EBELING ET AL.: ANNUAL VARIABILITY OF REEF FISH 



Table 4. — Analysis of variance of kelp-bed fish counts (logj^ 

 transformed) and species counts from cinetransects composing 

 yearly samples filmed during 1971-74 in canopy and bottom 

 habitats. For the two-way ANOVA's, samples were classified by 

 sites (Santa Barbara mainland and Santa Cruz Island, southern 

 California) and years (four sequential Septembers). For the 

 one-way ANOVA's, samples were classified by years for each site 

 separately (Mainland and Island subheads). 



•P==0.02; "P = 0.01; — P<0.001. 



pies, on the other hand, included relatively more 

 black perch, pile perch, and rainbow seaperch, 

 Hypsurus caryi. Whereas one species — the black 

 perch — usually dominated mainland samples, 

 several species — the kelp bass; opaleye; 

 blacksmith; garibaldi; and California sheep- 

 head — were often equally abundant in island 

 samples. This more equitable spread of numbers 

 over several common species resulted in signifi- 

 cantly greater species diversity ih) by increasing 

 the evenness component (J) (Table 3), and is 

 reflected in the flattened tops of dominance- 

 diversity curves (Figure 3). Neither total species 

 (Table 3,S) nor mean number of species per tran- 

 sect (Table 3, species counts) were significantly 

 larger (Table 4) in island samples. 



Mainland and island study sites differed sig- 

 nificantly in certain characteristics of structural 

 habitat (Table 5). Scored relief of reef bottom was 

 significantly greater at the island site, although 

 scored densities of giant kelp and bottom algae did 

 not differ significantly between sites. Even though 

 depth of reef over which bottom transects were 

 filmed did not differ significantly between sites, it 

 was more variable at the island site (Table 5) be- 

 cause the shore there sloped more steeply (Figure 

 4). Discounting 1973, when water at the island site 

 was unusually turbid, underwater visibility was 

 significantly greater by some 2.0 m at the island 

 site (Table 6). Island water temperatures were 

 significantly greater, though only slightly so, in 

 all yearly sampling periods except 1973. 



Yearly Differences 



Species composition of bottom assemblages at 

 mainland and island sites was more uniform 

 (showed greater resemblance among years) than 

 the corresponding canopy assemblages (Table 2), 

 although significantly so only for the island site 

 (^-tests, / between habitats, P<0.05; tau, 

 P<0.002). Consequently, both measures of yearly 

 resemblance (/, tau) of bottom-species arrays 

 within sites were significantly greater than those 

 between sites (Table 2). Furthermore, among-year 

 variances of both resemblance measures for main- 

 land- and island-bottom assemblages were less 

 than those for both canopy assemblages, though 

 significantly so only for island measure tau (F- 

 test, P~0.05). Comparing bottom assemblages 

 only, the island assemblage was significantly 

 more uniform (^-tests, / between sites, P~0.05; 

 tau,P<0.001). 



Fish and species counts also reflected the great- 

 er annual variability of canopy assemblages. We 

 computed coefficients of variation (CV) — 

 percentage ratios of standard deviation to 



Table 5. — Means of habitat variables measured with each cinetransect for all bottom samples filmed during 1971-74 at mainland 

 and Santa Cruz Island study sites off Santa Barbara, southern California. Scored from 1 (low) to 5 (high), plant density, other algae 

 includes all understory forms. Time of day is the 4-yr range of median times at which transects making up yearly samples were filmed. 

 CV is coefficient of variation and CI, confidence interval. Significance levels are from Mann- Whitney [/-tests of rank differences 

 between mainland and island values. 



••p. 

 'n = 



0.01. 



130, excluding unusually low values for 1973 (see Table 6). 



369 



