Berkman et al.: Effects on survival of Oncorhynchus tshawytscha in rivers of Southeast Alaska 209 
variability in the data set. The first RC (RC1) loaded most 
heavily on parr length and fall discharge, and the second 
one (RC2) loaded most heavily on early spring discharge 
and temperature (Table 2). Results from the principal com- 
ponent regression model indicate that the 2 RCs explained 
little of the variation in overwinter survival (adjusted 
R’=0.14, F=1.78, P=0.23). There were no significant rela- 
tionships between overwinter survival and RC1 (P=0.51) 
or RC2 (P=0.12) (Table 3). In contrast, smolt produc- 
tion was significantly related to both RC1 (standardized 
regression coefficient [B]=—0.10, P=0.05) and RC2 (B=0.19, 
P=0.03), indicating that smolt production was enhanced 
when parr were smaller and discharge was lower during 
the previous fall and when discharge was higher and tem- 
peratures were warmer in the spring (Table 3, Fig. 2). 
Marine survival Over the time series (BY 1999-2009), 
mean marine survival of Chinook salmon in the Chilkat 
Table 2 
Loadings of the first and second rotated components from 
principal component analysis of environmental and bio- 
logical variables used to explain variation in overwinter 
survival and smolt production of Chinook salmon (Onco- 
rhynchus tshawytscha) in the Chilkat River in Southeast 
Alaska. Data used in this analysis were for Chinook salmon 
from the brood years 1999-2009. 
Component RC1 RC2 
Parr length 0.88 -0.31 
Fall discharge 0.91 0.17 
Spring discharge 0.18 0.87 
Fall temperature 0.48 0.47 
Spring temperature -0.35 0.88 
Proportion of variance explained 0.39 0.38 
Cumulative proportion 0.39 0.77 
River was 2.8% (SE 0.7) and ranged from 1.3% (SE 0.4) for 
BY 2002 and 1.3% (SE 0.7) for BY 2008 to 4.7% (SE 1.0) for 
BY 2000 (Fig. 3). Although not significant at the 95% level, 
there was a negative temporal trend in marine survival 
for fish in the Chilkat River (R?=0.34, P=0.06). 
The annual mean mid-eye fork length for smolt Chinook 
salmon in the Chilkat River was 73 mm (range: 69-79 mm), 
and there was no significant trend in smolt length over the 
study period. Mean summer SST over the study period was 
12.0°C, and summer SSTs ranged from 10.5°C to 13.0°C. 
Mean temperature in the Chilkat River during the spring 
migration period (April-May) was 4.3°C (range: 3.3-5.5°C). 
The index value for standardized mean spring discharge in 
SEAK was —0.02 (range: —1.5—2.3). There was no significant 
temporal trend over this period (temperature: P=0.16; dis- 
charge: P=0.57). On average across BYs, smolt outmigra- 
tion in the Chilkat River ended on ordinal date 1386 (15 or 
16 May), and the end date ranged from ordinal date 129 
(8 May) in 2004 (BY 2002) to ordinal date 143 (23 May) in 
2001 (BY 1999). Outmigration timing (end date of outmi- 
gration) was positively related to smolt length (r=0.73) and 
negatively related to spring discharge (r=—0.69) (Table 1). 
Three PCs were retained and used to explain variation in 
log-transformed marine survival of Chinook salmon in the 
Chilkat River. The first PC (PC1) loaded heaviest on migra- 
tion timing, smolt length, and discharge, the second PC 
(PC2) loaded primarily on SST, and the third PC (PC3) 
loaded heaviest on average river temperature (Table 4). 
Log-transformed marine survival of Chinook salmon in 
the Chilkat River was significantly negatively related to 
PC1, indicating that marine survival was higher when out- 
migration ended later and smolt length was greater and 
that marine survival was lower when spring discharge was 
higher (R7=0.5, P=0.01) (Table 5, Fig. 3). 
Stikine River 
For Chinook salmon in the Stikine River, mean marine 
survival was 1.7% (SE 0.3) and ranged from 0.6% (SE 0.1) 
Table 3 
Results from the principal component regression model used to examine the relationships 
between the first and second rotated components and overwinter survival and smolt pro- 
duction of Chinook salmon (Oncorhynchus tshawytscha) in the Chilkat River in Southeast 
Alaska. Model statistics include the standardized regression coefficient (8), standard error 
of the coefficient (SE), coefficient divided by its standard error (t), and the coefficient of 
multiple determination (R”). Data used in this model were for Chinook salmon from the 
brood years 1999-2009. 
Response 
variable 
Explanatory variable 
Freshwater overwinter survival RC1 
RC2 
Smolt production RC1 
RC2 
Model statistic 
