1000 



100 



ro 



I 



e 



UJ 



o 

 > 



10 - 



Figure 3. — Relationship between displacement volume and 

 ash-free dry mass determined from the same samples, on 5 

 cruises {N = 160). A, Both size fractions (<505 Jim = " - ", 

 >505 Jim = " + ") plotted on a logio-logio scale. B, <505 

 (jLm on a linear scale. C, >505 |xm on a linear scale. Regres- 

 sion equations are reported in Table 1. 



0.1 



10 



100 



800 



600 



400 



200 



20 40 60 80 20 40 



ASH-FREE DRY MASS (mgm-3) 



descriptor at low AFDM values. The F-statistic 

 was inappropriate for testing the goodness-of-fit 

 of a nonlinear curve (Draper and Smith 1981), 

 but the F-ratio reported in Table 1 provided a 

 relative indicator of the adequacy of the regi'es- 

 sion equation. The fitted curve in Figure 3C was 

 merely an empirical fit that was not meant to 

 imply any theoretical relation. Because of this 

 nonlinearity, perhaps caused by the compressi- 

 bility of larger organisms at high biomass levels, 

 AFDM was used for subsequent comparisons. 



Since both displacement volume and AFDM 

 have natural variability, a functional regression 

 (Laws and Archie 1981) would be a more appro- 

 priate statistical model to apply to the data in 



Figure 3. However, the need for a nonlinear 

 relationship for the >505 (Jtm fraction compU- 

 cates the use of a functional regression model. 

 For consistency, the fit for both size fractions is 

 therefore based on predictive regressions with 

 the attendant possibility of introducing biased 

 estimators (Sokal and Rohlf 1981). Users of 

 functional regi'essions should note that the equa- 

 tion for estimating the confidence interval of the 

 regression slope reported in Jensen (1986; equa- 

 tion 14) is in error. 



A comparison of mesozooplankton and macro- 

 zooplankton standing stocks for four pooled 

 CalCOFI cruises is illustrated in Figure 4. The 

 median contribution of the mesozooplankton to 



970 



