Berkman et al.: Effects on survival of Oncorhynchus tshawytscha in rivers of Southeast Alaska 207 
to represent conditions and run timing experienced by 
a given BY. All statistical analyses were conducted in R, 
vers. 3.6.3 (R Core Team, 2020). 
Stikine River 
Biological data Marine survival, smolt abundance, and 
smolt fork length (in millimeters) of Chinook salmon 
returning to the Stikine River (BYs 1998-2009) were 
estimated with data collected by the ADFG, Fisheries 
and Oceans Canada, and Tahltan First Nation during 
the mark-recapture survey of adult Chinook salmon (we 
used data for the period 2001—2016) and during the CWT 
survey conducted in spring (we used data for the period 
2000-2011) (Pahlke et al.*; Jaecks et al.*; Jaecks’®). Data 
from the mark-recapture program were used to estimate 
in-river abundance of large Chinook salmon in the Stikine 
River, and those estimates, in turn, were used to calcu- 
late marine survival. The mark-recapture experiment was 
completed in 2 phases. During the first phase, Chinook 
salmon were captured during May—July with drift gill nets 
on the lower Stikine River; during the second phase, fish 
were captured from June through August with the same 
gear on the spawning grounds upriver (Jaecks et al.1). 
Captured Chinook salmon were measured for mid-eye 
fork length, externally examined to determine sex, sam- 
pled for scales, marked, and then released. Fish in both 
phases with missing adipose fins were processed similar 
to such Chinook salmon caught in the Chilkat River. The 
estimated abundance parameter was tabulated by using 
Chapman’s modification of the Petersen estimator, as long 
as all assumptions were met (Seber, 1982). When assump- 
tions were violated, a variety of stratification and mod- 
eling techniques were used on data to correct for bias as 
much as possible (Jaecks et al.'"). 
Data from the CWT study of juvenile Chinook salmon 
were used to estimate the number of smolts emigrating 
from the Stikine River. This study was designed as a mod- 
ified Petersen 2-event mark-recapture effort, where the 
tagging of juveniles (>50 mm in fork length) in the spring 
was the first sampling event and the mark-recapture of 
adults (described previously) was the second sampling 
event. Juvenile Chinook salmon were sampled annually in 
mid-April in the lower reaches (U.S. side) of the Stikine 
River (Jaecks et al.*). Salmon were caught by using beach 
seines that were 18 by 2 m or 24 by 2 m (depending on 
the crew), all with 63.5-cm mesh, and by using Gee min- 
now traps baited with disinfected salmon roe. Smolt 
abundance was estimated, by using adult mark-recapture 
data, for the 5 years it takes for each brood of smolts to 
return to the river. The fraction of the population that had 
been originally marked was estimated each year. If those 
10 Jaecks, T. 2016. Personal commun. Div. Sport Fish Commer. 
Fish., Alaska Dep. Fish Game, P.O. Box 115526, Juneau, AK 
99811-5526. 
1 Jaecks, T., P. Richards, and P. Etherton. 2013. Spawning 
escapement of Chinook salmon in the Stikine River, 2013. 
Alaska Dep. Fish Game, Reg. Oper. Plan SF.1J.2013.05, 27 p. 
[Available from website.] 
estimates were similar for all years, data were pooled and 
Chapman’s modification of Petersen’s estimator was used. 
If the estimates for the fraction of the population that had 
been marked were not consistent, the estimates were aver- 
aged over the years (Jaecks et al.*). Marked adult fish from 
the Stikine River were primarily recovered in marine, 
troll, and drift gill-net fisheries (Pahlke et al.°). 
Physical data To assess the relationship between marine 
survival and regional environmental conditions during 
the early marine period for Chinook salmon from the 
Stikine River, SST and river discharge were included in 
the model. Sea-surface temperature, as described previ- 
ously, was similarly used to represent conditions of the 
inshore marine waters of SEAK (in degrees Celsius) upon 
ocean entry of Chinook salmon (June—July). To represent 
river conditions during outmigration, data for monthly 
mean river discharge in the Stikine River was averaged 
for April-May (USGS gauging station no. 15024800). Data 
on river water temperatures were not available for the 
Stikine River and, therefore, not included in the model. 
Data analysis For Chinook salmon in the Stikine River, 
multiple and simple linear regression analyses were 
used to test for significant relationships between log- 
transformed marine survival and spring discharge, sum- 
mer SST, and smolt length. Environmental variables were 
lagged (BY+2) to represent experienced conditions of a 
given BY. The statistical approach used to examine factors 
for fish in the Stikine River was more simplistic than that 
used in analyses conducted for fish in the Chilkat River 
because there were fewer explanatory variables (n=3) 
and no correlation between those variables according to 
a pairwise comparison (Table 1). Results of the prelimi- 
nary exploration of variables included in the analyses 
for salmon in the Stikine River indicate the presence of a 
possible outlier; therefore, robust linear regression anal- 
yses were used as implemented in the function ImRob of 
the robust package (vers. 0.5-0.0; Wang et al., 2020) in R. 
A backward-stepwise approach, which was based on the 
Akaike information criterion for small sample sizes and 
removed terms one at a time on the basis of the largest 
reduction in deviance, was used to select the top models. 
All statistical analyses were conducted in R. 
Results 
Chilkat River 
Overwinter survival and smolt production Over the time 
series (BY 1999-2009), mean freshwater overwinter sur- 
vival for Chinook salmon in the Chilkat River was 36% 
(standard error of the mean [SE] 10) and ranged from 21% 
(SE 5) for BY 2000 to 53% (SE 15) for BY 2005 (Fig. 2). 
Mean smolt abundance was 172,616 individuals and 
ranged from 105,300 individuals in 2000 to 282,700 indi- 
viduals in 2003 (Fig. 2). There was no significant linear 
trend over time in annual estimates of overwinter survival 
