Bentley-Mowat, 1982, Garvey et al., 2007, Steinberg et at. 2011); conversely, the error 
associated with assuming all intact phytoplankton are viable was completely unknown (Paerl, 
1978). Since dinoflagellates were so abundant in our samples, errors in determining viable vs. 
dead status could have significant impact on the results. While this issue lies outside of our 
analytic methods comparison, it may have significant implications for ballast water monitoring 
studies more generally, suggesting that examination of appropriate methods for enumerating 
viable phytoplankton in the > 50pm size class could be considered further in future work. 
Finally, it is important to highlight that each CFA device has a specific 
calibration/conversion coefficient that relates its results (F v ,F m , or F 0 ) to cell concentrations. 
Consequently, different devices can generate different estimates even if they have the same 
fluorescence reading. In reality, the true relationship between fluorescence and cell 
concentration varies across phytoplankton communities depending on cellular chlorophyll 
content (can vary by a magnitude of 5x due to photo-adaption and an additional 5x due to 
nutrient limitation/saturation) and the size of cells in the community (range of cell volumes for 
10-50 pm group varies 125x)(Veldhuis et a 1,1997). Thus, chlorophyll content per cell is the 
major source of error for estimating cell numbers on the basis of bulk chlorophyll fluorescence 
and variation in calculated cell concentrations among instruments may depend on the similarity 
of the ballast sample community to that used to develop the calibration factor. Nonetheless, 
present results for these methods were promising for both size classes examined here. We 
expect that future studies to examine and quantify the variability in the relationship between 
fluorescence and cell concentration for ballast water communities would provide valuable 
insight.