Fishery Bulletin 119(4) 
Table 1 
Mean estimates of and Pearson product-moment correlation coefficients (r) from pairwise comparisons of explan- 
atory variables used in analyses of freshwater survival and production (smolt abundance) and marine survival 
of Chinook salmon (Oncorhynchus tshawytscha) in the Chilkat River (with data for brood years 1999-2009) and 
in analyses of marine survival of Chinook salmon in the Stikine River (with data for brood years 1998-2009) in 
Southeast Alaska. Variables include mean parr mid-eye fork length, mean river discharge and air temperature 
in the fall (September—November), mean river discharge and air temperature in early spring (March—April), 
mean smolt length, migration timing, sea-surface temperature (SST), and mean river discharge and water tem- 
perature in the spring (April—May). A standardized index, based on monthly mean discharge averaged across 
rivers in Southeast Alaska for the primary migration period of smolts (April—May), was used for discharge in the 
Chilkat River. Data for river water temperature were not available for the Chilkat River. An asterisk (*) indicates 
a significant correlation between variables (P<0.05). SD=standard deviation. 
Estimate 
Variable Mean SD 
Chilkat River: freshwater survival and production 
Parr length (mm) 68.85 2.67 
Fall discharge (index) —0.06 0.76 
Spring discharge (index) -0.04 1.05 
Fall temperature (°C) 3.16 0.74 
Spring temperature (°C) -0.83 1.73 
Estimate 
Variable Mean SD 
Chilkat River: marine survival 
Migration timing (ordinal date) 136.09 4.16 
SST (°C) 12.03 0.79 
Smolt length (mm) 73.42 3.37 
Spring discharge (index) -0.02 1.03 
Spring temperature (°C) 4.34 0.76 
Estimate 
Variable Mean SD 
Stikine River: marine survival 
SST (°C) 73.58 3.03 
Smolt length (mm) 12.07 0.77 
Spring discharge (m*/s) 40,499 11,381 
variables (mean: 0; standard deviation: 1) were created by 
using principal component analysis with varimax rotation 
(Jackson, 2005). 
The number of rotated components (RCs) that contained 
interpretable information were chosen through a visual 
assessment of a scree plot. Principal component regression 
models were fit to test for significant relationships between 
response variables (log-transformed overwinter survival 
and smolt production) and RCs. In addition, to uncover 
potential density dependence during the pre-smolt fresh- 
water period, a simple linear regression was fit to describe 
the relationship between parr abundance and parr length. 
Principal component analysis in combination with prin- 
cipal component regression was used to estimate effects of 
Migration 
Fall Spring Fall 
discharge discharge temperature 
0.11 
Spring 
timing discharge 
Spring 
discharge 
biological and environmental factors on marine survival 
of Chinook salmon in the Chilkat River because of the 
low sample size, high number of explanatory variables, 
and multicollinearity. Results of pairwise comparisons 
of explanatory variables made by using the r indicate 
significant correlation between the timing of the end of 
the smolt migration and smolt length (positive) and late 
spring discharge (negative) (Table 1). For these analyses, 
log-transformed marine survival was used as the response 
variable, and principal component (PC) variables that 
combine spring river temperature and discharge, summer 
SST, ordinal date at the end of smolt outmigration, and 
smolt length were included as explanatory variables. Envi- 
ronmental and run timing variables were lagged (BY+2) 
