Methodology:
Plankton samples were collected from discrete depths during CTD casts using a high-capacity, diaphragm pump with an intake hose mounted to the CTD Rosette. Subsamples of sample contents were manually counted, identified, and measured in the laboratory

Sampling and analytical procedures:
For each sample, the pump was run for a timed period of nominally 5 min. The flow rate of the pump was usually measured immediately prior to sampling, by recording the time required to fill a barrel of known volume. Samples were preserved and stored in 5% buffered formalin. The laboratory protocol was to separate samples into two size fractions by passing them through large (200 or 500 μm) then small (64 μm) mesh sieves; split, dilute, and sub-sample each size fraction as needed; count and identify all plankters (minimum n = 50) in each subsample; and estimate the length and width of exactly 100 plankters (the first 50 from the large and small fraction, respectively) to the nearest 50 μm using a dissecting microscope. Three replicate small size fractions per sample were processed starting in 2006.

Data column "vol" was calculated as follows:
vol = length(unique(replicate)*aliquot/beaker/2^splits*50/s_per_50l*d_min*60*1e-3

Data column "flag" contains the following codes:
flag 1: length & width were not measured, but estimated from similar samples
flag 2: shallow & deep samples not labeled, but deduced from sample contents
flag 3: depth estimate less accurate than usual (± ~3 m)
flag 4: time estimate less accurate than usual (± ~15 min)
flag 5: pump flow rate estimated by nominal flow rate