686 
Fishery Bulletin 95(4), 1 997 
Figure 6 
Monthly counts of newly settled A. lineatus on the outer reef flat at 
three sites on Tutuila Island, 1994-95. 
steadily from early summer through the cooler sea- 
son (Fig. 8). These losses resulted in CF values that 
were about 10% below peak summer levels and in fat 
bodies that were about two thirds below peak levels. 
Seasonal changes in five environmental factors 
paralleled changes in fish condition (Fig. 8). During 
the cooler season, nearshore water temperatures 
dropped below maximum summer values (-9%), as 
did available feeding hours (-14%), calm water con- 
ditions (-44%), rainfall (-76%), and daylength 
(-13%). Most physical factors were significantly 
autocorrelated (7 out of 10 comparisons), indicating 
that there is a distinctive seasonal signal in the 
nearshore environment, despite Samoa’s open-ocean 
location near the equator. Monthly CF values for im- 
mature and mature fish were significantly correlated 
with 3 of the 5 physical factors (Table 1). It seems 
clear that the fish were responding to a seasonal 
change, but causative factors are not known. 
Growth 
Otolith-based ages and size determinations of natu- 
rally marked fish provided two estimates of A. lineatus 
growth. Otolith weight was highly correlated with the 
number of annular bands in an otolith (Fig. 9). The 
fish grew rapidly, attaining 70-80% of their total growth 
by the end of their first year, and they were long-lived — 
up to 18 years (Fig. 10). Field estimates of fish growth 
confirmed the rapid early growth but underestimated 
later growth in comparison with the otolith-based de- 
terminations. Similar values of # and were derived 
from both the otolith-based age-length relation (Ford- 
Walford regression: #=0.7 , Lj= 20.3 cm, r 2 =0.93) and 
from size increments of marked fish (Gulland-Holt plot: 
#=0.8, L m = 21.0 cm, ^=0.58). 
However, L x and # were highly dependent on the 
age range of fishes examined (Fig. 11). Iterative Ford- 
Walford regressions of the smoothed mean size at 
age showed that # dropped progressively from 0.8 
for the whole sample (ages 0-12) to 0.1 when juve- 
nile fish ages 0-3 were excluded. Asymptotic length 
(L m ) increased in a similarly systematic manner from 
about 20 cm to 22 cm. Therefore, separate VBGF 
growth curves were generated for the juvenile phase 
(ages 0-3, #=1.1, t Q =-0.2, [L^=18.3 cm]) and adult 
phase (ages 4-12, #=0.12, L=22. 1 cm, t 0 =-15.6), a 
separation based on the age at which 50% of the popu- 
lation was mature (age 4, Fig. 4). The two-phase curve 
captured the precocious growth and attained a larger, 
more realistic that approached the maximum size 
of fish taken in the fishery (see Fig. 3). Using the 
relation that longevity is approximated by 3/# (Pauly, 
1983) and the adult #- value derived for ages 4-12, 
we predicted that the maximum age would be 25 
years, which compared favorably with the observed 
maximum age of 18 years. 
Mortality 
Mortality indices differed among the three size 
classes of naturally marked fish at the Afao site (Fig. 
12). Only 2% of the newly settled fish and 34% of the 
juveniles appeared to survive and grow into the next 
