FISHERY BULLETIN: VOL. 81, NO. 2 



Venrick (1982) discussed data on the vertical dis- 

 tribution of phytoplankton samples from four sta- 

 tions at one location in the central Pacific. For the 

 present study, counts from samples of 15 and 120 m 

 depths (representing shallow and deep phytoplank- 

 ton associations, respectively) were used to generate 

 values of / between the field samples. Appendix 

 Equations (7) and (9) were used to estimate the 

 magnitude of/ arising from laboratory subsampling 

 error. A predetermined relationship between labora- 

 tory sampling error and mean abundance (Venrick 

 1982) is available from which to estimate the value of 

 q. The parameters of each sample were used to calcu- 

 late the value of/ expected between replicate counts 

 of that sample and the maximum probable range (Ta- 

 ble 2). For the 15 m samples, one-half of the indices 

 observed between field samples fall within the range 

 expected from the equations. At least for these sam- 

 ples, it appears that differences between samples in 

 the field may be largely attributed to handling and 

 counting errors. For the 120 m samples, none of the 

 observed indices fall within the expected range. At 

 this depth there appear to be "real" differences be- 

 tween field samples. 



The indices observed at 120 m are lower than those 

 at 15 m. The extent to which this is due to hetero- 

 geneity of species abundances, as opposed to shifts 

 in number of species, diversity, or total abundance, 



may be assessed by calculating the standardized / 

 value: 



T = i/'i 



where / is the observed value and I is the maximum 

 expected value calculated from Appendix Equation 

 (7). For each observed value of/, two values of/ are 

 available, one from each sample. When two samples 

 are similar in species content, a representative value 

 of/ may be obtained by calculating a new value of/ 

 from pooled data. This is time consuming and, when 

 samples are dissimilar, the resultant value of/ may 

 not represent either of the original samples. In 

 general, it seems preferable to use the mean of the in- 

 dividual / values. 



The comparison of standardized /' values for the 

 phytoplankton data is presented in Table 3. In five of 

 the six cases, the/' values at 120 mare lower than the 

 corresponding value at 15 m. This shift in T values 

 with depth cannot be attributed only to changes in 

 number of species or diversity. Assuming no depth- 

 related change in the laboratory error, this indicates 

 an increase in the spatial or temporal variability of 

 abundances at greater depths. In the complete 

 analysis (Venrick 1982), the source of this hetero- 

 geneity is postulated to be vertical displacement of 

 vertically stratified populations. 



Table 2. — Similarity index (/) observed between replicate field samples compared 

 with maximum expected index calculated from Appendix Equation (7). The prob- 

 able range is based on the equation for 95'/c confidence interval using a variance 

 estimated from Appendix Equation (9) and the largest likely fl from Figure 4. All 

 samples were collected near lat. 28 N, long. 155°W. 



A. Predicted Laboratory Bias 



Sample 

 depth 

 and no. 



Max- 

 imum 



Date 



H' 



° 2 ('), 

 (X10- ) 



Probable 

 range 



B. Obse 

 from 



rved / between field samples. Underlined values are those within the expected range if / values 

 replicate counts from same sample. 



382 



