434 
Fishery Bulletin 99(3) 
Table 1 
Depth, sampling period, length and age range, and number of boats and individuals examined as part of the bluefish sampling in 
southern Brazil during the 1992-97 period. 
Bluefish target species 
Bluefish bycatch 
Gill net Purse seine 
Gill net 
Trawl (pair and otter) net 
Depth (m) 
10-35 
8-100 
270 
8-68 
Sampling period 
Jun-Aug 
May-Sep 
Sep 
Jan-Mar, May 
Length range (mm) 
301-602 
265-711 
650-770 
262-711 
Age range (years) 
1-5 
1-7 
5-10 
1-7 
Number of boats examined 
33 
11 
1 
7 
Number of individuals examined 
853 
233 
13 
60 
(SPSS, 1998). T - tests were used to test for gear- and sex- 
specific length-weight relationships. 
Scales of captured bluefish were removed from under 
the pectoral fin and stored in a coin envelop. The scales of 
1159 fish were mounted between glass slides and fish age 
was determined by following Krug and Haimovici (1989). 
Fish were considered to be 1 year old after the formation 
of the first winter ring and a further year was added to 
the age of the fish for each subsequent ring (Krug and 
Haimovici, 1989). 
Two different readers assigned ages for the fish, with no 
knowledge of fish length or month and gear of capture. If 
two readings agreed, then the age of that reading was ad- 
opted as definitive. If not, the scale was reread after two 
months, concurrently by both readers. If two of the three 
readings were the same, the age of these two readings was 
assumed as definitive. If all three readings differed, the 
scale was omitted from further analysis. On some scales, 
a single reader assigned ages. This method was followed 
with a time lag (2 months) between readings. 
Estimation of growth parameters 
Sample sizes were sufficient for estimation of growth 
parameters from gill-net and purse-seine samples. Ages 
from the direct reading of scales were obtained for 816 
bluefish from gill-net samples and for 212 bluefish from 
purse-seine samples. 
Ricker (1992) pointed out that the inclusion of sampled 
fish taken at different times during the growing season 
should be avoided and that preferably fish collected be- 
tween the formation of an annulus and the start of the 
next year’s growth should be used in analysis. Only data 
obtained during the fishing season were included for esti- 
mation of growth parameters derived from direct reading, 
and data derived from purse-seine samples (the less selec- 
tive gear) were used to compare the growth parameters 
derived from mean observed lengths-at-age and back-cal- 
culated lengths-at-age. 
For back-calculation of growth parameters, 283 speci- 
mens ( 174 from gill-net samples and 109 from purse-seine 
samples) were used. Fish were processed by sex and gears 
used in their capture to avoid bias in results. 
We determined the relationship between fish length 
(TL) and scale radius (S). Krug and Haimovici (1989) pre- 
viously identified a nonlinear relationship between fish 
length and scale radius for the southern Brazilian bluefish 
( TL=aS h ). Back-calculated total lengths at age were deter- 
mined by using the formula of Monastyrsky (see Bagenal 
and Tesch, 1978; Francis, 1990): 
L n = (S n /S) b x TL, 
where L n = the length of fish when annulus “n” was 
formed; 
S n - the radius of annulus “n” (at fish length L n )\ 
and 
b = the slope from the body-scale relationship TL 
on S. 
This nonlinear method for back-calculation assumes that 
the relationship TL on S is of the form TL = aS, b where a 
and b are parameters to be estimated usually by nonlinear 
least squares. 
The periodicity of growth increments was examined by us- 
ing analysis of marginal increments [(S - S n )/(Sn - 
from fish 2-7 years old (mainly 2-3 years old) (following 
Krug and Haimovici, 1989). Because of the limited temporal 
range of samples, no marginal increments analysis was done 
for the months March-April and November-December. 
We derived theoretical growth parameters by fitting 
back-calculated lengths-at-age to the von Bertalanffy 
(1934) growth equation 
L, = ( 1 - exp 
where L t = 
L TC = 
k = 
the length at age t\ 
the asymptotic length; 
the brody growth coefficient; and 
the age when length would theoretically be 
zero. 
Estimation of growth parameters was based upon two 
criteria. Criterion I identified the type of summary statis- 
tic to be used: 
