Optics-Biomass Estimation

Biomass and Abundance Estimation Working Group

Chairman: Mark Berman 
Rapporteur: Jeff Napp

Participants: Daniel Davis, Tommy Dickey, Van Holliday, Hein Skjoldal, Ron Zaneveld.

Task Statement

This group undertook to identify optical techniques which would facilitate quantification of the structure of plankton communities and their relationships to other biotic and abiotic variables over a wide range of temporal and spatial scales.

Scientific Problem

A basic objective of U.S. GLOBEC is to understand the ways by which physical and biotic factors govern the abundance and distribution of planktonic organisms over a range of temporal and spatial scales. Numerous studies have shown that zooplankton respond to certain physical (e.g., temperature, turbulence), biological (e.g., prey distribution, predators), and chemical (e.g., amino acids) variables. Our understanding concerning the range and extent of effects of these variables is severely limited because of the usual continuous motion of animals and water in three dimensions, and our inability to follow and sample or observe animal populations (e.g., Marine Zooplankton Colloquium 1, 1989) and water masses. To a certain degree this latter shortcoming can be solved by the application of improved technology. For phytoplankton to macrozooplankton sized marine (or aquatic) organisms optical methods appear to offer promising solutions to this problem.

Goal

The primary goal is to quantify the biomass of planktonic populations (including phytoplankton, meroplankton, holoplankton, and ichthyoplankton) and communities, and to describe the distribution and species composition of the plankton in relation to biotic and abiotic variables over a range of temporal and spatial scales.

Specific Objectives

To measure total plankton biomass and its distribution on scales ranging from a few centimeters to tens of kilometers.
Biomass assessment should be sufficiently rapid and routine that surveys can be accomplished quasi-synoptically with results available in real, or near-real, time. This objective specifically includes the description of spatial, both horizontal and vertical, heterogeneity of plankton, its temporal persistence, and biomass distribution within a patch (i.e., distance to nearest neighbor).
To relate the distribution of plankton to the structure of its biological, physical and chemical environment.
This would include description and evaluation of the physical and biological mechanisms responsible for the formation and maintenance of plankton patches. Animal behavior as it is affected by physical environment and biological neighbors is a variable whose importance must be assessed (see Rate Processes Working Group Report).
To determine the composition of plankton communities in relation to their environment.
The structure of plankton communities can be crudely described by size-frequency distributions (e.g., Platt and Denman, 1978; Napp et al., 1993). At a more detailed level, plankton structure is described using relatively coarse taxonomic categories (e.g., separation of copepods, fish larvae, amphipods, and phytoplankton) (Berman et al., 1990). Ideally, the structure of planktonic assemblages is best described by species-specific, or even developmental stage specific identification of each planktonic individual.

Sampling Strategy

Eulerian and Lagrangian approaches have both been used in the past to describe the biomass and composition of plankton assemblages; rarely have both approaches been used simultaneously. The success of U.S. GLOBEC will eventually depend on the temporal and spatial resolution of biomass, abundance and species composition observations (e.g. Mackas et al., 1985), and the rapidity of sample analysis. Traditional zooplankton sampling using nets and pumps provides relatively few and rarely replicated observations. The samples require a long time to analyze and provide poor temporal and spatial resolution. With the improvement of acoustical and optical technology for biomass quantification and the use of long-term moorings and/or drifters, the spatial resolution, frequency of observation, and rapidity of sample processing will improve. This improvement will provide more timely and detailed data on zooplankton abundance and distribution, and a better understanding of the structure and dynamics of zooplankton and ichthyoplankton populations (see reviews by Dickey, 1988, 1991).

Required Sensor Technologies

Achieving the objectives listed above will require the application of different combinations of sensor technologies and data analysis techniques. The objectives are listed in order of least (biomass and size only) to most (biomass, size and taxonomic identification, coupled with concurrent physical structure observations) sophisticated data return. These objectives will, therefore, be discussed separately.

Technology to quantify total plankton biomass.

The most advanced optical sensor in routine use for zooplankton biomass and size structure assessment is the Optical Plankton Counter (OPC, Herman, 1988, 1992; Herman and Dauphinee, 1980) which is designed to count and size zooplankton in the size range from 0.25 to 30 mm. The OPC employs a parallel light beam of 640 nm wavelength and uses the maximum cross-sectional area of a zooplankter as it passes through the beam to estimate size. It is useful for identifying the dominant copepods in boreal environments where relatively few species dominate much of the planktonic biomass. Recent modifications to the OPC have extended its suitability for sizing small (down to 0.12 mm) zooplankton (Herman, 1992). It can be deployed in conjuction with net sampling systems, on remotely operated vehicles, or by itself. When the OPC is coupled with other sensors as a single deployable instrument package, it can provide reasonably detailed spatial resolution of the interaction between plankton and the physical structure of their habitat. In this regard, it is useful in addressing objective (b) below. It is commercially available and represents one of the more mature optical instruments available for the assessment of zooplankton biomass and distribution. It is not however, in its commercially available form, capable of imaging individual zooplankton, and thus is somewhat limited in its ability to define the taxonomic structure of zooplankton assemblages. It is adaptable for towing as part of an undulating vehicle, and a moored version is under development.

Technology to determine distribution of plankton and their relation to the structure (biological, physical, chemical) of their fluid environment.

This objective focuses on the quantification of plankton distribution, supported by the quantification of major physical (e.g., temperature, salinity) and biological (e.g., distribution of food organisms, predators) variables. Distribution implies abundance measurements in up to three dimensions and through time. To accomplish this, various instruments employing acoustics and optics have been developed. Shipboard or moored Acoustic Doppler Current Profilers (ADCP) can measure acoustic backscatter, which if calibrated properly, can provide data on the vertical distribution of zooplankton biomass over extended periods of time (Flagg and Smith, 1989) or quasisynoptically for a spatial grid or transect (e.g., Heywood, et. al., 1991). The vertical distribution of 21 independent size classes of organisms can be measured using the Multifrequency Acoustic Profiling System (MAPS, Holliday et al., 1989). The report of the U.S. GLOBEC workshop on Acoustical Technology (GLOBEC Report Number 4) provides an overview of sonar and acoustical sampling techniques and makes recommendations for further research in the area of acoustical instrumentation. However, as discussed below, a shortcoming of current acoustical instruments is their general inability to positively identify acoustic targets.

Several optical instruments currently exist that can provide data on plankton abundance (and to some extent size structure) at spatial and temporal scales similar to those at which physical structure is measured. The above-mentioned OPC in its towed version is commercially available and can be used to determine plankton abundance and distribution vertically and horizontally (Herman, 1988, 1992). The data can be immediately processed and displayed. When coupled with other sensors (e.g., for temperature, salinity, oxygen concentration, fluorescence, etc.) the OPC is useful for describing plankton distributions in relation to their physical, chemical, and biotic environment.

The Video Plankton Recorder (VPR, Davis et al., 1992a,b) has been developed to quantify abundance of zooplankton on scales from microns to kilometers, and is now in the prototype stage. It consists of a video camera/strobe unit and an image processing system. Four video cameras are synchronized at 60 frames sec-1 to a red strobe light positioned 1 meter away. The field of view of each camera is adjustable from 0.5 to 10 cm, at 10 to 300 microns resolution, respectively, with corresponding depths of field of 4 to 20 cm. Imaged volumes are concentric with their center located 0.5 m between camera and strobe. Each one microsecond strobe pulse permits highly resolved images of plankton and seston. Plankton abundance is determined by counting the number of animals per field of videotape and dividing by the field volume.

The Optical Plankton Recorder (OPR, Kils, 1981; 1989) is a compact, high-speed, underwater video microscope with optional preconcentration nets. It is designed primarily for small-scale, high-resolution observations of plankton distributions. Prototype instruments have been deployed free-falling in Antarctic krill studies (Kils, 1981); towed from small vessels in mesoscale monitoring of fish schools; anchored (moored) for plankton orientation and ecotoxicology studies; and used in aquaculture for particle flow quantification (Kils et al., 1991). When towed, free-falling, or hovering, each image is exposed to two short (10 盜) strobes separated by 20 ms. Three different cameras with nested magnifications allow for observation of both predators and prey simultaneously, and for taxonomic identification (Kils, 1989).

To understand the distribution of zooplankton in their physical and chemical environment requires simultaneous collection of physical, chemical and biological data at common spatial and temporal scales. The distribution of phytoplankton has been found to be closely correlated with physical phenomena such as density gradients and shear layers. Moreover, at times phytoplankton distribution influences the distribution of zooplankton (e.g. Paffenhofer, 1983; Roman et al., 1986; Napp et al., 1988). Several bulk optical parameters are useful in describing the in situ distribution of phytoplankton or particulate biomass at scales comparable to physical structure measurements:

Beam attenuation: Beam attenuation at 660 nm has been used to estimate the volume of total particulate matter. For more than 20 years it has been applied to estimate total mass of autotrophic, heterotrophic and detrital particles. The standard instrument is the transmissometer.
Particle scattering: Back scattering of light from particles can be used to estimate total particle volume. Recent tests of prototype instruments have shown that particle scattering in a volume of 1 ml can be achieved for particle concentrations ranging over five orders of magnitude. Because the sensor is small it can be deployed in a variety of modes.
Red fluorescence: Stimulation of red fluorescence by ultraviolet and blue light has been widely used to estimate phytoplankton biomass (Yentsch and Menzel 1963). Low sampling rates (about 1 Hz) and internal averaging limit the spatial resolution of current in situ fluorometers.
Spectral fluorescence: The completion of development of in situ spectral fluorometers is about two years away. Tests of prototype laser-induced spectral fluorometers have shown their usefulness in measuring chlorophyll and other phytoplankton pigment signatures at vertical spatial resolutions on the order of 1-2 cm (Cowles et. al., 1989; Carr et. al., 1992). These instruments have the potential to estimate the biomass of major taxonomic groups of phytoplankton, such as diatoms, cyanobacteria and green algae.
Spectral absorption: Prototype measurements indicate that in situ absorption can be measured at a single wavelength. A prototype spectral absorption meter has been tested in situ (Moore et al., 1992; Zaneveld et al., 1992). The device has been used to determine the concentration of chlorophyll, detritus and cyanobacteria at sufficiently high frequencies to delineate centimeter-scale structures.

Technology to determine the species composition of plankton communities.

A major conclusion of the discussions at the U.S. GLOBEC Acoustical Technology Workshop (GLOBEC Report No. 4, 1991) was that while acoustical methods are very good for quantifying the abundance and distribution of plankton, they cannot currently identify organisms to species level. Identification to species is desirable within the context of plankton studies generally, and U.S. GLOBEC specifically, because animals of similar size but different taxonomy behave and perform differently (e.g., Paffenhofer and Stearns, 1988). It became clear during the above-mentioned workshop that optics would be an excellent tool for identification purposes. This led to the statement that "the integration of acoustical and optical technology could yield synergistic benefits and that the technologies are complementary " (GLOBEC Report No. 4, 1991, p.28).

Zooplankton can be organized in various taxonomic groups ranging from phylum to species. We consider here a range from family to genus, species and developmental stage. Morphometric features, mainly body shape and dimensions have been used for computer-automated identification of net collected, preserved zooplankton (Berman, 1991). To identify living organisms in situ several optical approaches have been used with varying degrees of success and identification: cameras (Ortner et al., 1981), video adapted nets (Welsch et al., 1991), video/OPC (Ortner et al., pers. comm.), the VPR (Davis et al., 1992a,b), and the CritterCam (Bergeron et al., 1988).

The Plankton Camera (Ortner et al., 1981) is towed at approximately 2 kt and obtains a silhouette photograph every two seconds. The film records are quantified manually after the cruise. In addition, measurements of temperature, conductivity, depth and fluorescence are obtained every second.

The Video Adapted Gulf III Net (Welsch et al., 1991) was designed for surveys of herring recruitment in the North Sea. A videocamera images organisms as they pass into the cod end of a plankton net; the video is transmitted in real-time via conducting cable to the research vessel. The images enable the identification of organisms from 0.5-20 mm in length to major taxa (Schulze et al., 1992). This device is expected to be commercially available within one year.

The Video/OPC (Ortner et al., pers. comm.) is a device combining the Optical Plankton Counter (OPC) with a frame-synchronized strobe light silhouette camera. It employs "smart sampling". When the OPC detects the presence of an organism or particle, it triggers the video camera to image the particle during passage through the instrument. The video image is used for specific target (taxonomic) identification. Not every possible target needs to be visualized, but merely a sufficiently large random subsample.

The towed VPR (Davis et al., 1992a; 1992b) has been tested as a prototype. Images from the video cameras on the VPR are digitized and processed in real-time by an image processor and transmitted to a host computer where morphometric indices (e.g., lengths of major/minor axis, area, etc.) are computed, and organisms sorted to major taxa (e.g., copepod, euphausiid, chaetognath, etc.) using discriminant analysis (Berman et. al., 1990).

The CritterCam (Bergeron et al., 1988; Schulze et al., 1992) is a video system developed by Rudi Strickler for imaging plankton distributions and behavior underwater. It is based on modified Schlieren optics and achieves very high resolution at sufficiently long working distances (0.15 to 0.4 m) so that organisms' behaviors are minimally affected by the instrument. It views a field of ca. 6 x 4.5 cm with a resolution of 5 痠 and uses a pulsed (<1 盜) near-infrared diode illumination to freeze the motion of organisms. In moored configuration it can produce images of zooplankton sufficient in quality for the organisms to be taxonomically identified. The CritterCam is presently being readied for commercial production.

The 3-D Bioluminescence Mapping System (Greene et al., 1992; Widder, 1992; Widder et al., 1992) is used to identify and map bioluminescent organisms based on the spatial and temporal patterns of their stimulated bioluminescent displays. Using species-specific bioluminescent displays enables bioluminescent organisms ranging in size from 50 痠 to 1 m to be mapped simultaneously with a single video camera.

Recommendations

A variety of optical sensors designed to study the distribution and behavior of planktonic plants and animals are currently available or in development (Schulze et al., 1992). Support for new optical samplers should be based on their ability to meet a need not filled by those systems already under development. Existing optical sensors should be improved and adapted to sampling platforms that will allow them to sample on the time and space scales needed to meet U.S. GLOBEC's basic scientific goals. This improvement should include interfacing optical sensors for both zooplankton and phytoplankton with acoustical and physical oceanographic sensors to allow simultaneous collection of various types of data from the same parcel of water (e.g., using moorings and drifters).

From the discussions of this working group, it is apparent that the increasing sophistication of video sampling technology is outpacing improvements in post-collection data analysis (e.g. automated image analysis and pattern recognition). Video sensors typically collect 30-60 samples (images) per second. Even though most of these images (depending on the magnification of the image and the depth of field) may be devoid of animals, visual examination of each image soon becomes impractical. Higher priority should be given to the development of systems to analyze the output of the video sensors. Ultimately, we need integrated optical collection, analysis and recognition systems that can measure and identify plankton animals to the species level in real time. However, a more realistic near-term goal is the near-real time identification of video images of plankton to major taxonomic groups.