Methods of identifying and counting zooplankton in preserved samples have changed little since Johannes Muller first towed a fine-mesh net through the ocean more than 100 years ago. Instrumentation basically consists of a microscope and a ruled counting chamber. Because time of analysis for a single sample ranges from hours to days, delay is inevitable. We are attempting to modernize the procedure, using improved image formation devices and computerized pattern recognition techniques. Total counts and size-frequency distributions can now be made in minutes with a simple, processor-controlled vidicon system, but calibration remains a problem awaiting sharpened contrast by optical edge enhancement and spatial filtering. Major species of North Atlantic zooplankton can be accurately classified to group (e.g. copepods, fish eggs, fish larvae, cladocerans, chaetognaths, euphausiids) by discriminant analysis of simple morphometric relations (length:width:perimeter:area). Identification of commonly occurring copepod species appears theoretically possible for samples taken with a 333 μm aperture net. Syntactic analysis, a complementary approach to pattern expression, may solve difficult problems of classifying developmental stages. Automated image analysis with improved photo-optical systems should meet present day needs.
[1]
A. Hart,et al.
The Abundance, Seasonal Occurrence and Distribution of the Epizooplankton between New York and Bermuda
,
1962
.
[2]
W. Uhlmann,et al.
Automated Phytoplankton Analysis by a Pattern Recognition Method
,
1978
.
[3]
John H. Steele,et al.
The Structure of Plankton Communities
,
1977
.
[4]
H. Edgerton,et al.
Silhouette photography of oceanic zooplankton
,
1979,
Nature.
[5]
C. Price,et al.
Automatic sorting of zooplankton by isopycnic sedimentation in gradients of silica: Performance of a “Rho Spectrometer”
,
1977
.
[6]
H. Bigelow.
Plankton Of The Offshore Waters Of The Gulf Of Maine
,
1926
.
[7]
K. S. Fu,et al.
Syntactic (Linguistic) Pattern Recognition
,
1976
.