FISHERY BULLETIN; VOL. 83, NO. 2 



Table l. — Three-way analysis of variance on log-transformed abundances (m~ ) (see Appen- 

 dix). To balance the sampling design, the 13th profile for zooplankton and chlorophyll (6 April) 

 was omitted from the analysis. A significance level of P < 0.01 was used to compensate for 

 multiple testing of the same hypotheses for many taxa. 



Classification of sample 



1. Day vs. night 



2. Before vs. after storm 



3. Deptfi 

 Interaction 1 x 2 

 Interaction 1 x 3 



Interaction 2x3 

 Interaction 1x2x3 



Significance of ANOVA probably indicates: 



Taxon migrated dielly from below 50 m into sampled range. Diel variation in 



avoidance of hose intake would create spunous significance. For very short-lived 



taxa, strong diel variation in birth, death, or maturation could also cause significant 



differences. 



Taxon changed in mean abundance coincident with storm. Cannot distinguish 



advective from biological causes. 



Taxon was nonuniformly distributed 0-50 m in a consistent manner (or avoidance 

 vaned with depth). 



Taxon migrated dielly into sampled range from below 50 m before or after storm, but 

 not both. 



Taxon had some kind of diel migration. If classification 1 was not significant, migra- 

 tion occurred within upper 50 m. Variation in avoidance both dielly and with dep'h 

 could create spunous significance. 



Depth distribution of taxon changed coincident with storm. Change could either 

 result in greater or lesser uniformity with depth or a change in depth of the maximum. 

 Pattern of diel migration of taxon changed coincident with storm. 



Table 2. — Questions, hypotheses, and nonparametric statistical tests for taxa with heterogeneous variances (designated by 

 asterisks in Appendix). A significance level of P < 0.01 was used to compensate for multiple testing. 



Question 



Null hypothesis (Hq) 



Test and comments 



No difference in abundance 

 (m~^), day vs. night. 



No difference in depth of 



median animal, day vs. 

 night. 



No difference in abundance 

 (m ~^) before vs. after 

 storm. 



No difference in range 

 of abundances (m~^) 

 0-50 m. 



No relation between 

 strength of rank 

 correlation between two 

 profiles and timing of 

 these profiles with 

 respect to the storm. 



What taxa migrated dielly 

 from below 50 m into 

 the sampled range? 

 What taxa migrated dielly 

 within the upper 50 m? 



What taxa changed in 

 abundance coincident 

 with the storm? 



What taxa became 

 more or less uniform 

 In vertical distribution 

 following the storm? 

 Did community 

 structure change 

 coincident with 

 the storm? 



Mann-Whitney U test for difference in median abundances (m ) of each taxon, day vs. night. 

 Compare to ANOVA classification 1 . 



For each taxon for which H,, 1 is accepted. Mann-Whitney U test for difference in depth of median 

 animal, day vs. night. Compare to ANOVA interaction 1 x 3. 



For each taxon for which Ho 1 is accepted. Mann- Whitney U test for difference in median abun- 

 dances (m ~^), 29 March-1 April profiles vs. 3-6 April profiles. Compare to ANOVA classification 2. 



For each taxon for which Ho 1 is accepted, short-cut F test on ranges in median profiles 

 (Appendix). Compare to ANOVA interaction 2x3. 



From abundances (m ) of each taxon in each profile, profiles are grouped by dendrogram based 

 on rank correlation coefficients of abundances. 



categories: A. Zooplankton; B. Phytoplankton 

 and protozoa; C. Relations between zooplankton 

 and phytoplankton; and D. Food for larval fish. 

 Figure 1 shows that the storm was not remark- 

 able in the wind records from San Diego, but was 

 quite apparent in the winds at San Clemente Is- 

 land and in records of wave height at La Jolla and 

 Oceanside. The generally lower wind speed and 

 greater variability in speed and direction within 

 each day at San Diego than at San Clemente Is- 

 land are general phenomena (Dorman 1982). The 

 wind at San Diego is probably more typical of the 

 actual wind off Dana Point, while the San 

 Clemente winds are more typical of the offshore 

 condition generating the swell arriving there. 

 Since the energy appearing as wind-induced tur- 

 bulence increases as the cube of the wind speed, a 



doubling of wind speed increases turbulent energy 

 eightfold. 



Following the storm, surface temperatures and 

 the thermal gradient in the upper 30 m were re- 

 duced at the sampling location off Dana Point, 

 though the change in thermal gradient was not 

 apparent until more than a day after the storm, 

 and water temperatures at La Jolla were higher 

 after the storm (Fig. 1). Minimal thermal gra- 

 dients in the upper 30 m were also observed by the 

 second vessel working at Dana Point on the night 

 of 4-5 April. Though there was pronounced day- 

 to-day variation in depths of isotherms, isotherms 

 tended to be shoaler after the storm. For example, 

 the poststorm median depths of 12°, 13°, and 14°C 

 isotherms all were shoaler by 4.7-5.9 m than were 

 the prestorm median depths (Pommeranz foot- 



154 



