Performance of the human "counting machine": evaluation of manual microscopy for enumerating plankton

In this study, live plankton (≥50 µm in minimum dimension) in samples were counted by three analysts to determine counting rates, accuracy and precision of manual microscopy. Counting rates were compared with sample characteristics (e.g. concentration of dead organisms). In separate experiments, plankton proxies, spherical microbeads (49 and 150 µm in diameters), were added to samples with varying degrees of debris loading, including samples from full-scale, replica ballast tanks. These analyses were used to test the hypothesis that as debris loading increases, counting rate increases and accuracy decreases. Highly concentrated samples and samples with high concentrations of dead plankton resulted in significantly slower counting rates. The recovery of 50-µm microbeads was lowest (75%) in laboratory samples that contained the highest debris load. The recovery of 150-µm microbeads was very high (>98%) in laboratory samples with and without debris. Field samples from the replica ballast tank were highly turbid, and microbead recoveries were low for both microbead sizes. Sample quality, therefore, will affect counting rates and accuracy and will limit the volume of the sample that can be analyzed within short-time windows available for counting live plankton samples.