Lucena and O'Brien: Effects of gear selectivity and different calculation methods on growth parameters of Pomcitomus sciltcitrix 
433 
54° 53° 52° 5f 
at depths greater than 35 m, the operation- 
al limit of the fleet at that time. Given to- 
day’s increased exploitation of bluefish, and 
the lack of older fish in the earlier study, we 
decided to examine the age structure and 
growth parameters of the bluefish stock off 
southern Brazil. New data are compared to 
previously published information and the 
effects of gear selectivity on the estimation 
of growth parameters are also evaluated. 
Annulus formation occurs in winter (June 
to September) and is associated with lower 
water temperature (Haimovici and Krug, 
1996). Rings do not become visible until 
growth resumes in January or February. 
During peak spawning, January-February, 
the true age of individuals with one ring is 
7 months on average (Haimovici and Krug, 
1992). 
Even though most sampled bluefish have 
usually been caught during the short period 
of the fishing season (June to September), 
the real age of a fish may vary by between 4 
and 9 months from that notionally adopted 
because the bluefish is a multiple spawner 
and because this species’ scale ring forma- 
tion is associated with lower temperatures 
(the exact timing of ring formation varies 
from year to year and can also be influenced 
by environmental phenomena such as El 
Nino events). Hence, because sample col- 
lections were spaced unevenly throughout 
the fishing season for both within-year and 
between-period comparisons, growth rates 
based on changes in observed lengths were 
deemed inappropriate and the estimation 
of growth parameters was based on back- 
calculated lengths. 
Many early writers, using the back-calcu- 
lation technique, made no mention of the 
reason for choosing a particular procedure 
(e.g. Johnson and Saloman, 1984; Barger, 
1990; Vieira and Haimovici, 1993). Valuable 
reviews on back-calculation were carried out 
by Hile (1970), Tesch (1971), Casselman (1987) and Francis 
(1990) and, although this technique is widespread, it still 
does not appear to be well understood. Moreover, there is 
a common tendency to plot only mean, rather than indi- 
vidual, back-calculated length-at-age to produce the growth 
curve (Hilborn and Walters, 1992). The effects of different 
approaches on growth parameter estimates are discussed in 
this paper. 
Materials and methods 
Sampling, age structure, and scale reading 
We sampled commercial landings from three fleets at Rio 
Grande from 1992 to 1997: catch from the gill-net fleet. 
catch from the purse-seine fleet, and bycatch from the 
trawl (pair and otter) fleet (Table 1). We sampled 13 addi- 
tional bluefish from the gill-net fishery targeting wreck- 
fish (Polyprion americanus ), choosing larger (older) fish 
in order to investigate the size of the bluefish that move 
along the shelf break at depths >200 m, where the wreck- 
fish fishery operates. We collected an additional 57 individ- 
uals (75-135 mm in TL) with experimental trawls in the 
estuary of Patos Lagoon in order to calibrate the weight- 
length relationship for juveniles. 
Total length (TL) in mm of 1159 fish (260 to 711 in TL), 
measured to the end of the extended tail were recorded. 
The total weight (TW, in grams) and sex (macroscopically 
determined) of 580 bluefish, ranging in size from 75 to 711 
mm TL, were determined to calculate the weight-length 
relationship, by using nonlinear least-squares regression 
