ysis, sufficient time for the plankton to reach 

 equilibrium volume and weight (Steedman 

 1976). 



In the laboratory, displacement volumes were 

 determined by using the technique outlined by 

 Ahlstrom and Thrailkill (1963), with slight mod- 

 ifications. All organisms larger than 2.5 cm and 

 nonplanktonic matter, i.e., small adult fishes, 

 juvenile fishes, and seaweed, were removed prior 

 to pouring the sample into a 1 1 graduated cylin- 

 der, with 1 ml increments. After recording the 

 volume, the sample was poured into a cone of 

 0.253 mm mesh suspended over a second gradu- 

 ated cylinder, and allowed to drain until the in- 

 terval between drops was 15 s. The water volume 

 was recorded. The difference between readings 

 was recorded as the displacement volume. Dry 

 weight was measured using the procedure out- 

 lined by Lovegrove (1966). Samples were dried 

 at 60°C to a constant weight (2-5 d); a weight loss 

 of 1 mg or less was considered constant. Samples 

 were weighed to 0.01 mg on an analytical micro- 

 balance. Before analysis, all values were ex- 

 pressed as ml or gm/100 m 3 of water filtered and 

 logarithmically transformed (base 10). 



The geometric mean (GM) regression was used 

 to express the relationship between displace- 

 ment volume and dry weight because it is appli- 

 cable to short series of measurements that have 

 moderate or large variability, where the nature 

 of error sources in the measurements is pri- 

 marily natural, when compared with measure- 

 ment error (Ricker 1973). The convention of 

 using the GM regression for relating pairs of 

 biomass measures was introduced by Wiebe et 

 al. (1975), where a more detailed discussion of the 

 theoretical and mathematical considerations of 

 the GM regression as it applies to biomass mea- 

 sures is presented. 



Using samples collected in 1977, conversion 

 equations for displacement volume to dry weight 

 were calculated using GM regressions for groups 

 of measurements divided according to station 

 and/or time of sampling as follows: 



1) General: All measurements regardless of 

 location or season (1). 



2) Area: All measurements from a distinct 

 oceanic region throughout the year (3). 



3) Seasonal: Measurements within an area for 

 a particular season (18). 



Individual displacement volume readings from 

 1978 samples were converted to dry weight using 



each type of conversion equation. The predictive 

 accuracy of the different equations was calcu- 

 lated by measuring the difference between the 

 estimated and the directly measured dry weight, 

 and expressing this difference as a percentage of 

 the latter. The absolute values of the percent de- 

 viations for each conversion factor were then 

 averaged for each group of seasonal and area- 

 specific samples to determine which equation 

 was most accurate. This method for evaluating 

 different conversion factors was used instead of 

 comparing differences between measured and 

 estimated means because of the cancelling effect 

 very high or low estimates would have on each 

 other. 



Results 



A strong linear relationship exists between 

 zooplankton displacement volume and dry 

 weight (Table 1). Slopes of the GM regression 

 lines were significantly different from zero 

 (P<0.001), and correlation coefficients were 

 high (0.885-0.977) for all lines within each class. 

 The range of slope and elevation values for the 18 

 seasonal equations was significantly wide (P< 

 0.05) to conclude that the different lines were not 

 expressing the same biomass relationship. It was 

 hypothesized that from among these regression 

 lines there might exist discrete groups of signifi- 

 cantly similar lines which could be combined to 

 describe a fourth category of conversion equa- 

 tions. A Neuman-Keuls multiple range test (Zar 

 1974) pinpointed lines which differed signifi- 

 cantly (P<0.05) in slope and/or elevation, but 

 other lines could not be accurately assigned to 

 distinct groups because of overlapping similari- 

 ties. Increasing the amount of data might yield 

 more acceptable conclusions, but it is more likely 

 that these results reflect a gradient of changing 

 trophic conditions which gradually alter the bio- 

 mass relationship from season to season. 



The seasonal class of conversion equations 

 yielded significantly (P<0.05) more accurate 

 estimates than either the general or areal equa- 

 tions. For the 18 groups of seasonal and areal 

 samples collected in 1978, predicted dry weights 

 on the average deviated 15.98% (range: 2.31- 

 34.4%) from the actual values using the seasonal 

 equations, as opposed to 29.27% (range: 12.62- 

 66.34%) and 31.4% (range: 15.90-53.45%) for the 

 areal and general equations, respectively. How- 

 ever, for certain seasons, the appropriate regres- 

 sion equation did not accurately convert dis- 



634 



