382 
Fishery Bulletin 99(2) 
Distribution 
square=516, 
Table 1 
(%) of length groups (10-cm intervals) of southern bluefin tuna across bigeye tuna 
»=8416, df=24, <0.001). 
(BE) indices (Pearson chi- 
Length (cm) 
BE indices 
Total no. 
Total 
0.0-0. 2 
0.2-0. 4 
0. 4-0.6 
0.6-0. 8 
O 
cq 
o 
140-149 
13.3 
6.7 
13.3 
26.7 
40.0 
15 
100.0 
150-159 
2.8 
10.1 
17.0 
18.6 
51.4 
247 
100.0 
160-169 
8.7 
15.4 
19.5 
24.5 
31.8 
1019 
100.0 
170-179 
12.7 
25.7 
20.8 
20.8 
20.0 
2442 
100.0 
180-189 
17.9 
26.6 
18.5 
19.3 
17.8 
3520 
100.0 
190-199 
27.0 
23.7 
18.9 
14.7 
15.7 
990 
100.0 
200-209 
35.9 
21.8 
19.6 
10.9 
12.0 
184 
100.0 
No. of landings 2100 
3585 
3876 
4421 
1900 
For each landing we calculated a bigeye (BE) tuna index 
as 
BE index = 
Weight of bigeye / (weight of bigeye + weight of yellow fin) . 
This equation was used as a proxy for the depth of fish- 
ing, with an index of 1 = deep and 0 = shallow. Landings 
were grouped into one of five levels of this index, i.e. 0-0.2, 
0.2-0. 4, etc., and then the length-frequency distributions 
of SBT within landings at each level were compared. 
In order to investigate patterns of distribution of fish 
size with depth, we grouped fish into 10-cm length class- 
es and calculated their relative abundance across the five 
levels of the BE index. Because of uneven sampling with 
depth, the number of fish in each BE index were first 
weighted inversely by the effort (number of landings) at 
each level of the index. 
The ovaries of 475 SBT were collected during monitor- 
ing from 1992 to 1995. These were examined histologically 
for evidence of recent or imminent spawning (Farley and 
Davis, 1998). Spawning fish were classed as those having 
spawned less than 24 hours previously (postovulatory fol- 
licles present in ovary), or about to spawn that day (ova- 
ries containing oocytes at migratory nucleus or hydrated 
stage). Postspawning SBT were identified by the propor- 
tion and type of atretic oocytes present (see details in Far- 
ley and Davis, 1998). Nonspawning SBT were mature fish 
on the spawning ground that were neither spawning nor 
postspawning individuals. 
Chi-square contingency analyses were used to test for 
differences in length classes of SBT, and for differences in 
the proportion of spawning and nonspawning SBT at dif- 
ferent levels of the BE index (the proxy for depth). 
Results 
The length-frequency distribution of SBT caught at five 
levels of the BE index shows a trend of increased propor- 
tions of small SBT with an increase in this index (Fig. 1 ). 
Fish <165 cm ranged from 3.3% of catch at an index <0.2 
to 15.7% at a index >0.8. 
Chi-square contingency analyses indicated significant 
differences in the proportion of length classes with the 
BE index (Table 1, Fig. 2). The chi-square test ignores the 
ordered and continuous nature of the categories, making 
it less powerful than it could be. However, we obtained 
a highly significant test result despite this weakness, re- 
flecting how strong the size-with-depth patterns are. The 
smaller length classes (150-169 cm) were better repre- 
sented in the deep catches (BE index >0.8) than they were 
in the shallow catches (BE index <0.2). Conversely, the 
larger length classes (190-209 cm) were better represent- 
ed in the shallow catches (BE index <0.2) than they were 
in the deep catches (BE index >0.8). Smaller fish were 
more likely to be caught in the deepest sets, which target 
bigeye, whereas the bigger fish were more likely to be 
caught in the shallow sets. Significantly, there is a system- 
atic change in depth distribution with size over the whole 
size range of SBT that occur on the spawning ground. This 
pattern is very clear when comparing the proportion of 
fish caught in shallow (BE index of 0.0-0. 2 or 0.0-0. 4) ver- 
sus deep (BE index of 0. 8-1.0 or 0. 6-1.0) sets for each 
length class. The proportion of SBT caught at the surface 
increases with size (Fig. 3). 
The proportion of spawning and nonspawning fish 
(based on the subset of histological data) was then deter- 
mined for each level of BE index (Fig. 4). Chi-square con- 
tingency analyses indicated significant differences in the 
proportions (Table 2). Spawning fish were better repre- 
sented in the shallow catches than in the deep catches. 
Conversely, nonspawning fish were better represented in 
the deep catches than in the shallow catches. There were 
insufficient numbers of SBT in the smaller size classes 
(only seven SBT<160 cm) to use the histology data to ex- 
amine directly the relation between size and proportion of 
spawning fish or spawning frequencies. Because spent fish 
were rarely encountered on the spawning ground, Farley 
and Davis ( 1998) concluded that they move south soon af- 
