564 
Fishery Bulletin 96(3), 1998 
scribed for whole otoliths. 
Scales were mounted intact 
between two glass slides and 
observed under a dissecting 
microscope at different magni- 
fications with transmitted 
light. 
Organisms were aged by 
counting the number of growth 
marks found in scales (scale 
age), whole otoliths (whole- 
otolith age), and sectioned 
otoliths (sectioned-otolith age). 
Because a large number of 
fishes occurred within a re- 
stricted length range (Fig. 2), 
age estimates for whole-otolith 
ring counts were made on 50 
randomly subsampled otoliths 
from each 2-cm fish-length in- 
terval (72=1356) (FAO, 1982). 
After whole-otolith age deter- 
minations were made, a strat- 
ified random subsample of 
these otoliths was sectioned 
(n= 151) and the corresponding 
scales were used for ageing ( 10 
per fish). Only nonregenerated scales were used for age 
determinations . 
Two independent whole-otolith age determinations 
were made at different times by two readers. A third 
estimate was then made three months later for those 
otoliths with different ages. Lack of consensus resulted 
in the otolith being discarded as “noninterpretable,” 
and excluded from the growth analysis. An effort was 
made to note the cause for rejection. Sectioned otoliths 
and scales were subject to two reading rounds by one 
reader separated by a three-month interval. 
Estimates of the precision (variation between dif- 
ferent reading rounds for each structure) and of the 
corroboration (variation among structures) were as- 
sessed by comparing the percentage of discordant 
determinations between two data sets (%D) and the 
index of average percent error of Beamish and 
Fournier (%E) (1981). 
Following the method proposed by Boehlert ( 1985) 
of using objective criteria and multiple regression 
models to determine fish age from otolith measur- 
able parameters, an effort was made to predict Pa- 
cific red snapper whole-otolith age by using otoliths 
and fish morphometries. This method, originally con- 
ceived to save time and costs as well as to reduce the 
subjectivity involved in otolith reading, was imple- 
mented on L. peru data and the resulting model used 
to determine the age of fishes whose otoliths could 
Total length (cm) 
Figure 2 
Length-frequency distribution of Lutjanus peru collected by hook-and-line and shrimp 
trawling activities in the southeast coast of Baja California Sur, Mexico. Only fishes 
from trawls, hook and line, and the market were used for the ageing study. 
not be read. Independent variables included otolith 
weight (to the nearest mg), width, length, ventral 
radius, anterior radius, fish TL, GW, their logarith- 
mic transformations and squared terms, and a modi- 
fied condition factor K (gutted weight/ TL 3 ). Data 
from a stratified random subsample (FAO, 1982) from 
the whole-otolith age determinations were used for 
this analysis (n=285). A stepwise procedure with an 
inclusion level ofP=0.05 was used to select variables 
for the model. Homoscedasticity and normality were 
evaluated by residual analysis (Zar, 1984). 
Growth 
Individual length-at-age (whole-otolith and sec- 
tioned-otolith) data were used to fit the von 
Bertalanffy growth function (VBGF): 
L t ^L„(l-e~ KU ~ to)) j 
where L f = length at age t\ 
asymptotic length; 
growth coefficient; and 
age at which fish length would be zero 
if it grows according to the model. 
L 
K 
tn 
Hotelling’s T 2 test (Bernard, 1981) was used to com- 
pare growth parameters. 
