466 
Fishery Bulletin 96(3), 1998 
the tank, placed in a padded tagging cradle, mea- 
sured, tagged, and released. Each anchor tag was 
inserted between and engaged behind the ptery- 
giophores of the dorsal fin. Fork length, tag number, 
capture depth, geographical position, and date of 
release were recorded for each fish. Only fish judged 
in viable condition were tagged. Sablefish of all sizes 
(20-110 cm) taken by the capture gear were tagged. 
Recovery 
A recovery program was promoted through the use 
of posters, news releases, and letters explaining the 
research and enlisting the cooperation of those who 
might encounter tagged sablefish, particularly fish- 
ermen, fish processors, and members of state fisher- 
ies agencies. Individuals finding a tagged fish were 
requested to return the tag with information about 
the date, location, fishing gear, depth of capture, and 
fish length. A reward and the release history of the 
tagged fish were provided to those who returned tags. 
By the end of 1993, 14,903 recoveries had been made 
(6.82%). The information garnered from these recover- 
ies constitutes the core of this paper. Owing to the wide- 
spread geographic availability of sablefish and the 
broad coverage of the surveys and fisheries, tagged 
sablefish were released and recaptured from locations 
throughout its range in the northeast Pacific Ocean. 
Data analysis 
Our philosophy in this paper is to present the data 
as they are and not to adjust the number of tag re- 
coveries for exploitation rates. Such adjustments 
require knowing the catch and population biomass, 
or standardized effort measures, by area. If these are 
incorrectly specified, the data may be further dis- 
torted rather than corrected. When the recovery data 
are being adjusted over an extremely broad geo- 
graphic range, where dominant gears are different 
and where even the same gears fish differently, such 
adjustments become particularly difficult. 
Because of the high numbers and long time series 
of tag returns, basic descriptive tools appear to re- 
veal large-scale migration and stock-structure pat- 
terns. Accordingly, we used basic descriptive tools 
such as mapping locations of tag recovery. Although 
such visual representations are useful, we also 
needed to aggregate the data so that general pat- 
terns could be more easily discerned. Therefore, we 
divided the range of sablefish (encompassing the 
eastern Bering Sea, the Aleutian Islands, Gulf of 
Alaska, and the west coast of the U.S. and Canada) 
into 27 areas of moderately small scale. We used large 
nominal regions (the eastern Bering Sea, the Aleu- 
tian Islands, and the Gulf of Alaska) to assist in these 
demarcations and also a straight line drawn down 
the Alaska Peninsula passing through 150°W, 60°N 
and 168. 5°W, 53°N and continuing on to 170°W lon- 
gitude. All areas west of 130°W longitude were di- 
vided into areas 5° wide in longitude, and areas east 
of 130°W longitude were divided into areas 2.5° wide 
in latitude (Fig. 1; Table 2). The habitat of adult sable- 
fish includes the 400-m depth contour shown in Fig. 1. 
Once these areas were described we were able to 
tabulate the tag recoveries in a table whose rows are 
the areas of release and whose columns are the ar- 
eas of recovery. We feel this is the single most useful 
piece of information that can be garnered from any 
study of migration based on a tagging experiment. 
We also characterized net movements by mapping 
average positions of release and recovery. The aver- 
age position of tagging and recovery was calculated 
for all fish tagged in area i, for each area i. An arrow 
was then drawn from the area of tagging to the area 
of recovery. The back end of the arrow was in area i, 
and the position of the arrow point provided some 
indication of the average movement of the tagged 
sablefish (i.e. the arrow points to where the tagged 
fish in area i go). Similarly, the average position of 
tagged and recovered fish was calculated for all fish 
recovered in area i, for each area i. The pointed end 
of these arrows were in area i, and the back end of 
the arrow gave some indication of where these fish, 
on average, originated (i.e. the arrow indicated where 
the tagged fish recovered in area i came from). 
Distance traveled was examined by calculating a 
histogram and empirical cumulative distribution 
function (cdf) of the distance traveled between tag- 
ging and recovery. Movement of tagged fish in rela- 
tion to depth was examined by simply tabulating the 
depth of tagging and the depth of recovery. 
Modeling ENSO growth effects 
Modeling approaches to analyzing ENSO growth ef- 
fects was difficult because the best model for this 
apparently simple task was unclear. Assuming an 
initial size at tagging of S p a size at recovery of S 2 , 
and a time at liberty of A t , the most direct model 
would seem to be something like S 2 = S 1 + Afi. How- 
ever, this model ignores the common sense notions 
that the growth increment per unit time would be 
expected to decrease as both Sj and A t increase and 
suggests the model S 2 = S x + A t exp (fi enso + P l S 1 + 
/3 2 4 ? ), where P enso represents an ENSO effect, and ^ 
and (5 0 can be presumed negative. A tagged fish was 
assumed exposed to ENSO if it was at liberty any- 
time during a recognized ENSO event: 1972-73, 
1976-77, 1982-83, 1986-87, and 1991-93. 
