226 



Fishery Bulletin 88(2). 1990 



Survey analyses 



From the three transects at each site, a mean number 

 ( + 95% CI) of fish per transect was computed for each 

 species and size category. To determine whether den- 

 sities of rockfishes differed among the four habitat 

 types and within the two replicate sites within each 

 habitat type, I used a nested analysis of variance 

 (ANOVA) (Zar 1984) (Fig. 2) that tested two null 

 hypotheses: (1) There is no difference in rockfish 

 densities between habitats (F„o.5( us (habitat types), 4 (.sites)), 

 and (2) there is no difference in rockfish densities be- 

 tween the two replicate sites within each habitat type 

 (F().05(i)4(sites).i6(error))- The degTces of freedom for error 

 was calculated as (4 (types) • 2 (sites)  3 (transects) - 4 

  2) = 16 (Zar 1984, p. 147). To reduce the number of 

 analyses, data were analyzed as a separate ANOVA 

 for each of six seasonal periods using the middle month: 

 Winter 1 = December 1986-February 1987; spring = 

 March 1987-May 1987; summer 1 = June 1987-August 

 1987; fall = September 1987-November 1987; winter 

 2 = December 1987-February 1988; and summer 2 = 

 June 1988- August 1988. Although density estimates 

 were similar within quarters (Figs. 3-6), the sedentary 

 behavior of rockfish could mean that the same indivi- 

 duals were counted in February that had previously 

 been counted in January, which would constitute a 

 repeated sample. Thus to eliminate the possibility of 

 non-independence of data, as I have no evidence that 

 rockfish redistribute themselves randomly from month 

 to month, I used only the middle month of each quarter. 

 Because many months and sites contained zero den- 

 sities and, in some cases, the variances were propor- 

 tional to the means, the data were transformed using 

 a square root transformation (A' ' = square root (A' + 

 0.5)). If a significant difference in densities was de- 

 tected among the four habitat types, the types were 

 then compared using Tukey's multiple range test (Zar 

 1984) to determine which types were different. 



To determine whether densities were different be- 

 tween years at each site, quarterly estimates were com- 

 pared using a Student's t-test (Zar 1984). For these 

 tests, densities from the summer of 1987 were com- 

 pared with the summer of 1988 for each site testing 

 the null hypothesis: There is no difference between den- 

 sities of each site between summer 1987 and summer 

 1988. Similarly, densities from the winter of 1987 were 

 compared with the winter of 1988. I conducted in- 

 dividual tests for copper, quillback, brown, and YOY 

 rockfishes for summer and winter comparisons. 



Results 



Habitat type and site comparisons 



All size categories of copper, quillback, and brown 

 rockfishes and YOY rockfish were seen at the high- 

 relief natural habitat type (Fig. 3). Consistent monthly 

 densities of large (>200 mm) copper (mean 3.7-13.0 

 fish/90-m'^ transect) and quillback (mean 3.0-7.0 fish/ 

 transect) rockfishes were observed throughout the year 

 on the high-relief rocky reefs. Brown rockfish (>200 

 mm) (mean 1.3-4.0 fish/transect) were seen at HRl (Or- 

 chard Rocks) but infrequently at HR2 (Blakely Point). 

 Low densities (mean<1.3 fish/transect) of subadult 

 (80-200 mm) copper, quillback, and brown rockfishes 

 were observed on both reefs until the late summer and 

 fall 1987. YOY (<80 mm) rockfishes were observed 

 (mean 8.0-57.0 fish/transect) from August-November 

 1987 but only on HR2. YOY were not observed in the 

 summer or fall of 1988. After the intlux of YOY dur- 

 ing summer and fall 1987 on HR2, an increase in 

 80-200 mm rockfishes was subsequently observed 

 (mean 6.0-15.0 fish/transect). 



Rockfishes primarily utilized low-relief rocky reefs 

 during the summer, coincident with peak Nereocyt^fis 

 leufkeana and understory cover, and were infre(iuent- 

 ly observed during other seasons (Fig. 4). Large cop- 

 per (up to a mean 6.0 fish/transect) and brown rock- 

 fishes (up to a mean 7.7 fish/transect) were observed 

 mainly during the summer, whereas large quillback 

 rockfish were not seen at LRl and only infrequently 

 at LR2. Low densities (mean 0-3.3 fish/transect) of 

 small copper and quillback rockfishes were seen on 

 LR2, while small brown rockfish were never seen on 

 either reef. YOY rockfishes were observed (mean 16.7 

 fish/transect) during August 1987 on LRl and (mean 

 1.3 fish/transect) on LR2 during October 1987, but not 

 during the summer or October of 1988. 



On the artificial reefs, densities of large copper rock- 

 fish fluctuated throughout the year. During both 1987 

 and 1988, high densities (mean 12.0-56.3 fish/transect) 

 were seen September through May; however, low den- 

 sities (mean 0.3-2.3 fish/transect) of copper rockfish 

 were observed June through August (Fig. 5). Variable 

 densities (mean 0-12.3 fish/transect) of large quillback 

 rockfish were observed on ARl while they were lower 

 in number (mean 0-2.3 fish/transect) on AR2 (Fig. 5). 

 Large brown rockfish were infrequently (mean 0-0.3 

 fish/transect) seen on either artificial reef. Low den- 

 sities (mean 0-5.0 fish/transect) of subadult (80-200 

 mm) copper rockfishes were observed on both artificial 

 reefs throughout the year. Extremely high densities 

 (mean 8.0-420.0 fish/transect) of subadult quillback 

 rockfish were observed on both ARl and AR2, al- 

 though densities were much higher on AR2 (up to 420 



