FISHERY BULLETIN: VOL. 87, NO. 3, 1989 



that area. Thus, we calculated an "expected" 

 biomass of euphausiids at the end of 1967, £„, 

 from the mean biomass before the start of the 

 fishery, Ef,. The maximum expected increase 

 was 50% in the northern California inshore area. 



To test for significant differences by two- 

 sample i-test between the means of nonnormal 

 data sets, we would transform to logarithms and 

 verify that this normalized the data (as was true 

 for our data sets) before performing the test. If 

 the means of the log-transformed data are E* 

 and Eh*, the critical value for the ^test, for 

 given numbers of samples, then indicates how 

 much E* must exceed E^ for the difference to be 

 significant, given the variance around E* and 

 Eg, and thus defines a critical ratio, (EalEh)cr, 

 for nontransformed data (see Appendix A). Con- 

 versely, we can ask how many samples would 

 need to be analyzed from a particular area such 

 that EJE„ > (EJEh)„- 



We did this calculation for the entire areas, 

 and the inshore portions, of northern Cahfornia 

 and Oregon. In order to detect significance, we 

 would have had to analyze between 80 and 1,600 

 samples for each block of time within an area, 

 depending on the area. Therefore, the numbers 

 of samples we analyzed, and the data set from 

 Oregon, were insufficient to detect the simplest 

 expectation. 



Next, to normalize the euphausiid biomasses, 

 we multiplied the ratio, E„/Eh, by the compar- 

 able ratio of means of observed, log-trans- 

 formed, small zooplankton or copepod biomasses 

 to obtain a ratio of euphausiid biomasses, 

 (Ea/Eh)corr, Corrected for the environmental or 

 climatic change reflected in these biomasses, 

 which increased significantly off northern Cali- 

 fornia (Fig. 3), and compared {Ea/Ef,)corr to 

 {Ea/Eh)cr- In the Oregon data set, {E,JEh)c,„r < 

 {Ea/Eh)ci; the expected change could not have 

 been detected with the available data. In both 

 inshore and total areas of northern California, 

 however, (Ea/Eh)corr > iE„/Eh)cr meaning that 

 a t-test should have been able to detect a signifi- 

 cant increase in biomasses of euphausiids in 

 northern California due to the combined effect of 

 environmental change and the whiting fishery, if 

 only these two factors operated in the most 

 simple, additive fashion. 



DISCUSSION 



Our results, like those of Colebrook (1977), 

 show considerable similarity in the large-scale 

 geographic and interannual variations in biomass 



of euphausiids (at least those caught by the 

 CalCOFI net at night) and smaller zooplankton. 

 It is also clear that the impact of a chmatic event 

 hke El Nino greatly exceeds, on these scales, 

 any effect of the whiting fishery. 



The failure to find greater biomasses of whit- 

 ing prey after the beginning of the fishery in the 

 Cahfornian area north of Monterey, or off cen- 

 tral Oregon, could result from undersampling, 

 from a mismatch between the effect of the fish- 

 ery on the whiting population and the zooplank- 

 ton sampled by plankton nets, and/or from the 

 complexity of the ecological relations affecting 

 euphausiid biomass. 



"Undersampling" means that variability 

 within each space/time group of samples is so 

 great that we cannot statistically detect differ- 

 ences between groups even though differences 

 actually exist which would be detected if more 

 samples were available. As we have shown, the 

 change in euphausiid biomass calculated from a 

 model of the biomass and food consumption of 

 exploited and virgin Pacific whiting stocks could 

 not have been detected statistically without at 

 least three times the number of samples we had. 

 Further, the actual biomasses of euphausiids off 

 both northern California and Oregon tended to 

 decrease, as did the ratios of euphausiid to small 

 zooplankton (or copepod) biomasses. Therefore, 

 we doubt that simply analyzing more samples of 

 the same kind (i.e., from the same sampling pat- 

 tern, using the same gear) would demonstrate 

 the anticipated increase in the biomass of 

 euphausiids. 



Because CalCOFI stations north of San Fran- 

 cisco were not sampled after April of 1960 during 

 that decade and because the whiting fishery was 

 centered off Oregon and Washington, the sam- 

 ples in our northern California area (Fig. 1) were 

 too far south to be ideal for this analysis, as well 

 as extending too far offshore. The transect off 

 Oregon was better placed latitudinally, but the 

 number of samples in the inshore zone (where 

 the Pacific whiting fishery was conducted) was 

 rather small. Therefore, our effort to increase 

 the number of samples to be analyzed resulted in 

 inclusion of areas outside that where the preda- 

 tors had been reduced by the fishery; we were, 

 in a sense, trying to detect advection or diffusion 

 of the effect into a larger area. 



Ecological complexity may have buffered the 

 response to a reduction of a predator such as 

 whiting in ways that do not amehorate climatic 

 effects. The relatively simple outcome — that 

 euphausiids became absolutely or relatively 



640 



