Caldarone et al.: Biological indices of growth rate and nutritional state of Salmo solar 
293 
r 
Q 
2 - 
I 
W 
w 
ill 
¥ 
TA 
▼ ; 
▼i 
7 11 15 
Days fasted 
19 
23 
B 
< 
z 4 
Q ^ 
Z 
a: 3 
nil 
48 12 16 
Days after refeeding 
Figure 2 
Each arrow represents changes in RNA/DNA 
values of individual laboratory-reared postsmolt 
Atlantic salmon (Salmo salar) that were either 
(A) fasted or (B) refed (fed after 11 days of fast- 
ing). Direction of the triangle is the direction of 
change in the values from (A) day 0 start val- 
ues (when fish were fed) through the number of 
days (23 d) that fish were fasted; and (B) day 11 
values (start of fasting) through number of days 
(16 days) that fish were refed after fasting. The 
open and closed triangles differentiate between 
the two different directions of change. 
pie day were not statistically significant. Within each 
food treatment; DNA/pro did not change from mean ini- 
tial values (Dunnett P=0.295, 0.090, 0.071 fasted, fed, 
refed, respectively). However, repeated measurements 
of individuals indicated that DNA/pro in individual fish 
increased during fasting (Student’s repeated measure 
/-test P=0.033) (plot not shown). DNA/pro values were 
not correlated with growth rates in the 3 individual 
food treatment groups but had a significant, although 
Table 2 
Pearson product-moment correlations (r) between the 
instantaneous weight-based growth rate (per d) of 
postsmolt Atlantic salmon (Salmo salar) and the ratio of 
RNA:DNA (RNA/DNA, pg/pg), RNA and DNA on a pro- 
tein basis (SNA/pro, DNA/pro, respectively, pg/mg), and 
circulating plasma insulin-like growth factor 1 (IGFl, 
ng/mL). Fish were either fed or fasted for 23 days, or 
refed (fed for 16 days after 11 days of fasting). For each 
feeding treatment, boldface type highlights the high- 
est significant correlation with growth rate. *F<0.05, 
**P<0.005, ***P<0.0001. «,=number of fish sampled. 
Variable 
Fed 
Fasted 
Refed 
All data 
n 
19 
19 
17 
55 
RNA/DNA 
0.882*** 
-0.295 
0.853*** 
0.832*** 
RNA/pro 
0.765*** 
-0.190 
0.666** 
0.727*** 
DNA/pro 
-0.354 
0.090 
-0.238 
-0.507*** 
IGFl 
0.502* 
-0.403 
0.543* 
0.661*** 
low, negative correlation when all three groups were 
combined (Table 2). 
Within a sample day, IGFl values between feed- 
ing treatments were not statistically significantly dif- 
ferent (Fig. 3C). As with DNA/protein measurements, 
high variance (SD), coupled with a small sample size, 
yielded very low statistical power to detect differences 
between treatments. Beginning on day 15 and con- 
tinuing until the conclusion of the experiment, mean 
IGFl values in fed fish were significantly greater than 
day-0 mean values (Dunnett P<0.0001). Fasted fish 
showed no change in IGFl with time, either on a dai- 
ly mean (Dunnett F=0.722) or on an individual basis 
(Student’s repeated measure /-test P=0.065, Fig. 4A). 
Mean IGFl values of refed fish increased 12 days af- 
ter food was introduced (day 23, Dunnett F<0.0001), 
and repeated measurements of individuals indicated 
that final IGFl values in refed fish were greater than 
their start values (Student’s repeated measure /-test 
F=0.0001, Fig. 4B). IGFl values were positively and 
significantly correlated with growth rates in the fed 
group, refed group, and the all-data-combined group 
(Fig. 5, Table 2). 
A plot of growth rate vs. RNA/DNA by food treat- 
ment revealed a difference in the relation of RNA/DNA 
to growth rate between the fed and refed groups (Fig. 
6). ANCOVA results confirmed that the slopes of the 
regression lines of the two feeding treatments were the 
same but the intercept of the refed data was signifi- 
cantly greater. Slopes and intercepts of the regressions 
between the other 3 biochemical measures and growth 
rate did not differ between the fed and refed groups. 
Linear growth models containing all combinations of 
WWjnit, RNA/pro, DNA/pro, RNA/DNA, and IGFl were 
examined (31 models). On the basis of AICc values, the 
best candidate model for predicting growth rate in- 
