338 
Fishery Bulletin 106(4) 
Tanner crab), C. opilio (snow crab) and P. camtschaticus 
in pot fisheries, and a fixed value of 80% for the discard 
mortality of both red king crab and snow crab in Bering 
Sea trawl fisheries (Siddeek, 2003). 
Estimates for discard-related mortality in fishery spe- 
cies normally require experimental research whereby 
individuals or populations captured in different fishing 
operations or under different conditions are monitored 
for subsequent survival. Mortality can be evaluated 
through tag recovery studies, but this method requires 
handling large numbers of individuals, often for low 
returns. Holding fish or crabs in tanks or cages is a 
good method for direct evaluation of mortality in dif- 
ferent fishing operations and handling, but experiments 
are often limited to relatively small numbers, short fol- 
low-up periods, and a few specific treatments or fishing 
variables. 
A useful predictor for discard mortality may be crab 
condition. Blood and tissue chemistry (e.g., lactate, 
glucose, glycogen levels) have been used to evaluate 
stress in crustaceans (Crear and Forteath, 2001; Har- 
ris and Andrews, 2005; Ridgway et ah, 2006). Recent 
experiments with marine fishes, however, show that 
chemical measures are often poor predictors of mor- 
tality because they typically reach peak values before 
any mortality occurs and because chemical measures 
often respond differently to physiological stress and 
physical injury (Davis et al., 2001, Davis and Schreck, 
2005). Two other types of condition indicators have 
been explored for predicting mortality rates in crabs: 
injuries to the exoskeleton and behavioral impairments. 
In the first category, correlations between externally 
visible physical injuries and mortality were reported 
for Chionoecetes spp. (Rosenkranz, 2002) and red king 
crabs (Zhou and Shirley, 1995). However, internal in- 
juries and bleeding that result in mortality can occur 
without apparent external injuries. Given that limita- 
tion, Stevens (1990) considered both external injuries 
and an index of spontaneous activity, termed vitality, 
as possible correlates with delayed mortality in red 
king crabs and Tanner crabs caught incidentally in 
Bering Sea trawls. The crabs were assessed and held 
for 48 hours in shipboard tanks to observe mortality. 
Logistic models showed that the vitality index was 
useful in predicting delayed mortality, and injury in- 
formation did not significantly improve model power. 
Subsequently, others proposed that impaired right- 
ing behavior provides a sensitive indicator of freeze- 
related stress in snow crabs (Warrenchuk and Shirley, 
2002), and might be used to predict mortality (van 
Tamelen, 2005). However, complex behaviors are diffi- 
cult to quantify at sea because they require space and 
controlled conditions, and often do not yield graduated 
results. As an alternative to assessments determined 
by complex behavioral patterns, Davis and Ottmar 
(2006) recently discovered that easily acquired obser- 
vations on a suite of simple reflex actions can provide 
excellent predictions of mortality in fishes related to 
both physical (i.e., wounding) and physiological (e.g., 
thermal stress, air exposure) injury. 
We hypothesized that an assessment of reflex ac- 
tions in crabs would directly reflect their condition and 
provide a good predictor for mortality, independent of 
external injury. The goals of this study were to identify 
reflexes for potential use in assessing crab condition, 
and then test their ability to predict mortality of Chi- 
onoecetes bairdi and C. opilio captured in Bering Sea 
trawls. Crabs were tested for reflex actions and injuries, 
and monitored for mortality in shipboard tanks. These 
experiments yielded excellent predictions of mortality 
based on reflex actions. 
Materials and methods 
Identification and scoring of individual reflexes 
Laboratory studies were conducted at the Kodiak Fish- 
eries Research Laboratory (National Marine Fisheries 
Service, Alaska Fisheries Science Center) during April 
2007, to identify reflex actions of Chionoecetes spp. that 
could be reliably used in evaluating their likely survival. 
Thirty-two individuals of C. bairdi were collected by 
divers from Kodiak nearshore waters between February 
and April 2007. Males (n - 24) ranged from 79 to 128 mm 
carapace width (CW). Females (n = 8) ranged from 88 to 
98 mm CW. The crabs were transported to the Kodiak 
Laboratory in buckets and placed in large fiberglass 
tanks (1.8 m diam., 1 m deep) where they were held 
with flowing seawater. Temperature ranged 1.5° to 4.6°C 
during the holding period. The crabs were fed twice 
each week with a diet of frozen chopped fish and squid. 
At the time of testing all of the crabs were in new shell 
condition, having molted in the past few months, and 
they were active and feeding. The crabs were marked 
with vinyl spaghetti tags tied loosely around the basi- 
ischium of the second or third walking leg. During the 
preliminary experiments seawater temperature was 
5.0°C in the laboratory, and air temperature ranged 
16° to 18.6°C. 
From prior experience with various crab species, we 
expected that C. bairdi would demonstrate stereotypic 
responses to being lifted and to manipulation of their 
appendages. The goal was to identify simple reflex ac- 
tions that could be evaluated rapidly in the tester’s 
hand (out of water), during shipboard operations, with 
a high degree of reliability. After one day of manipula- 
tions we identified six reflex actions that appeared to be 
reliable (Table 1). These were tested on every individual 
crab on two consecutive days and were scored as strong, 
weak, or as no response. 
Field experiment 
A field experiment was conducted in June and July 
2007 to evaluate the feasibility of using reflex actions 
as predictors of delayed mortality in Chionoecetes spp. 
Trawling operations for this study were conducted on 
the Bering Sea shelf east of St. Paul Island in the Pribi- 
lof Islands of Alaska (57°12' to 57°25'N, 169°30' to 
