170 



Fishery Bulletin 105(2) 



from 9 through 12 October 1993 (boreal fall), between 

 the hours of 07:30 (1 hour after sunrise) and 17:00 (1 

 hour before sunset). Thirty-three strip-transects (2 m 

 wide by 10 minutes in duration) were surveyed within a 

 10 X 12 km study area (Fig. 1). During each transect, the 

 scientific observer made observations from the central 

 starboard porthole while the pilot drove the submersible 

 about 1 m above the seafloor at a speed of 0.4-0.9 knots 

 depending on currents and topography. Three broad- 

 scale strata (hard, mixed, soft substratum), which had 

 been identified from seafloor maps by using geophysical 

 data (Eittreim et al., 2002; Anderson et al., 2005), were 

 sampled at depths ranging from 72 to 252 m. 



Within each 10-minute transect, all demersal fishes 

 within 2 m of the submersible were identified to the 

 lowest taxon, measured (total length was visually es- 

 timated to 5-cm size classes), and counted vocally by 

 the scientific observer. An external starboard mounted 

 Hi-8 video camera simultaneously recorded the seafloor 

 along each transect and the scientific observers' vocal 

 commentary on the audio track. A hand-held sonar gun 

 was used to gauge transect width, and paired lasers, 

 set 20 cm apart and projected into the observers' field 

 of view, were used to gauge fish size. A Pisces Video 

 Plus 11 data-logger {Pisces Design, San Diego, Califor- 

 nia) superimposed time, date, depth, and altitude of 

 the submersible onto the video image. Final fish sizes 

 and counts were derived from the videotape, by using 

 the audio commentary as supporting information. All 

 video analyses were conducted by the same person to 

 reduce between-observer variability. Individual fish 

 that could not be distinguished to species were as- 

 signed to a taxonomic group, for example: to subgenus 

 (e.g., young-of-year Sebastes spp. [YOY], Sebastomus 

 spp. [rosy-like rockfish species]), genera (e.g., Cithar- 

 ichthys spp. [sanddabs], Zaniolepis spp. [combfishes]), 

 family (e.g., Agonidae [poachers], Cottidae [sculpins]) 

 or order (e.g., Pleuronectiformes [flatfishes]). 



Benthic habitat characteristics within each transect 

 (intermediate scale) were categorized and delineated 

 from the videotape. Substratum composition (rocks, 

 boulders [>25.5 cm], cobbles [6.5-25.5 cm], sand, and 

 mud) within a patch was categorized by using the 

 dominant (primary=>50'%) and subdominant (second- 

 ary=>20%) percentages of substratum cover used by 

 Stein et al. (1992) and Yoklavich et al. (2000). For 

 example, a patch comprising >50% rock and >20% 

 boulders was classified as rock-boulder (RB); a patch 

 comprising >70% rock was classified as rock-rock (RR). 

 Patches were delineated from videotape where patch 

 duration exceeded 3 seconds of elapsed video time (i.e., 

 where patch size >1.7 m). Habitat relief within each 

 patch was categorized as flat (0-5°), low (5-30°), or 

 high (>30°). These methods adequately defined interme- 

 diate scale habitat composition and patchiness within 

 transects (i.e., m's-100's m), yet logistically enabled 

 long transects (max. 585 m) to be quantified. To de- 

 scribe fine-scale (<1 m) microhabitat use by demersal 

 fish species, we recorded the type of substratum (rock, 

 boulders, cobbles, sand, or mud) directly beneath each 



fish. This multiscale approach enabled habitat asso- 

 ciations at each scale to be recorded independently of 

 associations at other scales. 



Transect length, independent of submersible speed, 

 was estimated by using the known distance between the 

 lasers (i.e., 20 cm) as a ruler, by counting the number of 

 lengths that occurred sequentially over a 15-s duration 

 within each minute of videotape, and then multiplying 

 by transect duration (i.e., 10 min). Patch lengths were 

 calculated by using the same method but were multi- 

 plied by patch duration (elapsed time per patch). 



Analysis 



The categorical measures of substratum type were 

 recoded as semiquantitative variables. Primary and 

 secondary categories were recoded so that each sub- 

 stratum type within a patch was given a percent cover 

 value of 0%, 20%, 50%, or 70%. For example, rock-rock 

 (RR) was recoded as 70%- rock (50%-i-20%) while all other 

 substratum types scored a value of 0%; similarly boul- 

 der-cobble (BC) was recoded as 50% boulder, 20% cobble 

 and all other types scored a value of 0%. Habitat relief 

 was recategorized as an ordinal variable with values of 

 1, 2, and 3 that corresponded with flat, low, and high 

 relief. The mean and standard error for substratum 

 types and relief, and median depth were then calculated 

 for each transect (broad-scale) and patch (intermedi- 

 ate-scale). Habitat patchiness at the broad-scale was 

 represented by "patch number" — the number of patches 

 within each transect, and "patch size" — calculated as the 

 \ogipatch length) within each transect. Benthic habitat 

 variables, with the exception of patch number and patch 

 size, were .r"^ transformed to improve data normality 

 and linearity between variables. Principal components 

 analysis (PCA) was run on the correlation matrix of the 

 transformed transect-level data to evaluate the validity 

 of the broad-scale strata classifications and to describe 

 the relationship between benthic habitat variables over 

 broad spatial scales. 



To examine the relationship between fish and habi- 

 tat, total abundance and species richness were calcu- 

 lated for all fish species and rockfish species at both 

 transect (transect length x 2 m width) and patch (patch 

 length X 2 m) scales: fish densities were then expressed 

 as numbers per 1000 m- (transects), and 200 m^ (patch- 

 es). To examine the fish assemblage in relation to har- 

 vest potential, we classified species as either commer- 

 cial (e.g., Sebastes paucispinis [bocaccio], S. ruberrimus 

 [yelloweye rockfish], S. flavidus [yellowtail rockfish], 

 Ophiodon elongatus [lingcod], and Microstomus pacificus 

 [Dover sole]) or noncommercial (e.g., S. wilsoni [pygmy 

 rockfish], Rhinogobiops nicholsii [blackeyed goby], and 

 Zaniolepis spp.). We also categorized fishes as small 

 (s20 cm) or large (>20 cm). Individual species and taxon 

 groups were included in analyses when they were pres- 

 ent in more than 5% of all patches. Consequently, 21 

 taxa (15 species and six groups) from nine families were 

 retained for analyses. The data on fish distributions 

 were examined by using histograms and Taylor power 



