Bullimore et al.: Study of catches in a fleet of "ghost-fishing" pots 
249 
B 
Pot entrance 
One way entrance 
to parlor 
Parlor 
c 
(A) Map of Skomer Island MNR, Pembrokeshire, Wales, UK, showing the location of the ghost fishing experiment, 
(B) a schematic diagram of the construction of a parlor pot, and (C) the configuration of a deployed fleet of pots. 
ming crab ( Liocarcinus puber) and fish were never tagged. 
However, fish were readily identifiable on consecutive oc- 
casions from the injuries sustained while resident in the 
pots. These injuries most frequently occurred as wounds 
on the head region that were sustained as the fish tried to 
escape through the pot meshes. Records were made of the 
presence and condition of bait on each occasion. Pot depth 
was estimated from divers’ depth gauges and corrected for 
the state of tide from appropriate tide-tables. 
Statistical analyses 
Variation in the total resident catch (all species) within 
the pots at each time interval was analyzed by using a 
GLM (general linear modeling) ANOVA with time as the 
independent variable. Catch rates were calculated as the 
number of newly captured animals caught per pot and 
recorded by divers on each consecutive sampling occasion. 
It was not possible to standardize the sampling time inter- 
val because weather conditions and water visibility during 
the winter period were not conducive to regular sampling. 
Because the intersample period and number of pots varied 
throughout the experiment, the catch data were expressed 
as catch-per-unit-of-effort (CPUE) data with the following 
formula: 
CPUE = N i l(E p (t l -t,)), 
where N t - the number of newly caught animals; 
E /( = number of pots fishing; and 
tj — t t = the time interval since the previous observa- 
tion (t t ). 
The existence of a relationship between the decline in 
catch rate with time was determined by using nonlinear 
regression analysis in the SPSS statistical software pack- 
age (SPSS, 1998). These analyses were performed only 
for brown and spider crabs because other animals were 
caught in insufficient numbers for meaningful analyses. 
Preliminary examination of the data suggested that the 
capture rate for the crab species differed with season; 
hence the regression analyses were undertaken by using 
only the data for the first 125 days of the experiment. Fur- 
thermore, there was a long period without direct obser- 
vations from 125 to 270 days after initial deployment of 
the gear. Catch rate does not provide an indication of the 
total actual mortality of individuals associated with the 
ghost-fishing pots. Mortality was confirmed when divers 
observed the remains of individuals and their tags either 
at the bottom of the pot, or on the seabed nearby, because 
both the tags and dead animal fragments were small 
