294 
Fishery Bulletin 116(3-4) 
classification method (Whitmire, 4 Wakefield 5 ). Untraw- 
lable habitats were considered to be primary and sec¬ 
ondary combinations of at least 1 high-relief type, e.g., 
boulder-boulder, rock-mud, or sand-pinnacle. Trawlable 
habitats were primary and secondary combinations of 
low-relief types, e.g., cobble-cobble, cobble-mud, mud- 
mud, flat rock-sand. Trawlable habitats within a tran¬ 
sect were considered to be proxies for the type of area 
surveyed by a trawl, although these patches were too 
small to be trawled. We noted the occurrence of each 
fish on untrawlable or trawlable habitat patches. 
The current trawl survey has been conducted an¬ 
nually off the U.S. west coast since 2003, from Cape 
Flattery, Washington (48.3°N latitude), to the border 
with Mexico (32.6°N latitude). Detailed descriptions of 
the survey design, sampling allocation, protocols, and 
equipment are provided by Keller et al. (2017). A strat¬ 
ified, random grid design and chartered commercial 
bottom trawlers were used to sample depths 55-1280 
m during daylight (after sunrise and before sunset). 
Cells within the grid were 3.7 km (2.0 nautical mile 
fnmi]) latitude by 2.8 km (1.5 nmi) longitude in size 
and were selected randomly from depth and latitudi¬ 
nal strata. Within a selected cell, the captain of the 
vessel surveyed the seafloor with sonar to find suit¬ 
able areas that were large enough to accommodate a 
15-min trawl haul conducted at a speed of 1.1 m/s (2.2 
kn). Trawlable habitat types were low to moderate in 
relief and included little substrata larger than cobble 
(Wakefield 5 ). 
The trawl net was an Aberdeen-type bottom trawl 
(NET Systems, Inc., Bainbridge Island, WA) with a 
14.0-cm (5.5-in) stretch mesh and 3.8-cm (1.5-in) mesh 
liner that extended from the middle of the intermedi¬ 
ate section to the codend. The spread of the net when 
deployed was approximately 5 m high and 14 m at the 
wing tips. The footrope had a continuous series of 25.4- 
cm (10-in) rubber disks that allowed the net to pass 
over cobbles (Wakefield 5 ). Predetermined species of 
management concern or interest were subsampled ran¬ 
domly for individual length measurements. Depending 
on the species, up to 100 individuals were measured 
(fork length [FL]) to the nearest cm from each haul 
(Keller et al., 2017). 
Trawl survey data were obtained from the NWFSC 
Data Warehouse: (website). We selected trawl hauls 
with the project name “Groundfish Slope and Shelf 
Combination Survey,” with a “satisfactory” performance 
(determined from sensors attached to the trawl net to 
monitor bottom contact and the net opening [Keller et 
al., 2017]), and from latitudes 36° to 37°N and years 
2003-2009, resulting in a total of 139 hauls conducted 
at depths 60-1208 m from June through October. 
4 Whitmire, C. 2017. Personal commun. Northwest Fish. 
Sci. Cent., Natl. Mar. Fish. Serv., NOAA, 99 Pacific St., Bldg. 
255-A, Monterey, CA 93940. 
5 Wakefield, W. 2017. Personal commun. Northwest Fish. 
Sci. Cent., Natl. Mar. Fish. Serv., 2032 SE OSU Dr., New¬ 
port, OR 97365-5275. 
We examined the length data from the trawl and 
submersible surveys for harvested deepwater rockfishes 
that commonly occur off central California within the 
overlapping depth range of the two surveys (55-326 m) 
(Love et al., 2002). Species with at least 50 length re¬ 
cords from each survey were considered for comparison. 
We also considered species that have different orien¬ 
tations to the seafloor (i.e., on-the-bottom dwellers, 
near-the-bottom dwellers) and habitat associations, as 
described in Yoklavich et al. (2000), Love et al. (2002), 
and Laidig et al. (2009). On the basis of these consider¬ 
ations, we selected 4 rockfishes for analysis: greenspot- 
ted rockfish (S. chlorostictus ), a bottom-dwelling species 
that occurs on a wide range of habitats; greenstriped 
rockfish ( S. elongatus), a bottom-dwelling species that 
occurs primarily on low-relief cobble and mud; canary 
rockfish (S. pinniger), a near-bottom species that oc¬ 
curs over high-relief rock; and vermilion rockfish (S. 
miniatus ), a near-bottom species that occurs over high- 
relief rock. 
From submersible transects and trawl hauls (i.e., 
samples) with positive occurrences and length data for 
each of these 4 species, we examined the number of 
samples, total area sampled, and numbers and depths 
of fish measured from each survey. Fish length data 
from depths <55 m in the submersible survey were 
eliminated to match the shallow depth limit of the 
trawl survey. Fish length data from the trawl sur¬ 
vey fell within the 326-m maximum depth of the sub¬ 
mersible survey; therefore none was eliminated. Fish 
lengths measured from the trawl survey were convert¬ 
ed from FL to TL by using conversions from Echeverria 
and Lenarz (1984). Within the 2 surveys, length data 
for each species were weighted by sampling effort. 
For each species, we compared 1) all lengths from 
trawl and submersible surveys; 2) lengths from the 
submersible survey associated with untrawlable and 
trawlable habitats; and 3) lengths from trawl and sub¬ 
mersible surveys associated with trawlable habitats. 
For these comparisons, we plotted lengths as the per¬ 
centage of total frequency, using trawl data binned to 
5-cm increments (bin as the midpoint) to match the 
format of the submersible data, and we added the 
trawl survey selectivity curve (not available for ver¬ 
milion rockfish) from the most recent stock assessment 
to the plots with trawl data. To test whether 2 length- 
frequency distributions came from the same distribu¬ 
tion, we used Pearson’s chi-square two-sample test in 
R statistical software, vers. 3.3.2 (R Core Team, 2016) 
and trawl data binned to 5-cm increments. The means 
of the length data and the 10%, 50%, and 90% quan¬ 
tiles were calculated (trawl data, however, not binned) 
with R statistical software (R Core Team, 2016). 
Results 
The spatial distribution of submersible transects and 
trawl hauls with 1 or more of the 4 species present 
in our study from depths >55 m was fundamentally 
