VENRICK: SMALL-SCALE DISTRIBUTIONS OF OCEANIC DIATOMS 



age of two variances, etc. The average varian- 

 ces were plotted against the sampling interval, i. 



In the case of one species, A^. turgiduloides, 

 this technique revealed a periodicity of si^ with 

 peaks separated by 1-mile increments of the 

 sampling interval (Figure 1 b-d). This indi- 

 cates a pattern of heterogeneity on a scale of 

 1 mile which was not apparent from a direct 

 plot of abundances (Figure la). The periodi- 

 city was best developed at 35 m (runs test sig- 

 nificant at P < 0.001) where it centered about 

 the population variance, as measured by the total 

 variance of the 10 samples. The periodicity was 

 also highly significant (runs test, P = 0.01) at 

 10-m depth. At 50 m the variance showed sig- 

 nificant periodicity (P = 0.10) only when the 

 sample from substation h was omitted from the 

 calculation. The high population densities of N. 

 turgiduloides, and other species, encountered at 

 substation h were comparable to densities ob- 

 served at shallower depths, and may represent 

 another scale of patchiness imposed upon the 

 deeper populations by vertical mixing. 



The horizontal pattern observed in N. turgid- 

 uloides was most highly developed along the top 

 of the seasonal thermocline, which, at Station 23, 

 extended between 30 and 50 m. Internal waves 

 travelling along the thermocline produce a reg- 

 ular series of vertical displacements, which may 

 occur on a scale of 1 mile. In species with 

 strong vertical gradients of density at the top 

 of the thermocline, such circulation patterns 

 would produce regular horizontal fluctuations of 

 abundance, such as were observed in the present 

 study. The effect of vertical displacement on 

 the less strongly stratified species may have 

 been obscured by their horizontal variations. 



This technique has been successfully used to 

 investigate patterns of terrestrial vegetation 

 (Grieg-Smith, 1964) . Once scales of heterogen- 

 eity have been defined, those environmental pa- 

 rameters that vary on scales of similar magni- 

 tude may be sought as possible determinants of 

 the species patterns. Because this approach is 

 not limited to factors which can be measured 

 simultaneously, it is very flexible. It is appli- 

 cable not only to parameters in eflfect at the time 

 of sampling, but also to those whose effect on 

 phytoplankton was exerted some time in the 



past, and which cannot therefore be directly cor- 

 related with abundance. It may for instance 

 prove to be a useful tool for examining the ef- 

 fect of vertically migrating herbivores on the 

 standing stock of phytoplankton. 



SUMMARY AND CONCLUSIONS 



Of the distributions examined in the present 

 study, less than half showed significant aggre- 

 gation. For these species the 90% confidence 

 interval about a single sample, x, could be esti- 

 mated from the interval 0.3x — 3.2a:. This ex- 

 pression was conservative for the nonaggregated 

 species. 



The inability to establish contagion for the 

 majority of the species investigated in the pre- 

 sent study does not prove randomness on this 

 or other scales. However, the prevalence of 

 nonaggregated distributions lends support to the 

 hypothesis that the oceanic environment is less 

 complex than that of the nearshore region. In 

 the oceanic environment, the numerous process- 

 es which bring about local variations in abund- 

 ance of phytoplankton appear to proceed more 

 slowly relative to the randomizing turbulent 

 processes. In such an environment, only the 

 most important local processes produce a mea- 

 sureable effect, and, thus, these may be rela- 

 tively easily isolated for further study. 



ACKNOWLEDGMENTS 



I am grateful to Professor E. W. Fager for 

 his help with the statistical analysis, and for 

 his criticism of the first draft of this paper. 



The work was based on part of a dissertation 

 submitted in partial fulfillment of the require- 

 ments for the Ph. D. degree at the University 

 of California at San Diego (Scripps Institution 

 of Oceanography) . The work was supported in 

 part by Scripps Institution of Oceanography and 

 the Institute of Marine Life Research Program, 

 the Scripps Institution of Oceanography's part 

 of the California Cooperative Oceanographic 

 Fisheries Investigations, which are sponsored by 

 the Marine Research Committee of the State of 

 California, and by the National Science Founda- 

 tion Grant GB 2861. 



369 



