Whenever possible,samples were collected within the mixed layer, in the thermocline, and between the thermocline and the bottom. Samples were preserved in buffered formalin. In the laboratory, each sample was washed free of formalin on a 53 μm sieve, transferred to a 250 ml beaker, and, with the aid of an electronic balance, made up to 200 ml (200 g). The sample was homogenized
by vigorous haphazard stirring, and a 12 ml subsample was removed with a Stempel pipette (Peterson et al. 1979, Omori & Ikeda 1984). Multiple 12 ml subsamples were counted until at least 100 individuals of the most common organisms were enumerated. This yielded a sample standard deviation of ca. 10% for the most abundant organisms (including the subjects of this study) and between 10 and 20% for the less common species (Venrick 1978). To test the subsampling technique, we compared the number of organisms in 4 samples by subsampling and by counting the entire sample. No statistically significant differences (Mann- Whitney U-test, p > 0.05) were found between the concentrations of organisms determined by the 2 methods, indicating that the subsampling technique adequately described the samples.