EBELING ET AL.; ANNUAL VARIABILITY OF REEF FISH 



Annual variability (AV) in numbers offish per 

 species was measured as variance in logjo ratios 

 (logR) of estimated numbers per hectare between 

 consecutive years. For each species, we estimated 

 number per hectare by summing mean counts per 

 bottom and canopy cinetransects after correcting 

 canopy means for greater area covered per tran- 

 sect, then multiplying by 47.39, the estimated 

 number of bottom transects covering 1 ha (which 

 approximates the average number per year, 

 44.12). AccordingtoWoldal 1978), logi? = log AT. - 

 log (iV,-i), where Nj is number of individuals for 1 

 yr and N, ^ j is that for the preceding year. The 

 mean log R for an array of species indicates the 

 average net change in species abundance, and the 

 variance of the logi?'s ( AV) measures the scope of 

 change in species abundances. For example, a 

 mean logi? near zero indicates that about as many 

 species increased as decreased in abundance be- 

 tween years, while a relatively low AV shows that 

 increases and/or decreases were generally small; 

 i.e., that annual variability was low. To increase 

 the reliability of R, only species with at least 5 

 individuals/ha per year were included in the anal- 

 ysis (Wolda 1978). Although samples covered 

 more than 2 yr, arrays must appear in calculations 

 only once (Wolda 1978). Thus, we computed AV's 

 for an array of 16 log R's for ratios of species 

 abundances between 1972 and 1971, and for a 

 similar array between 1974 and 1973 (separately 

 for mainland and island study sites). Then, we 

 computed overall AV between the years from the 

 array of 32 logiJ's: those for 1972-71 plus those for 

 1974-73. 



RESULTS 



Yearly sampling yielded 297 and 331 cinetran- 

 sects from mainland and island study sites, and 

 recorded 46 fish species in 21 families, although 

 only 31 species in 11 families were common 

 enough to be analyzed (Table 1 ).^^ On the average, 

 about 35 transects were needed to record 90% of 16 



"Additional species that were rarely recorded include: 

 Heterodontus francisci (Heterodontidae), Cephaloscyllium ven- 

 triosum (Scyliorhinidae), Myliobatis californica 

 (Myliobatididae), Torpedo californica (Torpedinidae), 

 Syngnathus spp. (Syngnathidae), Atherinops affinis 

 (Atherinidael, Cymatogaster gracilis (Embiotocidae — 

 sporadically common at island, see text), Phanerodon atripes 

 (Embiotocidae), Caulolatilus princeps (Branchiostegidae), Gib- 

 bonsia elegans (Clinidae), Coryphopterus nicholsi (Gobiidae), 

 Scorpaena guttata iScorpaenidaei, Sebastes auriculatus (Scor- 

 paenidae), P/euronic/zf/iys coenosus (Pleuronectidae), and Mo/a 

 mola (Molidae). 



species that were filmed in the kelp-canopy 

 habitat (the "canopy assemblage" of fishes), while 

 50 transects were needed to record 90% of 31 

 species that were filmed in the reef-bottom habitat 

 (the "bottom assemblage"). 



Although our primary objective was to measure 

 yearly variability, our analysis revealed sig- 

 nificant differences in species composition, diver- 

 sity, and abundance offish assemblages between 

 canopy and bottom habitats, and between main- 

 land and island study sites. Therefore, we describe 

 the observed spatial differences as a prelude to the 

 account of yearly differences. 



Spatial Differences 



Differences in composition between as- 

 semblages in canopy and bottom habitats were 

 obvious and easily demonstrated. For example, 

 canopy and bottom arrays from all years and both 

 sites were segregated in the cluster analysis 

 based on proportionate species abundances (Fig- 

 ure 2). Canopy samples contained relatively more 

 planktivores and kelp browsers like black- 

 smith, Chromis punctipinnis; kelp perch; blue 

 rockfish; juvenile olive rockfish, S. serranoides; 

 and sehorita (Table 1). Bottom samples contained 

 more bottom grazers and ambushers like pile 

 perch, Damalichthys uacca; black perch; gari- 

 baldi; California sheephead; gopher rockfish; and 

 black-and-yellow rockfish, S. chrysomelas. 



The canopy assemblage was simpler than the 

 bottom assemblage in the sense that more indi- 

 viduals of fewer species occurred in the canopy. All 

 of our 31 common species were recorded in bottom 

 cinetransects, but only 16 were filmed in the 

 canopy (Table 1). Furthermore, while the median 

 number of individuals (33 — corrected for greater 

 volume per transect) recorded in canopy transects 

 significantly exceeded that (24) for bottom trans- 

 ects, the median species count ( 5) was significantly 

 less than that (8) in bottom transects (Wilcoxon 

 tests based on 275 canopy and 353 bottom counts, 

 P<0.005). These differences were reflected in the 

 shapes of the abundance-diversity curves for the 

 two habitats (Figure 3). Those from the canopy 

 habitats sloped steeply, reflecting the fact that 

 only a few species were relatively common there, 

 while those from the bottom habitats had flatter 

 tops, reflecting a more coequal commonness of 

 several species. 



In general, the composition offish assemblages 

 differed markedlv between mainland and island 



365 



