All measurement values were standardized to equivalent concentrations in the prepared 
subsample where necessary to account for any concentration steps performed for any analytic 
method (see Appendix B). Thus, measurements shown represent those in the subsample water 
after concentration during initial sample collection/preparation. For example, if 10 ml of the 
prepared subsample was concentrated to 1 ml for measurement by microscopy (i.e. lOx 
concentration) and 100 individuals were counted in this concentrated replicate (i.e. 100 
individuals/ml), this value would be standardized to the concentration in the prepared 
subsample by dividing the measured value by the concentration factor (i.e. prepared subsample 
concentration = measured value in concentrated replicate/concentration factor). In this case, a 
value of 10 individuals/ml would be used for comparison against measurements by other 
analytic methods. One outlier measurement was dropped for SGS ATP (Aqua-tools) where a 
large organism was seen in the sample and a very high, concordant measurement was 
recorded; this replicate represents a rare event where one large organism was sampled, while 
this is of concern for ballast water sampling generally, it was excluded here as it is a real 
difference between samples, not a poor measurement, and thus not relevant for comparison 
between methods. 
To compare results for each analytic method versus microscopy (i.e. dissecting and/or 
epifluorescence (with FDA) microscopy), pairwise scatterplots were made that compared all 
measurements taken for each subsample with replicates randomly paired between methods. A 
line of best fit was generated using Deming's regression which accounts for errors in both 
variables (Ripley and Thompson, 1987) and has been shown to give unbiased slope estimates 
for method comparisons (Linnet, 1993). Explicit error values were specified based on the