292 
Fishery Bulletin 11 6(3-4) 
areas could increase understanding of trawl-survey se¬ 
lectivity, thereby improving model estimates of stock 
abundance. 
Length composition data are among the fundamen¬ 
tal sources of information used to assess fish popu¬ 
lations (Ono et al., 2015), and sampled lengths ide¬ 
ally would represent the true distribution of lengths 
in a population. Length distributions are used to es¬ 
timate critical population parameters (e.g., growth, 
mortality, recruitment), and the selectivity of fishing 
gear or scientific sampling methods. Length-depen- 
dent selectivity values are estimated from the fit of 
a stock assessment model to trawl survey data; se¬ 
lectivity can be thought of as a function of the avail¬ 
ability of all lengths in the population to the trawl 
gear and the efficiency with which the gear samples 
those available lengths (Sampson, 2014; Weinberg et 
al., 2016). Selectivity also can be considered as the 
probability of a fish being sampled in relation to its 
length (Maunder et al., 2014). The selectivity func¬ 
tion relates the index of abundance from the trawl 
survey to the estimate of total population abundance 
from the stock assessment model, and can interact 
with related model parameters, such as growth and 
natural mortality. Therefore, appropriate specification 
of selectivity is critical for reliable model outputs, 
evaluation of stock status, and resulting management 
recommendations (Maunder et al., 2014; Sampson, 
2014; Weinberg et al., 2016). 
Submersible surveys of demersal rockfishes that are 
most abundant in deepwater, untrawlable habitats can 
provide non-extractive, fishery-independent estimates 
of abundance, size composition, and biomass for stock 
assessments, e.g., cowcod (Yoklavich et al., 2007; Dick 
and MacCall, 2014); and yelloweye rockfish (O’Connell 
et al. 2 ). Such surveys provide spatially explicit data 
that reveal patterns in abundance, size, and biomass, 
as well as habitat associations and community struc¬ 
ture that are not possible with other survey methods 
(Yoklavich et al., 2000; Yoklavich and O’Connell, 2008; 
Wedding and Yoklavich, 2015). Length composition 
data from submersible surveys of rockfishes in areas of 
untrawlable habitat can be used to assess the extent to 
which length data from trawl surveys represent these 
populations on a regional basis, and to provide infor¬ 
mation to aid stock assessors with choosing a function 
that best represents trawl survey selectivity for a given 
species. 
In this study, we examined length data collected 
off central California in trawl surveys and from sur¬ 
veys conducted with a manned submersible in nearby 
untrawlable areas. Our objectives were to compare 
length distributions of demersal rockfishes sampled in 
these two surveys, to evaluate the extent to which they 
might differ, and thereby to inform trawl survey selec- 
2 O’Connell, V., C. Brylinsky, and D. Garble. 2003. Demer¬ 
sal shelf rockfish stock assessment and fishery evaluation 
report for 2004. Alaska Dep. Fish Game., Reg. Inf. Rep. 
1J03-39, 36 p. [Available from website] 
tivity functions used in stock assessments for selected 
species. 
Materials and methods 
Our study area was located off central California with¬ 
in the region bounded by latitudes 36°N (just south of 
Big Creek) and 37°N (Davenport), which was the geo¬ 
graphic extent of the most recent submersible surveys 
conducted during a 7-yr period 2003-2009 (Fig. 1). We 
chose the period 2003-2009 for our study because our 
initial examination of length data for several species 
sampled during the trawl survey indicated that we 
would need to combine data from multiple years to en¬ 
sure adequate data for comparison. 
Submersible surveys of fishes and habitats 
were conducted with the 2-person Delta (Delta 
Oceanographies, 3 Torrance, CA) during daytime hours 
(typically 0700-1700) between late August and early 
November in years 2003, 2004, 2007, 2008, and 2009. 
Surveys of a total of 919 strip transects 2 m in width 
and averaging 248 m in length (standard deviation 
[SD] 54.4) were conducted at depths ranging from 24 
to 326 m in submarine canyon and continental shelf 
locations. Strip-transect surveys of 10-min duration 
were directed by a scientific navigator aboard the sup¬ 
port FV Velero IV and were located in areas of rocky 
substrata determined from maps of bathymetry and 
interpreted seafloor habitat (Monterey Bay Aquarium 
Research Institute, website; California State Univer¬ 
sity Monterey Bay Seafloor Mapping Lab, website; 
Yoklavich et al., 1997; Eittreim et al., 2002). The posi¬ 
tion of the submersible was displayed in ArcGIS, vers. 
9.0-9.3 (Esri, Redlands, CA) and tracked at 1- to 3-s 
intervals with an ORE Trackpoint II ultra-short base¬ 
line (USBL) acoustic system (EdgeTech, West Ware- 
ham, MA) and WinFrog software (Fugro, Leidschen- 
dam, Netherlands). The length of each transect was 
estimated either from the edited and smoothed USBL 
navigation data, or from a MiniRLG2 ring laser gyro¬ 
compass (Teledyne TSS, Watford, UK) and NavQuest 
600 Micro Doppler Velocity Log (LinkQuest, Inc., San 
Diego, CA) mounted on the outside of the submersible. 
Details about the Delta survey vehicle, its associated 
equipment, and visual survey methods are described 
by Laidig and Yoklavich (2016) and Yoklavich and 
O’Connell (2008). 
From inside Delta, a pilot and a scientist conducted 
the transect surveys. The pilot operated the submers¬ 
ible within 1 m of the seafloor at a speed of 0.3-0.5 m/s 
(0.5-1.0 kn), while the scientist identified and counted 
all fishes within the transect, and estimated their total 
lengths (TL) to the nearest 5 cm by direct observation 
in situ. A video camera and lights (Laidig and Yoklav¬ 
ich, 2016), mounted externally on the starboard side of 
3 Mention of trade names or commercial companies is for iden¬ 
tification purposes only and does not imply endorsement by 
the National Marine Fisheries Service, NOAA. 
