44 
Fishery Bulletin 1 1 1 (1) 
Broadhurst et al. (2006) provided a thorough review 
of such studies. Although a great number of studies 
estimated mortalities of discarded catch, others dealt 
with mortalities of escaping animals not brought 
aboard the fishing vessel. Broadhurst et al. (2006) 
noted that studies of escaping animals, almost exclu- 
sively fishes, lately have emphasized methods where 
escaping animals are recaptured in cages that are then 
detached from the fishing gear while still at fishing 
depths. Those cages then are moved slowly to shallow- 
er depths, where they are maintained by divers long 
enough to assess delayed mortalities. Earlier methods 
involved capture of escaping animals in auxiliary nets 
before they were brought aboard and held long enough 
to evaluate mortality rates. However, stress and injury 
from recapture and extended towing and holding times 
could have easily masked or exacerbated the effects of 
the escape process, particularly for animals vulnerable 
to skin abrasion damage. More recent methods retain 
the experimental subjects in an environment closer to 
what they would experience after actual escape. The 
cost of these gains is that each collection of affected 
animals requires an extended series of activities that 
are time consuming and labor and resource intensive. 
These time and resource demands greatly restrict the 
number of experimental samples that can be collected 
and held and, hence, the number of experimental fac- 
tors that can be addressed. 
As an experimental subject, crab are significantly 
different from fish for which the in situ capture, trans- 
fer, and holding methods were developed. Exoskeletons 
protect crabs from the type of abrasion to which fish 
are particularly susceptible during net capture and 
crowded holding. As a trawl net approaches, fish con- 
tinue swimming, often to exhaustion, to avoid contact 
with the net and other animals, but crab, being much 
slower, can flee only briefly before being overrun (Rose, 
1995). 
Another difference is how crabs interact with fishing 
gear. Broadhurst et al. (2006), describing research on 
fishes, noted, “Because most experiments have quanti- 
fied escape mortality at the codend, the potential for 
mortalities as a result of collisions and escape through 
other parts of the gear have largely been ignored.” Be- 
cause of the sizes and behavior of Bering Sea crabs and 
the configurations of Bering Sea bottom trawls, most 
crabs escape under the forward parts of trawl systems, 
and interactions typically last only a few seconds as 
the crab passes the components of the net that directly 
contact the seafloor. Rose (1999) studied crab mortali- 
ties after such escapes under the forward sections of 
bottom trawls through assessment of visible injuries to 
red king crab that resulted from passes of crabs under 
different trawl footrope designs. The crabs were recap- 
tured in an auxiliary net fished behind the main foot- 
ropes. A control footrope, suspended with floats to allow 
crabs to pass beneath with minimal damage, also was 
used. A low rate of injuries for control crabs indicated 
that recapture of crabs to bring them aboard could be 
done without greatly increasing injury to crabs. The 
principal limitations of that study were the following: 
1) crabs were not held beyond the initial assessment of 
injuries to observe delayed mortality; and 2) observa- 
tions were limited to crabs that passed under the cen- 
ter section of the footrope, a small portion of the area 
swept during trawling. 
Studying mortality of crabs discarded from trawl 
catches, Stevens (1990) effectively applied a strategy in 
which all subject crabs were assessed for selected con- 
dition attributes and a sample was held long enough to 
relate those attributes to delayed mortality. Since that 
study, such methods have been expanded and improved. 
Davis and Ottmar (2006) used assessment of a range of 
reflexes of Pacific Halibut ( Hippoglossus stenolepis) to 
build a predictor of delayed mortality, the Reflex Action 
Mortality Predictor (RAMP). In a pilot study for this 
project, Stoner et al. (2008) found the RAMP technique 
effective for estimation of delayed mortalities for snow 
and southern Tanner crabs. 
Our research addressed unobserved mortality rates 
for 3 principal commercial crab species of the Bering 
Sea: red king crab, southern Tanner crab, and snow 
crab. We improved methods for collection of crabs im- 
mediately after trawl encounters as used by Rose 
(1999) and applied the RAMP technique as described 
by Stoner et al. (2008) to assess the mortality prob- 
abilities for crabs that passed under the sweeps, wings, 
and central footrope of a commercial groundfish trawl. 
Raised sweeps, which reduce seafloor contact yet main- 
tain herding of flatfishes (Rose et al., 2010), also were 
used at the red king crab sites to evaluate whether 
they would reduce crab mortality rates. Observations of 
control animals collected with identical recapture nets 
but no trawl encounter were used to adjust observed 
mortality rates for effects of capture and handling. 
Materials and methods 
A pilot study conducted in 2007 evaluated the RAMP 
and developed and tested techniques for 1) recaptur- 
ing crabs after encounters with trawl components, 2) 
handling and assessing those crabs on deck, 3) holding 
selected crabs to determine their survival over several 
days, and 4) using the RAMP to estimate the mortal- 
ity probability of each crab (Stoner et al., 2008). Our 
study followed those methods closely, and the following 
description summarizes them and highlights all modi- 
fications made to the methods of the pilot study for our 
later study. 
Experimental tows for southern Tanner and snow 
crabs were made in August of 2008 -111 km (-60 nmi) 
east of Saint Paul Island (Fig. 1). All tows included a 
mix of both species. Red king crab tows were made in 
August of 2009 at 2 sites in Bristol Bay, about 22 km 
(12 nmi) west of Amak Island and ~65 km ( —35 nmi) 
northwest of Port Moller. Operations were conducted 
aboard the FV Pacific Explorer, a 47-m, 1800-hp com- 
