Stoner et al.: An assessment of discard mortality for two Alaskan crab species, based on reflex impairment 
345 
injury, provides a neurological integration of both physi- 
ological stress and physical wounding (Davis, 2002). In 
fact, a variety of behaviors such as feeding (and growth) 
(Carls and O’Clair, 1990, 1995; Zhou and Shirley, 1995), 
molting (O’Brien et al., 1986; Carls and O’Clair, 1990), 
and righting behavior (Carls and O’Clair, 1990, 1995; 
Zhou and Shirley 1995; Warrenchuk and Shirley, 2002) 
are useful for evaluating crab condition. However, all 
these complex behaviors require some form of holding 
in seawater tanks for observation. For example, pre- 
liminary experiments at the Kodiak Fisheries Research 
Laboratory confirmed that lack of righting behavior in 
C. bairdi can be a useful predictor of mortality caused 
by both internal and externally visible injuries, but the 
evaluation can take >5 minutes, the behavior is highly 
variable, and yields only a binary response (i.e., yes or 
no). The vitality index described by Stevens (1990) also 
has just two possible outcomes — alive or moribund — for 
live crabs. In contrast, Davis and Ottmar (2006) and 
Davis (2007) have shown that testing a suite of reflex 
actions yields a graduated response variable with excel- 
lent potential for mortality prediction in fishes. They 
called the resulting relationship a reflex action mortal- 
ity predictor (RAMP). Our results show that a similar 
approach is possible for Chionoecetes spp. 
We identified six reflexes in Chionoecetes spp. that 
are stereotypic, repeatable, and easy to assess. These 
reflexes, associated with simple movements of the pri- 
mary limbs, mouth parts, and eye stalks, represent the 
most fundamental and involuntary responses of the 
crabs, and they can be rapidly assessed in hand (i.e., 
without holding a crab in water). The most sensitive 
indicators of trawl-related stresses observed in this 
study were kick and leg retraction. These reflexes were 
generally lost first, followed by leg flare and chela clo- 
sure, and reflexes associated with the eyes and mouth 
were least sensitive. The latter two reflexes probably 
require the lowest energy expenditures and movements 
of the mouth parts play an important role in ventilating 
the gills. Mouth movements were maintained in crabs 
near death and loss of motion could be used as the final 
determination of death (Stevens, 1990; this study). The 
order of reflex loss might shift somewhat under different 
fishing conditions or specific types of injury; however, a 
reflex impairment index, calculated by summing reflex 
actions without weighting individual reflexes (as in this 
study), has the advantage of representing condition over 
a wide range of stressor types in fishes (Davis and Ott- 
mar, 2006; Davis, 2007). The same representation over 
a wide range of stressors is likely for crabs. 
The most important finding of this study was that 
mortality in Chionoecetes spp. was closely correlated 
with reflex impairment. Although others have evalu- 
ated injuries to the exoskeleton in an attempt to predict 
mortality (Stevens, 1990), a thorough assessment of 
exoskeletal injuries is time consuming and minor and 
potentially fatal injuries, such as finely cracked cara- 
paces, are easily missed. On the other hand, internal 
injuries and bleeding that can occur without external 
injury can contribute significantly to delayed mortality 
(Grant, 2003). Rose (1999) reported exoskeletal injury 
rates for red king crabs captured or struck with dif- 
ferent types of trawl footropes, but it is now clear that 
these injury rates may or may not correlate well with 
mortality. Injury and reflex impairment were correlated 
in the present study, but our more general assessment 
of crab condition acquired through testing reflex actions 
provided a better predictor of mortality and an obvious 
improvement over tedious and subjective evaluations 
of shell damage. As observed with fishes (Davis, 2005, 
2007), reflex impairment appears to integrate the ef- 
fects of different kinds of stress and injury that can 
occur in gear encounters and in routine handling and 
discard practices. 
Another key finding of this study was that the reflex 
impairment index provides a relatively universal indi- 
cator of condition and likely mortality for Chionoecetes 
spp. For example, shell condition can affect mortality 
rates in crabs handled in fishery operations and re- 
turned to sea (Kruse et al., 1994). However, our logistic 
analyses showed that reflex impairment indices pro- 
vided the best predictors of mortality in both C. bairdi 
and C. opilio, regardless of shell condition and crab 
gender and size. Similarly, reflex impairment indices 
have provided excellent predictions of mortality for 
several different fish species over a wide range of sizes, 
and a good integration of cumulative stress (Davis and 
Ottmar, 2006; Davis, 2007). 
Reflex action mortality predictors have also proven 
to be robust over a wide range of different stressor 
types, including tow time in a trawl, air exposure, 
and temperature shock, at least for fishes. The present 
study emerged from an interest in evaluating impacts 
of trawling operations on crab mortality rates and we 
believe that the response curves provided for C. bairdi 
and C. opilio should be robust for the kinds of injuries 
sustained under typical trawl operations in Alaska. 
The results, however, should be regarded primarily as 
a first proof of principle for crab species and the best 
possible RAMP models will depend upon additional 
fishing experiments conducted over a broader range 
of stressor types and fishing conditions. For example, 
other variables, such as freezing temperatures and 
windchill stress (Carls and O’Clair, 1995; Warrenchuk 
and Shirley, 2002) encountered in pot fishing warrant 
experimental investigation and may necessitate modi- 
fications to the mortality prediction models. 
Once a RAMP curve is well developed for a species it 
should be widely applicable in fishing experiments de- 
signed to improve fishing practices and reduce discard 
mortality. Instead of the traditional approach, where 
fishing and handling variables are evaluated directly 
through multiday holding, either in deck boxes (Stevens, 
1990) or in sea cages (Grant, 2003), crabs can be evalu- 
ated immediately in the field, without holding, and prob- 
abilities of mortality can be estimated with the RAMP 
model. For example, we have been interested in mortal- 
ity of crabs that encounter trawls, but are not captured. 
In the future, we should be able to use recapture nets 
in front of and behind different trawl components (see 
