. In developing countries there are well-trained scientists with invaluable knowledge of understudied areas of the world's oceans who could make a major contribution to biological oceanography, if they could be given access to medium- or high-tech instruments. One way to achieve this might be to persuade international funding agencies (e.g. the World Bank) to contribute to the acquisition of these instruments, thereby widening the scope of international programmes. Another approach would be to simplify the funding of multi-national efforts to attack specific research objectives. Finally, it would be valuable to find ways of funding the development medium-tech, low-cost, low-maintenance sensors that could be used by all nations to expand the ocean areas from which data can be currently obtained. If successful, such an initiative could well triple or quadruple the number of biological oceanographers worldwide.
Topics discussed during the PICES technical theme session included systems of managing and merging complex data (difficult to implement with biological data), data collection using large arrays of optical instruments at locations with large concentrations of zooplankton, and global sensing using passive floats with satellite data retrieval. Instruments included a prototype system for collecting discrete samples in thin layers, which was being developed in Japan and consisted of a set of small packed tubes in a towed body, and an optical plankton camera, which had been developed in Russia. The camera, which produced silhouettes of plankton using a narrow sheet of light, had been abandoned because results showed major disagreements with net catches.
Gaby Gorsky began his talk by reminding the PICES meeting of the close connections between top predators and plankton and the increases in the abundance (e.g. Black Sea) of phytoplankton and gelatinous zooplankton that could follow overfishing. He also reminded them of the avoidance problem and the difficulties of obtaining representative samples with a conventional plankton sampler. He then reviewed the relative merits and capabilities of modern optical plankton sampling instruments, which included towed instruments (e.g. optical plankton camera, video plankton recorder), vertical profilers (UVP, ZOOVIS, LAPIS) and towed platforms, such as SIPPER, FLOWCAM and ZOOSCAN, as well as various holographic cameras. ZOOSCAN, for example, could sample up to 6000 organisms per day and produce a master data table with measured parameters and an image for each individual. Identification, which was largely automatic, was based on a training set (look-up table) of images and a neural network. A new look-up table was needed for each region with the images in the same proportion (e.g. 80% copepods) as the local organisms.
Despite substantial progress with optical instruments in recent years, there were still a number of major problems to be solved, mainly related to resolution, field of view, identification and the probability of non-detection. Fields of view, for example, were generally very small except for LAPIS, which had a field of view of 2 x 4 m2. Rapid advances in image processing and computing indicated, however, that significant further advances were possible in the near future, particularly if optical instruments were used in combination with acoustics to increase the sampling volume and laser illumination.
Fisheries Acoustics, Japan