varying concentrations of zooplankton. The community composition may also affect how well 
our size-segregated samples conform to those defined in the D-2 convention (i.e. > 50 pm and 
10-50 pm in minimum dimension). As detailed in our methodology, samples were separated 
with a series of filters and all instruments except microscopy and the Satake Pulse Counter 
measured the sample without any further consideration of whether the organisms contained 
therein were of the expected size. Depending on the flexibility and shapes of organisms 
contained within our samples, it is possible that our samples did not exclusively contain 
organisms of the specified size since these properties can impact the efficiency of filtration. Any 
individuals excluded from microscopy counts based on size would thus not have been similarly 
excluded from measurement by the indicative tools; this could lead to higher estimates for the 
indicative tools versus microscopy. 
In a related matter, it is interesting to note that previous ballast water counting 
protocols for the > 50 pm size class have counted phytoplankton at the same time as motile 
zooplankton using a dissecting microscope, with the assumption that structurally intact 
phytoplankton cells are viable (e.g. First et al., 2015). Since preliminary trials indicated that 
dinoflagellates were present in our samples at high abundance in the > 50pm size class (on 
average 85% of individuals), we modified our analytic methods to include two sets of 
microscopy counts for this size class: (i) motile taxa counted on dissecting microscope, and (ii) 
non-motile phytoplankton counted with FDA 'stain' on the epifluorescence microscope, to 
avoid relying on the assumption that intact cells were viable. While we concede that the error 
rates (false positives and false negatives) of the FDA 'staining' were not quantified during our 
experiment, we assumed they would be similar to those reported in previous studies (e.g.