268 
Fishery Bulletin 117(3) 
sockeye salmon, and steelhead, O. mykiss ) that exhibit die! 
vertical movement patterns in stratified and unstratified 
waters (Drenner et al., 2012). Limited available data indi¬ 
cate that salmon in estuaries are often associated with 
thermo-haloclines (Brawn, 1982; Westerberg, 1982; B0ving 
et al., 1985; Potter, 1988; Olson and Quinn, 1993; Strange, 
2013), although vertical distribution is not always related 
to oceanographic conditions (Quinn et al., 1989; Ruggerone 
et al., 1990; Walker et al., 2000; Ishida et al., 2001). In the 
Columbia River estuary, individual Chinook salmon made 
complex vertical movements relative to varying thermal 
and salinity gradients (Olson and Quinn, 1993). Vertical 
placement near dines presumably provides advantages 
for behavioral thermo-osmoregulation and adjustment 
of ambient salinity as well as detection of olfactory clues 
needed for homing at shear zones (Westerberg, 1982). Tag¬ 
ging studies conducted during the adult migratory stage 
have revealed rapid transitions from saline to fresh waters 
and tolerance for substantial, brief changes in salinity 
among salmonids (Potter, 1988; Olson and Quinn, 1993). 
Some Chinook salmon populations migrate upriver 
during the highest temperatures in, for example, the Sac¬ 
ramento (Alabaster, 1989), Columbia (Keefer et al., 2015; 
Keefer et al., 2018), and Fraser (Waples et ah, 1990) River 
systems, although other populations in these large rivers 
migrate later or earlier. In contrast, coastal populations of 
sockeye salmon in systems subject to high temperatures 
(mean: 19°C) tend to migrate prior to peak temperatures 
(Hodgson and Quinn, 2002). This pattern, combined with 
the more southerly distribution of Chinook salmon com¬ 
pared with that of sockeye salmon, indicates substantially 
greater thermal tolerance in Chinook salmon, as part of 
the suite of adaptations of salmonids to stressful or lethal 
effects of warm temperatures along migration routes (Cro- 
zier et al., 2008). However, population-specific variation in 
thermal tolerance has been documented in sockeye salmon 
(Eliason et al., 2011) and likely plays an important role in 
Chinook salmon as well. 
Individual Chinook salmon use a variety of behavioral 
adaptations to high temperatures in rivers, such as delay¬ 
ing migration until temperatures decline, migrating at 
greater depths, and utilizing cool water refuges (Berman 
and Quinn, 1991; Strange, 2013; Keefer and Caudill, 2016). 
The Lake Washington Chinook salmon migrated near 
the annual peak temperatures, but individuals delayed 
in the estuary, occupied refuges in the estuary and lake, 
and moved vertically during migrations in those refuges. 
Because most Chinook salmon entering fresh water in late 
summer are near sexual maturity, a long delay in migra¬ 
tion at high temperatures may result in lost reproductive 
opportunity. In contrast, the timing of steelhead and some 
early arriving sockeye salmon populations is more flexible 
before spawning, where they may delay, seeking cool water 
refuge sites when temperatures are stressful (Keefer et al., 
2009; Mathes et al., 2010). Individual sockeye salmon 
may be less adaptable in their response to temperatures, 
instead relying on evolved traits such as earlier migration 
timing and use of vertical stratification in lakes (Hodgson 
and Quinn, 2002; Hinch et al., 2012). The Lake Washington 
sockeye salmon, studied by Newell and Quinn (2005) using 
methods similar to our study, displayed both responses, 
entering prior to peak temperatures, migrating rapidly 
through the ship canal, and residing for extended periods 
below the thermoeline in the lake. Entry into fresh water 
prior to peak temperatures, as observed in the Lake Wash¬ 
ington sockeye salmon, is an example of so-called prema¬ 
ture migration, a pattern widely distributed in salmonids 
(Quinn et al., 2016). In some species and populations, this 
pattern is related to thermal conditions, although other 
factors are also important. The use of cool water below the 
thermoeline in the lake by sockeye salmon starkly con¬ 
trasts the vertical distribution, hence thermal experience, 
of Chinook salmon in the Lake Washington basin. 
Exposure to warm water temperatures and the accu¬ 
mulation of DD have been correlated with prespawning 
mortality (Crossin et a!., 2008; Keefer et al., 2009; Eliason 
et al., 2011), which can reach 90% for Chinook salmon in 
the upper reaches of the Willamette River (Keefer et al., 
2015). Temperatures outside the acceptable range of 
14-20°C for Chinook salmon can deplete energy that could 
have been used for migration, reproduction, and immune 
responses (Richter and Kolmes, 2005; McCullough et al., 
2001; McCullough et al., 2009). The Lake Washington Chi¬ 
nook salmon accumulated 350-1050 BB, with most hav¬ 
ing between 500 and 750 DD, compared with 700-900 DD 
(maximum: 1500 DD) for those in the Willamette River 
(Keefer et al., 2015). Lake Washington has gotten warmer 
in the summer and fall over the past 5 decades (Fig. 2), 
and in the future the watershed is expected to see greater 
increases in temperature than any watershed in Puget 
Sound (Mantua et al., 2010). How Chinook salmon will 
respond to this marked change in water temperatures is 
of concern, in part because this population is part of an 
evolutionarily significant unit listed as threatened under 
the U. S. Endangered Species Act (Federal Register, 1999). 
Responses to warm water may include genetic and pheno¬ 
typic changes in migration timing to avoid peak tempera 
tures or more extensive behavioral thermoregulation. Of 
ultimate concern is how warming will affect the persistence 
of the salmon populations. The temperatures at the upper 
estuary in most summer months already approach or 
exceed levels associated with disease and energy depletion 
in migrating salmon; therefore, en route or prespawning 
mortality may result from further increases (McCullough 
et al., 2001; Crossin et al., 2008). Cool, hypolimnetic water 
will remain in Lake Washington if salmon can migrate 
successfully from the locks through the ship canal. In the 
future, temperatures could delay their migration or fur¬ 
ther stress them prior to reaching the lake. Therefore, if 
water temperatures in Lake Washington continue to rise, 
the success of migration and spawning by Chinook and 
sockeye salmon will likely be compromised. 
Acknowledgments 
We thank D. Seiler (WDFW) and K. Fresh (National 
Marine Fisheries Service) for their support in collaborative 
