4-92 



by the other method. As with all other taxa, the 1/2 m net had the 

 lower detection threshold; like copepod and Cirripedia nauplii, the high 

 equivalence value (14,437/in ) was exceeded only in peaks of abundance. 

 The scatter-plot was distinctly non-linear and deviation from the 

 regression line by an order of magnitude or more was common. 



To summarize the comparability study results, it was clear 

 that the two methods yielded radically different density estimates. We 

 believe that the CB method was less accurate, underestimating most taxa 

 when densities were near or below detection threshold densities (100- 

 200/m ) and overestimating peak densities, which were generally those in 

 excess of the equivalence value (Appendix Table 4-1) . Underestimates are 

 attributed to patchiness of the zooplankton assemblage and low sample 

 voliomes from the small net; overestimates are explained by patchiness 

 and inaccuracies in the estimation of sample volume. Since the CB and 

 1/2-meter net sample different volumes of water they are likely to result 

 in different estimates of population densities that are reflective of 

 the dispersion characteristics of the species being smapled. Both 

 methods are also limited in accuracy by net clogging and by the error 

 resulting from subsampling in the laboratory. We conclude that the 1/2-meter 

 net was the best practical sampling method for application to zooplankton 

 assemblages of New Haven Harbor. The Clarke-Bumpus net did not meet the 

 basic guideline for plankton sampling recommended by EPA (1973) . 



