440 
Fishery Bulletin 99(3) 
Selectivity-related bias from gill-net back-calculation 
is thus incorporated in the growth parameter estima- 
tion. The growth curve calculated from gill-net back- 
calculation reports unrealistic and k. Method 4 (in- 
dividual back-calculated lengths-at-age derived from 
all annuli) for the gill net failed because no feasible so- 
lution could be found for the parameters of the curve. 
High variability of back-calculated lengths-at-age de- 
rived from all annuli (see standard deviations in Table 
3) in the presence of Lee’s phenomenon may have been 
the reason for the unsuccessful attempt to obtain esti- 
mates of the parameters of the von Bertalanffy growth 
curve with this method. 
In our study, the growth parameters estimated from 
individual back-calculated length-at-age derived from 
the last annulus only (method 3) on purse-seine sam- 
ples were considered to be the most appropriate for 
future assessment of the southern Brazilian bluefish 
stock. This gear is less selective than others and data 
comprise a wider range of total length groups and age 
groups. 
Changes in Brazilian commercial fishing operations 
for the bluefish over the last decade have led to chang- 
es in the length and age structure of commercial land- 
ings. From 1976 to 1983 the maximum age and total 
length of bluefish in the fishery were 7 years and 630 
mm TL, respectively, and less than 5% of fish were 
4 years and older (Krug and Haimovici, 1989). Cur- 
rently the fishery lands fish up to age 10, and 16% 
of fish are aged 4 and older. The inclusion of older in- 
dividuals in recent catches is due to the expansion 
of the fishery into deeper areas. In 1990, the purse- 
seine fishery began moving to deeper waters, catching 
a wider range of sizes of bluefish. Hilborn and Walters 
(1992) have pointed out the necessity to understand 
fleet dynamics in order to assess the population dy- 
namics of a species. 
Our mean back-calculated lengths-at-age (calculat- 
ed from the last annulus only) are similar to those 
found by Krug and Haimovici (1989) (except for ages 
1 and 3) (Table 5), but estimated growth parameters 
for the two studies are different. We believe these dif- 
ferences are due to differences in fishing operation be- 
tween the two periods (fleet operated in shallower wa- 
ter in 1977-83 than in 1992-97) rather than to fishing 
pressure. Also, the use of individual back-calculated 
lengths-at-age for the last annulus only (method 3) 
rather than the mean back-calculated lengths-at-age 
for all annuli (weighted mean, method 2, Krug and 
Haimovici, 1989) could have led to such a difference. 
Data from the literature indicate considerable vari- 
ation in length-at-age of bluefish between areas (Ta- 
ble 5). Terceiro and Ross (1993) attributed this varia- 
tion to a sampling bias in available fish and to the 
differential proportion of spawned fishes in collected 
samples (e.g. due to multiple spawning). Champagnat 
(1983) suggested that differences in size at first ma- 
turity could lead to differences in growth parameters. 
The reported size at first maturity of bluefish varies 
from 250 mm TL in South Africa (Van der Elst, 1976) 
