370 
Fishery Bulletin 106(4) 
used. Both the Spearman’s p and Kendall’s r had the 
same result of no significance (Table 5). 
Correlations between sablefish and grenadier or rock- 
fish catch rates in trawl gear were either positive or not 
different from zero. A scatter plot of sablefish and gren- 
adier catch rates showed that there were very few hauls 
when no sablefish or no grenadier were caught (Fig. 5C). 
It also demonstrated that when sablefish catch rates 
were relatively high, grenadier catch rates were high 
as well. The positive trend in the data is illustrated 
by the 90% density ellipse. However, the scatter plot 
of sablefish and rockfish catch rates (Fig. 5D) showed 
that there were three distinct groups of data: one where 
there were very few sablefish and many rockfish, one 
where there were few rockfish and many sablefish, 
and one where there were high catch rates of both 
species. The 90% density ellipse showed that the 
overall trend was neutral. The data for the sable- 
fish and rockfish comparison in trawl gear did not 
conform to a normal distribution after the data 
were transformed, although the transformation 
was retained for ease in viewing the data. 
Lengths of sablefish, grenadier, shortraker, 
and rougheye rockfish were similar in trawl and 
longline gear. Because large fish were caught 
by both types of gear, we can assume that both 
gears are catching adult cohorts of fish. Neither 
gear appears to be selecting only small juveniles 
(Fig. 6). 
Discussion 
Negative correlations on the longline surveys indi- 
cated that there was likely competition for hooks 
between sablefish and grenadier and also possibly 
between sablefish and rockfish. The catch rate cor- 
relations on longline gear between these species 
were negative in all six of the management areas, 
with the exception of the correlations for sablefish 
and rockfish in the Aleutian Islands. The compari- 
son of longline and trawl catch rates in the Gulf 
of Alaska demonstrated that, in the same areas, 
depths, and habitats, the relationship between 
sablefish and grenadier and sablefish and rockfish 
catch rates on longline gear were negative, but 
positive for sablefish and grenadier, and neutral 
for sablefish and rockfish caught in trawl gear. 
The lack of negative correlations for trawl gear and 
the presence of negative correlations for longline 
gear indicated that longline gear was not catching 
these species in proportion to their abundance. 
The trawl comparison is important because trawl 
gear is thought to catch fish in proportion to their 
abundance, even if the gear may be size selective 
(Gunderson, 1993). Therefore, hook competition 
is likely the cause of the negative correlations 
between sablefish and grenadier catch rates and 
is also likely one of the factors causing negative 
correlations between sablefish and rockfish catch 
rates. 
If the cause of the negative correlations on 
longline gear was due to differing habitat prefer- 
ences or direct competition, the trawl correlations 
would also have been negative. In the trawl data, 
there were very few hauls when zero sablefish or 
grenadier were caught, demonstrating that sable- 
40 n A 
35 - 
30 - 
3 25 
O 
,<S 20 - 
~G 
03 
§ is ^ 
C5 
10 - 
5 
0 
40 -i 
{ . <*.♦ * ..*.»••. « 
,•**. . » ** .*.♦ . 
. . ..V \ 4 »«,•.♦. .••• • 
B 
35 - 
30 - < 
25 - 
20 - 
15 
10 
5 
0 
0 
«• ♦ ♦♦ ♦ ♦ 
♦ ♦ ♦ • ♦ 
♦ ♦ *♦**/♦♦ 
># ♦ ,*♦ **% ♦♦ ♦ 1 1 , 
* |t •• ♦ 
**5*.\r . 
10 
15 20 25 
Sablefish CPUE 
30 
35 
40 
Figure 3 
Untransformed catch rates, in number of fish caught per 45 
hooks (one skate of gear), for (A) grenadier (giant grenadier 
[Albatrossia pectoralis ]) versus sablefish (. Anoplopoma fimbria) 
and (B) rockfish (shortraker [Sebastes borealis ] and rough- 
eye [ Sebastes aleutianus ] rockfish) versus sablefish caught 
during the National Oceanic and Atmospheric Administration 
Alaska Fisheries Science Center annual longline surveys from 
1979-2003. Catch per unit of effort is abbreviated as CPUE. 
Because multiple data points coincide, a random number was 
added to each point so that they could be spread slightly to 
illustrate data point frequency. Only catches from the East 
Yakutat and Southeast management area from 2003 are shown 
as an example. Only preferred depths were included for grena- 
dier (550-1000 m) and rockfish (300-600 m). 
