Oversummer growth and survival of juvenile coho salmon (Oncorhynchus kisutch) across a natural gradient of stream water temperature and prey availability: an in situ enclosure experiment

Abstract

Conservation efforts for Pacific salmon (Oncorhynchus spp.) increasingly prioritize maintenance of cool water temperatures that protect all freshwater life stages. However, development of appropriate temperature standards requires a robust understanding of the interactions among water temperature, ecosystem productivity, and fish performance. We used a series of in situ enclosures to examine how natural spatiotemporal gradients in thermal conditions and prey availability affected the summer growth and survival of age-0+ coho salmon (Oncorhynchus kisutch). Coho salmon absolute growth rates peaked at a mean daily average water temperature (mean T) of 16.6 °C and an associated maximum weekly maximum temperature (MWMT) of 21.1 °C. Juvenile growth under these thermal conditions was sixfold greater than the growth rates observed for conspecifics rearing in the coolest study reach (mean T = 13.0 °C; MWMT = 16.0 °C). Even at the highest rearing temperature (mean T = 18.1 °C; MWMT = 24.0 °C), growth rates remained positive and above the study-wide average, although overall survival was reduced. Among the predictor variables examined, invertebrate prey abundance was the predominant factor influencing age-0+ coho salmon growth. These results suggest that abundant prey resources may mitigate the negative effects of elevated water temperature on fish growth in riverine environments. Given the likelihood of increasing stream temperatures with climate change, productive ecosystems may provide critical refuges for juvenile salmonids.

Full text

ARTICLE
Oversummer growth and survival of juvenile coho salmon
(Oncorhynchus kisutch) across a natural gradient of stream water
temperature and prey availability: an in situ enclosure
experiment
Robert A. Lusardi, Bruce G. Hammock, Carson A. Jeffres, Randy A. Dahlgren, and Joseph D. Kiernan
Abstract: Conservation efforts for Pacific salmon (Oncorhynchus spp.) increasingly prioritize maintenance of cool water temper-
atures that protect all freshwater life stages. However, development of appropriate temperature standards requires a robust
understanding of the interactions among water temperature, ecosystem productivity, and fish performance. We used a series of
in situ enclosures to examine how natural spatiotemporal gradients in thermal conditions and prey availability affected the
summer growth and survival of age-0+ coho salmon (Oncorhynchus kisutch). Coho salmon absolute growth rates peaked at a mean
daily average water temperature (mean T) of 16.6 °C and an associated maximum weekly maximum temperature (MWMT) of
21.1 °C. Juvenile growth under these thermal conditions was sixfold greater than the growth rates observed for conspecifics
rearing in the coolest study reach (mean T= 13.0 °C; MWMT = 16.0 °C). Even at the highest rearing temperature (mean T= 18.1 °C;
MWMT = 24.0 °C), growth rates remained positive and above the study-wide average, although overall survival was reduced.
Among the predictor variables examined, invertebrate prey abundance was the predominant factor influencing age-0+ coho
salmon growth. These results suggest that abundant prey resources may mitigate the negative effects of elevated water temper-
ature on fish growth in riverine environments. Given the likelihood of increasing stream temperatures with climate change,
productive ecosystems may provide critical refuges for juvenile salmonids.
Résumé : Les efforts de conservation des saumons du Pacifique (Oncorhynchus spp.) accordent une priorité croissante au maintien
de températures de l’eau fraîches, qui protègent les poissons à toutes les étapes de la vie en eau douce. L'élaboration de normes
de température convenables nécessite toutefois une bonne compréhension des interactions entre la température de l’eau, la
productivité des écosystèmes et la performance des poissons. Nous avons utilisé une série d’enclos en place pour examiner
l’influence de gradients spatiotemporels naturels des conditions thermiques et de la disponibilité de proies sur la croissance et
la survie estivales de saumons cohos (Oncorhynchus kisutch) d’âge-0+. Les taux de croissance absolus des saumons cohos atteig-
naient un maximum à une valeur moyenne de la température de l’eau moyenne quotidienne (Tmoyenne) de 16,6 °C et une valeur
maximum de la température maximum hebdomadaire (MTMH) associée de 21,1 °C. La croissance des juvéniles dans ces condi-
tions thermiques était six fois plus grande que les taux de croissance observés pour les congénères croissant dans le tronçon
étudié le plus frais (Tmoyenne = 13,0 °C; MTMH = 16,0 °C). Même à la température de croissance la plus élevée (Tmoyenne =
18,1 °C; MTMH = 24,0 °C), les taux de croissance demeuraient positifs et supérieurs à la moyenne pour l’ensemble de l’étude,
malgré une survie globale plus faible. Parmi les variables prédictives examinées, l’abondance de proies invertébrées est le facteur
qui influençait de manière prédominante la croissance des saumons cohos d’âge-0+. Ces résultats donnent à penser que des
ressources de proies abondantes pourraient atténuer les effets négatifs de températures de l’eau élevées sur la croissance des
poissons en milieu fluvial. Étant donné la forte probabilité d’augmentation des températures des cours d’eau en raison des change-
ments climatiques, les écosystèmes productifs pourraient offrir des refuges critiques aux salmonidés juvéniles. [Traduit par la
Rédaction]
Introduction
Efforts to conserve at-risk populations of anadromous Pacific
salmon (Oncorhynchus spp.) often center on the restoration of phys-
ical habitat attributes (e.g., stream sinuosity, pool frequency and
depth, large wood abundance) associated with enhanced juvenile
production in fresh water. However, there is growing recognition
that restoration and protection of suitable hydrologic and ther-
mal regimes will be critical to the long-term viability of many
salmonid populations (Mantua et al. 2010;He and Marcinkevage
2017;Obedzinski et al. 2018). This is especially true in California
(USA), where numerous salmon stocks are listed as threatened
or endangered under the US Endangered Species Act (ESA), and
Received 6 December 2018. Accepted 2 July 2019.
R.A. Lusardi. Center for Watershed Sciences/California Trout, University of California, One Shields Avenue, Davis, CA 95616, USA; Department of
Wildlife, Fish, and Conservation Biology, University of California, One Shields Ave., Davis, CA 95616, USA.
B.G. Hammock. Aquatic Health Program, School of Veterinary Medicine, Department of Anatomy, Physiology, and Cell Biology, University of
California, One Shields Ave., Davis, CA 95616, USA.
C.A. Jeffres and R.A. Dahlgren. Center for Watershed Sciences, University of California, One Shields Ave., Davis, CA 95616, USA.
J.D. Kiernan. Fisheries Ecology Division, Southwest Fisheries Science Center, National Marine Fisheries Service, National Oceanic and Atmospheric
Administration, 110 McAllister Way, Santa Cruz, CA 95060, USA; Institute of Marine Sciences, University of California, 1156 High Street, Santa Cruz,
CA 95064, USA.
Corresponding author: Robert A. Lusardi (email: ralusardi@ucdavis.edu).
Copyright remains with the author(s) or their institution(s). Permission for reuse (free in most cases) can be obtained from RightsLink.
413
Can. J. Fish. Aquat. Sci. 77: 413–424 (2020) dx.doi.org/10.1139/cjfas-2018-0484 Published at www.nrcresearchpress.com/cjfas on 30 July 2019.
Can. J. Fish. Aquat. Sci. Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Davis on 02/03/20
For personal use only.

extant populations are universally challenged by both anthropo-
genic impacts and extreme climatic conditions (Moyle et al. 2013,
2017).
Water temperature is a key environmental variable that influ-
ences juvenile salmonid production through direct and indirect
pathways. While direct adverse effects of elevated stream temper-
atures can be acute (i.e., lethal) due to extreme thermal stress,
more frequently effects are chronic (i.e., sublethal) and associated
with short-duration exposure to small or moderate increases in
temperature. Chronic effects can include reduced growth and de-
velopment (Martin et al. 1986;Reeves et al. 1989), altered behavior
(Goniea et al. 2006), increased susceptibility to disease (Becker and
Fujihara 1978;Miller et al. 2014), or changes in life-history phe-
nology (Kovach et al. 2013;Spence and Dick 2014). Furthermore,
water temperature can indirectly influence juvenile salmon growth
and production via changes in riverine food web structure (e.g.,
rates of primary and secondary production) or by increasing the
frequency of negative density-dependent interactions such as
competition and predation (Wenger et al. 2011;Lehman et al.
2017).
Coho salmon (Oncorhynchus kisutch) are recognized as the most
temperature-intolerant of the Pacific salmonids and experience
thermal stress at water temperatures as low as 16 °C (Brett 1952).
Laboratory studies have demonstrated that juvenile coho salmon
growth and performance are optimized at temperatures between
11.4 and 14.5 °C (e.g., Coutant 1977;Reiser and Bjorn 1979;Bell
1986), and field observations generally confirm a relationship
between cool thermal regimes and the presence of juvenile coho
salmon (Welsh et al. 2001). However, fish growth rates in nature
are determined by the interactions among water temperature,
prey availability, and the energetic costs of living in a lotic environ-
ment. Thus, for juvenile salmonids, enhanced trophic resources may
offset the metabolic costs associated with elevated water temper-
ature to some extent (Brewitt et al. 2017), when temperatures
remain below the critical thermal maximum (estimated range 27
to 29 °C for wild coho salmon; Konecki et al. 1995). There is in-
creasing evidence that juvenile coho salmon can occupy and per-
sist in ostensibly thermally stressful habitats when food resources
are abundant. For example, Osterback et al. (2018) reported positive
growth by juvenile coho salmon in a central California coastal
freshwater lagoon despite mean daily water temperatures > 20 °C
and attributed these results, in part, to high standing stocks of
invertebrate prey. It is also likely that local adaptation and (or)
acclimation contribute to the ability of coho salmon populations
at the southern end of their North American range (i.e., Califor-
nia) to cope with elevated and highly variable thermal regimes.
While the effects of stream temperature on juvenile salmonid
growth and performance have been extensively studied (e.g., Brett
1952;McCullough 1999;Myrick and Cech 2004), it remains uncer-
tain how ecosystem productivity influences the interaction be-
tween water temperature and salmon performance (Myrvold and
Kennedy 2015). Several authors have called for a broader under-
standing of how prey availability influences salmonid persistence
in the wild (e.g., Weber et al. 2014;Lusardi et al. 2016,2018), par-
ticularly where high water temperatures may limit growth and
production. To address this knowledge gap, we used a series of
in situ mesocosms to quantify differences in the oversummer
growth and survival of juvenile (age-0+) coho salmon across a
natural gradient of stream temperature and prey abundance. Our
aim was to examine whether enhanced food availability mitigated
the adverse effects of elevated water temperature on the summer
growth and survival of age-0+ coho salmon. This work has impli-
cations for contemporary salmonid conservation and recovery
efforts and for predicting the response of salmonids to future
climate change.
Methods
Study system
This study was conducted in a 10 km segment of Big Spring
Creek and the Shasta River within the Klamath River watershed
in northern California, USA (Fig. 1). The Shasta River (watershed
area = 2070 km
2
) originates in the Scott Mountains and flows
northward ⬃93 km before joining the Klamath River. The climate
is semi-arid and characterized by cool, wet winters and warm, dry
summers. Annual precipitation averages 48.3 cm (water years
1981–2015, GHCND rainfall gage USC0004 49866; http://www.ncdc.
noaa.gov) and occurs as both rain and snow, primarily between
October and April. Mean monthly maximum air temperature ranges
from ⬃7 °C in January to 33 °C in July. Historically, favorable thermal
conditions for Chinook salmon (Oncorhynchus tshawytscha), coho
salmon, and steelhead (anadromous Oncorhynchus mykiss)inthe
Shasta River basin were maintained by large-volume volcanic
spring sources that provided year-round inputs of cool, nutrient-
rich water. Prior to water development in the Shasta River basin,
the largest natural spring complex (Big Springs) provided a steady
flow of 2.9 m
3
·s
−1
of 11–12 °C water to the mainstem Shasta River
(Mack 1960) and accounted for more than half of the 5.7 m
3
·s
−1
historical (unimpaired) summer base flow (NRC 2004;Null et al.
2009).
Land-use change and intensive water management have ad-
versely affected salmonids and other native cold-water fishes in
the Shasta River basin (NRC 2004). The construction of Dwinnell
Dam on the mainstem Shasta River (river kilometre (rkm) 65.4) in
1928 eliminated access to more than 22% of the historical salmon
spawning and rearing habitat (Wales 1951). Downstream of Dwin-
nell Dam, the Shasta River flows predominantly through a low-
gradient alluvial valley with little riparian shading. The dominant
land use is agriculture, and widespread water abstraction (and
subsequent tailwater returns of warm water) during the April to
October irrigation season result in highly altered summer stream-
flow and temperature regimes that are generally unfavorable for
salmonids, and juvenile coho salmon in particular (Jeffres and
Moyle 2012;Moyle et al. 2017). The Southern Oregon/Northern
California Coast (SONCC) coho salmon evolutionary significant unit,
which includes the Shasta River population, is listed as threatened
under the US ESA. The federal SONCC coho salmon recovery plan
(NMFS 2014) explicitly identifies summer stream temperatures as a
critical stressor affecting the Shasta River population.
Experimental design
We used in situ mesocosms (hereinafter enclosures) to examine
how natural gradients of stream water temperature and prey avail-
ability in the Shasta River basin affected the oversummer perfor-
mance of age-0+ coho salmon. Our experiment, which began on
9–10 July 2013 and was terminated on 8 –9 September 2013 (63 days), was
conducted during the summer dry period when stream discharge
in the Shasta River basin is low, water temperature is elevated,
and juvenile salmonid mortality is presumed to be high (Chesney
et al. 2009;Nichols et al. 2013;NMFS 2014).
Our study was carried out in five 50 m study reaches in the
Shasta River basin (Fig. 1). The study reaches were chosen to rep-
resent a gradient in thermal conditions, and reach selection was
informed by previous research in the watershed (e.g., Jeffres et al.
2009;Nichols et al. 2013). Four study reaches, coded BS1, BS2, SR2,
and SR3, were located 0.5, 2.7, 5.9, and 10.0 km, respectively,
downstream of natural cold-water spring sources. A fifth study
reach (coded SR1) was located on the mainstem Shasta River
⬃200 m upstream of the Big Springs Creek confluence and was
not influenced by major spring sources (Fig. 1). In each of the five
study reaches, we constructed five replicate enclosures (254 cm
long × 133 cm wide; area = 3.4 m
2
), resulting in 25 enclosures total.
The enclosures were constructed by hammering T-posts into the
streambed in a rectangular pattern and encircling the perimeter
414 Can. J. Fish. Aquat. Sci. Vol. 77, 2020
Published by NRC Research Press
Can. J. Fish. Aquat. Sci. Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Davis on 02/03/20
For personal use only.

of the T-posts with heavy delta knotless netting (6.35 mm aperture;
Memphis Net and Twine Co., Inc., Memphis, Tennessee, USA). Tube
sandbags (12.6 cm diameter × 152.4 cm long; mass ⬃30 kg) were
used to form a seal between the bottom of the perimeter netting
and the natural streambed. The net walls of each enclosure ex-
tended vertically from the streambed to a height > 60 cm above
the water’s surface. Enclosure walls were permeable to most drift-
ing invertebrates but not fishes. While we recognize that enclo-
sure walls may have inhibited the drift of some invertebrate taxa,
such effects were likely minimal due to the relatively large aper-
ture size of the enclosure netting, and any realized effect was
expected to be similar among enclosures and across study
reaches. The natural streambed served as the bottom of each en-
closure and the tops of each enclosure were open to permit insect
emergence and inputs of allochthonous material. However, a
small amount of fluorescent cordage was strung across the tops to
deter roosting behavior and (or) entry by aerial predators. To min-
imize the potential influence of upstream enclosures on water
velocity and the delivery of organic material (e.g., detritus and
drifting invertebrates) to downstream enclosures, we constructed
each enclosure with its long axes parallel to streamflow, and the
in-stream position of each successive enclosure was offset (i.e.,
river left, midchannel, and river right) longitudinally within the
study reach.
Biotic and abiotic conditions
We quantified a select set of physical and biological parameters
during the 63-day experiment to characterize differences in rear-
ing conditions within and among study reaches. Stream water
temperature was continuously recorded at 15 min intervals using
HOBO Pro v2 (Onset Computer Corporation, Bourne, Massachu-
setts, USA) data loggers affixed to the upstream-most enclosure in
each study reach. Water velocity (±0.01 m·s
−1
) and depth (±0.01 cm)
were measured within each enclosure near the midpoint of the
experiment (day 31; 10–11 August 2013) using a Marsh-McBirney
Flo-Mate (Hach Co., Loveland, Colorado, USA) velocity meter
attached to a top-setting wading rod. We visually estimated the
percentage cover (±5%) provided by aquatic macrophytes (predom-
inantly shortspike watermilfoil (Myriophyllum sibiricum) and white-
water crowfoot (Ranunculus aquatilis)) within each enclosure at the
midpoint of the experiment.
Benthic invertebrate sampling
As a measure of prey availability in each study reach, benthic
invertebrates were collected at the start (day 0), midpoint (day 31),
Fig. 1. Location map showing study reach locations on Big Springs Creek and the Shasta River (Siskiyou County, California, USA).
Lusardi et al. 415
Published by NRC Research Press
Can. J. Fish. Aquat. Sci. Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Davis on 02/03/20
For personal use only.

and end (day 63) of the experiment. Although juvenile coho salmon
are known to preferentially forage in the water column rather than
from the benthos (e.g., Nielsen 1992), invertebrate drift is roughly
proportional to benthic invertebrate density and thus a suitable
proxy for prey availability (Hildebrand 1974;Hammock and Wetzel
2013;Kennedy et al. 2014). Within the Shasta River, Lusardi et al.
(2018) reported a positive relationship between invertebrate den-
sity and invertebrate drift. Benthic invertebrate samples were
collected proximate (but external) to each enclosure to avoid dis-
rupting conditions and biota inside the experimental units. On
each sample date, we randomly selected four transects within
each study reach and used a modified Hess sampler (335 ␮m mesh;
sample area = 0.12 m
2
) to collect three benthic samples along each
transect (25%, 50%, and 75% wetted width). The three samples were
combined in the field to produce one composite sample per tran-
sect and resulted in four composited invertebrate samples per
study reach on each sample date. All samples were preserved in
90% ethanol.
In the laboratory, benthic invertebrate samples were processed
using a subsampling procedure (Folsom plankton splitter or Caton
(1991) gridded tray) to randomly extract a minimum of 500 organ-
isms. In cases where a sample contained fewer than 500 organisms,
all invertebrates were removed. Invertebrates were subsequently
dried at 60 °C for ≥48 h and weighed on an analytical balance
(±0.01 g). Dried samples were ashed in a muffle furnace at 475 °C
for ≥90 min, cooled to room temperature, and reweighed to de-
termine ash-free dry mass (AFDM, g). We did not determine the
taxonomic composition of invertebrate samples.
Coho salmon growth and performance
On 8 July 2013, 150 subyearling (age-0+) coho salmon were ob-
tained from Iron Gate Fish Hatchery (California Department of
Fish and Wildlife, Hornbrook, California, USA), located on the
Klamath River, ⬃21 rkm upstream of the mouth of the Shasta
River. Hence, the juvenile coho salmon used in our study were a
local stock and presumed well-adapted to local environmental
conditions, including the thermal regime. During the 60 days
preceding our study, our experimental fish were exposed to water
temperatures in the hatchery that ranged from 8.3 to 14.4 °C
(mean = 12.2 °C). Fish were transported to the study area in insu-
lated and oxygenated 114 L plastic containers. After a 24 h holding
period, individual fish were anesthetized using a buffered solu-
tion of tricaine methanesulfonate (MS-222), weighed (wet mass
±0.1 g), measured for fork length (FL; ±1.0 mm), and distinctively
marked using a 12.5 mm × 2.1 mm, 134.2 kHz full-duplex passive
integrated transponder (PIT) tag (Biomark Inc., Boise, Idaho, USA).
Following PIT tagging, fish were separated into two size groups
based on the distribution of fork lengths: 70–78 mm (small; mean
FL = 75 mm) and 79–90 mm (large; mean FL = 82 mm). Three
individuals from each size group were then randomly selected,
positively identified via PIT tag number, and assigned to each
enclosure (n= 6 fish per enclosure; density = 1.8 age-0+ coho
salmon·m
−2
). While this density is high relative to what would be ex-
pected basin-wide, juvenile coho salmon densities > 2.0 individuals·m
−2
have been documented in the watershed (e.g., Big Springs Creek;
Jeffres et al. 2009). Coho salmon were acclimated to water temper-
atures at each reach prior to being transferred to experimental
enclosures on 9–10 July 2013. There were no significant differences
in initial FL or mass among study reaches or among enclosures
within reaches at the start of the experiment (nested analysis of
variance (ANOVA), p> 0.52 for all variables).
During the experiment, each enclosure was cleared of external
debris, inspected for physical damage, and checked for the pres-
ence of dead fish at least every 2 days. Two enclosures (one each in
study reaches SR2 and SR3) were damaged and all fish escaped;
consequently, these enclosures were excluded from all subse-
quent analyses. Individuals found dead during the experiment
were not replaced, as coho salmon survivorship was a response
variable of interest. We acknowledge that live fish remaining in
enclosures where mortalities occurred may have experienced en-
hanced growth rates due to reduced density, and we accounted for
this possibility by using enclosure as a random effect in the anal-
ysis (see Data analysis section). On 8–9 September 2013 (after
63 days), surviving fish in each enclosure were collected using a
backpack electrofisher, whereupon they were anesthetized, iden-
tified via PIT tag number, and weighed to determine final mass.
Data analysis
We extracted the average, maximum, and minimum stream
temperature for each day of the experiment and used these data
to calculate the mean daily average water temperature (mean T)
and mean diel variation (mean ⌬T) at each study reach. Addition-
ally, we converted each time series to 7-day moving averages and
calculated maximum weekly average temperature (MWAT), max-
imum weekly maximum temperature (MWMT), and maximum
weekly minimum temperature (MWMinT). These summary met-
rics were subsequently used to model the effect of water temper-
ature on coho salmon growth (see Model development). Benthic
invertebrate density (no. individuals·m
−2
) and biomass (g AFDM·m
−2
)
were extrapolated based on the area of streambed sampled and
the fraction of each sample processed in the laboratory. Inverte-
brate density and biomass samples collected from each study
reach at the start, midpoint, and end of the experiment (n= 12)
were averaged to produce an overall estimate of food availability
during the study period. Averaging the invertebrate density data
allowed us to regress invertebrate density against coho salmon
growth, a response variable for which we had a single measure-
ment per fish across the experiment. The response of age-0+ coho
salmon during the experimental period was assessed in terms of
absolute growth in mass (G, g·day
−1
) and proportion mortality,
with enclosures within study reaches serving as replicate experi-
mental units. While our analysis averaged across potentially im-
portant temporal variation in juvenile coho salmon growth and
its predictors, a major advantage is that the fish remained undis-
turbed for the entire experiment.
Statistical differences in invertebrate density, invertebrate bio-
mass, and coho salmon absolute growth rates between reaches
were assessed using ANOVA with a significance level (
␣
) of 0.05.
Each variable was transformed (natural-log) to meet assumptions
of normality and to correct for heteroscedasticity. Significant
ANOVAs were followed by Tukey’s honestly significant differ-
ences (HSD) tests. ANOVAs were run in JMP Pro version 12.0 (SAS
Institute Inc., Cary, North Carolina, USA).
Model development
Environmental predictors of coho salmon growth and mortality
We used an information-theoretic approach (Burnham and
Anderson 2002;McElreath 2016) to examine how select environ-
mental parameters contributed to the growth and survival of
age-0+ coho salmon across all study reaches. Candidate models
were compared using Akaike’s information criteria adjusted for
small sample size (AIC
c
;Burnham and Anderson 2002). Statistical
significance was assessed based on the relative ranking of the
models and whether the 95% confidence intervals of environmen-
tal parameter estimates overlapped zero. Models exhibiting a
delta (⌬) AIC
c
value > 2.0 were considered dissimilar (Bolker 2008).
All model comparisons were conducted in R (R Core Team 2015).
For each set of models considered, enclosure was included as a
random effect to account for individual enclosure effects within
study reaches. For all models, each fixed effect variable was scaled
using the “scale” command in R to ensure that all parameters
could be estimated and were of similar magnitude (“scale” sub-
tracts the mean and divides by standard deviation; Becker et al.
1988). To demonstrate effect sizes for the environmental variables
416 Can. J. Fish. Aquat. Sci. Vol. 77, 2020
Published by NRC Research Press
Can. J. Fish. Aquat. Sci. Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Davis on 02/03/20
For personal use only.

considered, we made predictions using models that received sub-
stantial AIC
c
support (⌬AIC
c
≤ 5.0) across the range in environ-
mental predictors.
We employed a three-step model development process to
determine drivers of coho salmon growth in our study. First, we
examined a set of candidate models to specifically assess whether
invertebrate density or invertebrate biomass was the best proxy
for coho salmon food availability (see online Supplementary
material
1
). Since the model with invertebrate density performed
conclusively better than other candidate models (Table S1
1
), all
subsequent analyses used invertebrate density as the measure of
prey availability. Second, again using coho salmon absolute growth
rate as the response variable, we compared eight models to assess
the influence of the summary temperature metrics derived in our
study (Table S2
1
). AIC
c
did not strongly differentiate between dif-
ferent potential temperature metrics, although minimum tempera-
ture models (m3 and m7) and mean ⌬T(m8) received substantially
less weight (Table S2
1
). Therefore, we used mean Tas our primary
water temperature metric for all subsequent analyses because it ad-
equately captures the range of thermal conditions experienced dur-
ing the entire study period. However, we also report MWMT values
because of the broad application of this temperature metric by reg-
ulatory agencies throughout the western USA (EPA 2003) and its
prevalence in the ecological literature (e.g., Welsh et al. 2001;Moore
et al. 2013;Mauger et al. 2017).
In the final step, we compared 11 models of coho salmon growth
rate: (m1) an intercept model, (m2) an invertebrate density model,
(m3) a temperature model (mean T), (m4) a water velocity model,
(m5) a water depth model, (m6) a percentage aquatic macrophytes
model, (m7) an invertebrate density plus temperature (MWMT)
model, (m8) an invertebrate density plus velocity model, (m9) an
invertebrate density plus depth model, (m10) an invertebrate den-
sity plus percentage aquatic macrophytes model, and (m11) an inver-
tebrate density by temperature interaction model (see Table 2). The
invertebrate density by temperature interaction model was in-
cluded because both predictor variables have been shown to in-
teract to affect salmonid growth in both laboratory experiments
(Wurtsbaugh and Davis 1977) and bioenergetics model simula-
tions (Railsback and Rose 1999). Invertebrate density was included
in most models to prevent the strong association between inver-
tebrate density and coho salmon growth from masking the effects of
other variables on growth (McElreath 2016). Furthermore, because
the influence of temperature on fish growth is nonlinear over wide
temperature ranges (e.g., Fonds et al. 1992;Cotton et al. 2003), we
assessed the shape of the relationship between growth and tem-
perature in our analysis by examining box and whisker plots of
the residuals from the invertebrate density model (i.e., the top-
ranked growth model) against temperature (Fig. S1
1
). This analysis
suggested a possible nonlinear response of coho salmon growth to
temperature. As such, a quadratic temperature model was also fit,
but it did not perform well and was therefore not included in the
model comparison. We used the R packages lme4 (Bates et al.
2015) and nlme (Pinheiro et al. 2018) to fit the growth rate models
with a Gaussian distribution of error.
Coho salmon mortality data were also analyzed using multimodel
inference; however, we used generalized linear models (GLMs) rather
than mixed effects Gaussian models because the response variable
was a proportion (McElreath 2016). Models included an intercept-
only model (null) and a model with MWMT as a predictor. We used
a binomial distribution of error for both the null and temperature
models because models with this distribution substantially out-
performed beta-binomial models in terms of AIC
c
in a preliminary
analysis. Maximum likelihood estimation was used to fit beta-
binomial mortality models (bbmle package, mle2; Bolker 2016),
while binomial models were fit using the glm command in R.
Results
Biotic and abiotic conditions
Physical habitat characteristics were variable both within and
among the five study reaches. Aquatic macrophyte coverage ranged
from 0% (reach BS1) to 100% (reach SR2) of the stream channel
(Table 1). Water depth averaged 42.2 cm (±1 standard error (SE) =
2.6 cm) across all enclosures (range = 21.3–56.5 cm; n= 25), and
reaches BS1 and SR2 were significantly different from all other
reaches (ANOVA, F
[4,20]
= 4.47, p< 0.01; Tukey’s HSD, p< 0.05;
Table 1). Water velocity also differed among study reaches, as SR3
1
Supplementary data are available with the article through the journal Web site at http://nrcresearchpress.com/doi/suppl/10.1139/cjfas-2018-0484.
Table 1. Abiotic and biotic variables associated with each study reach.
Variable
Study reach
BS1 BS2 SR1 SR2 SR3
Distance from springs (km) 0.5 2.7 NA
a
5.9 10
Mean depth (cm) 50.1±1.9 46.1±2.4 38.7±1.0 35.8±4.7 40.3±2.4
Mean velocity (m·s
–1
) 0.07±0.02 0.13±0.01 0.09±0.01 0.16±0.04 0.32±0.05
Temperature (°C)
Mean T
b
13.0±0.03 14.8±0.01 18.1±0.1 16.6±0.1 17.1±0.1
Range
c
11.1–16.1 11.1–19.2 13.1–24.7 12.3–21.6 13.3–20.9
Mean ⌬T
d
3.9±0.49 5.2±0.66 6.7±0.86 5.1±0.65 3.2±0.41
MWAT
e
13.3 15.7 19.9 18.0 18.7
MWMT
f
16.0 18.9 24.0 21.1 20.6
Mean macrophyte cover (%) 0 65±19.0 85.4±1.2 100 7.8±2.1
Mean invertebrate density (individuals·m
–2
) 32 019±2 408 39 451±6 129 51 474±7 469 64 106±9 597 34 219±4 997
Mean invertebrate biomass (g·m
–2
) 2.8±0.5 3.4±0.6 9.9±3.4 7.8±1.4 5.3±1.1
Note: Mean values are presented ±1 standard error. Water depth and velocity were measured and macrophyte cover was visually
estimated (±5%) inside each enclosure (n= 5 at each study reach) at the midpoint of the experimental period. Water temperature was
continuously recorded (15 min intervals) at a single location in each study reach. Invertebrate density and biomass values for each study
reach represent the mean of 12 samples (four transects × three sample dates; see Methods).
a
Study reach SR1 was located upstream of the Big Springs Creek – Shasta River confluence and was not influenced by source springs.
b
Mean daily average water temperature (n= 63 days).
c
Range is the instantaneous minimum and maximum water temperature values recorded during the study.
d
Mean ⌬Trepresents the average diel variation (n= 63 days).
e
Maximum weekly average temperature.
f
Maximum weekly maximum temperature.
Lusardi et al. 417
Published by NRC Research Press
Can. J. Fish. Aquat. Sci. Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Davis on 02/03/20
For personal use only.

exhibited significantly higher velocity at the midpoint of the ex-
periment relative to the other four reaches (F
[4,20]
= 8.9, p< 0.01;
Tukey’s HSD, p< 0.05; Table 1).
Stream water temperature was variable across all study reaches
(Table 1), but each time series exhibited a similar temporal pattern
of decline during the study period (Fig. 2). Mean Tvalues ranged
from 13.0 to 18.1 °C, and MWMT values ranged from 16.0 to 24.0 °C.
Four of the five study reaches exhibited MWMT values above 18 °C
(Table 1). The highest mean Tand MWMT values occurred at SR1,
which was located on the mainstem Shasta River above the Big
Springs Creek confluence, and thus was not influenced by cool
water delivered by Big Springs Creek. Maximum diel temperature
fluctuations occurred at SR1 (⌬T= 8.5 °C) on 17 July (day 8 of our
study), BS2 (⌬T= 7.4 °C) on 10 July (day 1 of our study), and SR2 (⌬T=
7.3 °C) on 17 July (day 8 of our study). In contrast, study reaches BS1
(⌬T= 4.8 °C) and SR3 (⌬T= 4.8 °C) exhibited substantially less
thermal variability, with maximum diel fluctuations occurring on
10 July (day 2 of our study; Fig. 2).
Fig. 2. Time series of daily maximum, mean, and minimum water temperatures during the study period at each study reach: BS1 (a), BS2 (b),
SR1 (c), SR2 (d), and SR3 (e).
418 Can. J. Fish. Aquat. Sci. Vol. 77, 2020
Published by NRC Research Press
Can. J. Fish. Aquat. Sci. Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Davis on 02/03/20
For personal use only.

Benthic invertebrates
There were significant differences in both invertebrate density
(F
[4,53]
= 3.96, p< 0.01; Tukey’s HSD, p< 0.05; Fig. 3a) and inverte-
brate biomass (F
[4,53]
= 7.21, p< 0.01; Tukey’s HSD, p< 0.05; Fig. 3b)
among the five study reaches. Mean invertebrate density was
greatest at SR2 (mean ± SE = 64 106 ± 9597 individuals·m
−2
) and
lowest at BS1 (32 019 ± 2408 individuals·m
−2
), a 2.0-fold difference.
Even greater variation occurred among individual samples. Max-
imum invertebrate density during the entire study period was
recorded at SR2 during August (139 434 individuals·m
−2
), while
the lowest density (20 122 individuals·m
−2
) was observed at BS1
during July, a 6.9-fold difference in abundance. Mean invertebrate
biomass was greatest at SR1 (9.88 ± 3.37 g AFDM·m
−2
) followed by
SR2 (7.76 ± 1.37 g AFDM·m
−2
), whereas the lowest mean inverte-
brate biomass was observed at BS1 (2.77 ± 0.45 g AFDM·m
−2
;
Fig. 3b). Invertebrate density and biomass were positively associ-
ated, but not strongly so (linear regression, r
2
= 0.34, p< 0.01),
indicating a decoupling between the two variables.
Coho salmon growth and performance
We recovered 113 (75.3%) of the 150 age-0+ coho salmon initially
placed in our experimental enclosures. Of the 37 unrecovered fish,
31 individuals escaped (confirmed via subsequent detections on
stationary PIT tag antenna arrays located in the basin), four car-
casses were recovered from SR1, and two carcasses were removed
from SR3. We found a significant difference in coho salmon absolute
growth rate among reaches (ANOVA, F
[4,15]
= 32.38, p< 0.01;
Tukey’s HSD, p< 0.05; Fig. 4). Mean absolute growth was greatest
at SR2 (0.15 ± 0.01 g·day
−1
), followed by SR1 (0.11 ± 0.01 g·day
−1
), and
lowest at BS1 (0.02 ± 0.01 g·day
−1
). Age-0+ coho salmon reared at
study reaches SR2 and BS1 exhibited a mean change in mass of 9.1
and 1.4 g, respectively, a 6.5-fold difference in growth. Nearly all
Fig. 3. Mean (+1 standard error) invertebrate density (a) and biomass (b; AFDM = ash-free dry matter) at each enclosure study reach. Different
lowercase letters above the bars indicate statistically significant differences between reaches (ANOVA followed by Tukey’s honestly significant
different test, p< 0.05). Study reaches sharing the same letter are not different. Parenthetical values under each study reach are the mean
daily average water temperature (mean T) followed by maximum weekly maximum temperature (MWMT) observed during the study period.
Note: SR1 is located upstream of Big Springs Creek and was not influenced by major spring sources.
Lusardi et al. 419
Published by NRC Research Press
Can. J. Fish. Aquat. Sci. Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Davis on 02/03/20
For personal use only.

coho salmon (106 of 113; 93.8%) recovered at the end of the 63-day
experimental period exhibited positive absolute growth rates.
The exception to this trend occurred at reach BS1 (the reach with
the lowest mean T, MWMT, and invertebrate density; Table 1)
where 7 of 29 fish (24.1%) exhibited absolute growth rates ≤ 0.0 g·day
−1
during the study.
Model comparisons
The top-ranked model describing the absolute growth of juve-
nile coho salmon during our experiment included scaled inverte-
brate density as the lone parameter (AIC
c
weight = 67.0%), and all
models that included a parameter for invertebrate density re-
ceived >99% of the AIC
c
weight (Table 2). The parameter estimates
for scaled invertebrate density were positive and did not overlap
zero (parameter estimate = 0.04, 95% CI = 0.03 to 0.05), indicating
a reliably positive effect of invertebrate density on coho salmon
absolute growth rate (Fig. 5). Model estimates for coho salmon
growth were 4.1 times greater at SR2 than at BS1, the reaches with
the highest (64 106 individuals·m
−2
) and lowest (32 019 individuals·m
−2
)
overall invertebrate densities, respectively. The food + tempera-
ture model received some AIC
c
weight (⌬AIC
c
= 1.6, AIC
c
weight =
30%; Table 2), indicating an additive effect of increasing water
temperature on age-0+ coho salmon growth. The parameter esti-
mates for both factors were positive and did not overlap zero (food
parameter estimate = 0.03, 95% CI = 0.02 to 0.04; temperature
parameter estimate = 0.02, 95% CI = 0.01 to 0.03). A plot of the
residuals from the invertebrate density model against mean T
confirmed that the influence of temperature was positive, but
showed a possible nonlinearity between growth and temperature
(Fig. S1
1
). Mean coho salmon growth peaked at a mean Tof 16.6 °C
and MWMT of 21.1 °C. Mean absolute growth rates from low to
high values were 0.02 g·day
−1
at BS1 (mean T= 13.0 °C; MWMT =
16.0 ° C), 0.05 g·day
−1
at BS2 (mean T= 14.8 °C; MWMT = 18.9 °C),
0.07 g·day
−1
at SR3 (mean T= 17.1 °C; MWMT = 20.6 °C), 0.11 g·day
−1
at SR1 (mean T= 18.1 °C; MWMT = 24.0 °C), and 0.15 g·day
−1
at SR2
(mean T= 16.6 °C; MWMT = 21.1 °C) (Fig. 5). Thus, while coho
salmon growth remained positive (and above average) at the high-
est mean T(18.1 °C) and MWMT (24.0 °C) (reach SR1; Table 1),
absolute growth was maximized at a mean Tof 16.6 °C and MWMT
of 21.1 °C (reach SR2; Table 1). Neither water velocity nor depth
were a strong predictor of coho salmon growth (Table 2).
While coho salmon mortality was generally low during our exper-
iment, mortality was elevated at SR1, the study reach with the high-
est MWMT (Fig. 6). In order of increasing MWMT values at each
study reach, mean mortality rates were 0.0% at BS1 (MWMT = 16.0 °C),
0.0% at BS2 (MWMT = 18.9 °C), 6.7% at SR3 (MWMT = 20.6 °C), 0.0%
at SR2 (MWMT = 21.1 °C), and 13% at SR1 (MWMT = 24.0 °C). The
MWMT model received markedly more AIC
c
support than the
intercept model (⌬AIC
c
= 6.2, 95.9% of the AIC
c
weight; Table 3).
The temperature parameter was positive and did not overlap zero
(parameter estimate = 0.57, 95% CI = 0.10 to 1.03), indicating that
age-0+ coho salmon mortality reliably increased with MWMT
(Fig. 6).
Table 2. Model comparison for juvenile coho
salmon growth rate.
Model df ⌬AIC
c
AIC
c
weight
G⬃D4 0 0.670
G⬃T+D5 1.6 0.302
G⬃Velocity + D5 8.2 0.011
G⬃P
Macro
+D5 9.3 0.006
G⬃Depth + D5 9.9 0.005
G⬃D+T+D×T6 11.1 0.003
G⬃T4 12.5 0.001
G⬃P
Macro
4 12.6 0.001
G⬃Depth 4 17.3 <0.001
G⬃(Intercept only) 3 20.1 <0.001
G⬃Velocity 4 29.1 <0.001
Note: The model parameter Grepresents absolute
growth rate of age-0+ coho salmon (g·day
–1
), Dis inverte-
brate density, Tis mean daily average water temperature
(mean T), and P
Macro
is the proportion of the stream chan-
nel occupied by aquatic macrophytes. df is the degrees of
freedom, and ⌬AIC
c
is the difference in AIC
c
between the
model of interest and the top-ranked model.
Fig. 4. Absolute growth rate of age-0+ coho salmon reared in experimental enclosures at five different locations in the Shasta River basin
(center line, median; box limits, upper and lower quartiles; whiskers, maximum and minimum interquartile range). Different lowercase
letters above box and whiskers indicate a statistically significant difference between study reaches (ANOVA followed by Tukey’s honestly
significant different test, p< 0.05). Study reaches sharing the same letter are not different. Parenthetical values under each study reach are
the mean daily average water temperature (mean T) followed by maximum weekly maximum temperature (MWMT) observed during the
study period. Note: SR1 is located upstream of Big Springs Creek and was not influenced by major spring sources.
420 Can. J. Fish. Aquat. Sci. Vol. 77, 2020
Published by NRC Research Press
Can. J. Fish. Aquat. Sci. Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Davis on 02/03/20
For personal use only.

Discussion
Our study describes how natural spatiotemporal gradients in ther-
mal conditions, prey availability, and select environmental factors
affected the growth and survival of enclosure-reared age-0+ coho
salmon during the critical summer period. We found that variabil-
ity in food resources (invertebrate prey densities) among study
reaches most strongly explained the growth of age-0+ coho salmon
during the summer low flow period, across mean Tvalues ranging
from 13.0 to 18.1 °C and MWMT values ranging from 16.0 to 24.0 °C.
Even at the highest temperatures observed during our study
(mean T= 18.1 °C; MWMT = 24.0 °C), juvenile coho salmon growth
rates were positive and above the study-wide average, although
mortality rates peaked at this elevated temperature. Our results in-
dicate that ecosystem productivity may help buffer the negative
effects of elevated water temperature on juvenile salmonid
growth and survival, with important implications for climate
change.
We found that juvenile coho salmon growth rates peaked at a
mean Tof 16.6 °C and MWMT of 21.1 °C and were six times greater
than those observed at the coolest reach (BS1), which exhibited a
mean Tof 13.0 °C and MWMT of 16 °C. Previous studies have sug-
gested that optimal stream temperatures for coho salmon growth
are less than those in our experiment. McCullough (1999) reported
that juvenile coho salmon growth was maximized at 15 °C, while
Hicks (2002) proposed that weekly average temperatures of 14–
15 °C would likely be the most beneficial for coho salmon growth.
Moyle (2002) suggested an optimal temperature range between 12
and 14 °C for juvenile coho salmon populations in California,
while Stenhouse et al. (2012) recommended an instantaneous
maximum temperature of 15.5 °C for coho salmon in the Shasta
River (our study system). Differences in growth and temperature
across disparate studies and geographies may be a function of
local adaptation and (or) acclimation. However, the aforemen-
tioned recommendations suggest that water temperatures exceed-
ing ⬃16 °C likely constrain juvenile coho salmon growth and, by
extension, diminish habitat suitability to some degree. The results of
our study, however, are inconsistent with these recommendations.
To reconcile our findings with reported temperature optima,
we propose that abundant trophic resources can help mitigate the
negative effects of elevated water temperature on juvenile salmonid
growth, particularly in cases where instantaneous maximum tem-
peratures remain well below upper thermal limits and (or) where
sufficient diel temperature fluctuations occur. There are clear
metabolic trade-offs between water temperature and prey con-
sumption (Hanson et al. 1997). Growth of ectotherms is a function
of the amount of energy consumed and net energy expended
through respiration, in addition to metabolic costs associated with
egestion and excretion. Energy consumption is dictated by prey
availability (capture efficiency), while energy expenditure in-
creases with temperature and discharge, primarily through inten-
sified respiration. Most salmon-bearing streams are oligotrophic
and abundant food sources are typically rare, particularly at temper-
ate latitudes (Myrvold and Kennedy 2015). Consequently, factors
other than food (both physiological and ecological) often con-
strain salmonid growth in many riverine ecosystems, and this
may explain the paucity of empirical evidence directly linking
juvenile salmonid growth to food abundance.
Nevertheless, a few field studies have provided evidence that
food availability influences juvenile salmonid growth and (or) habi-
tat selection. Bisson et al. (1988) reported robust rates of coho
salmon production in several Washington (USA) streams where
maximum temperatures ranged from 24.5 to 29.5 °C, and the
authors hypothesized that food resources played an important
role in those observations. Railsback and Rose (1999) found that
variability in the summer growth of rainbow trout (resident
O. mykiss) in the Sierra Nevada of California was a function of food
availability rather than temperature, and Boughton et al. (2007)
demonstrated that augmented food resources enhanced the growth
rate (but not survival) of juvenile steelhead reared in enclosures in a
southern California stream. In an experiment conducted in artifi-
cial channels adjacent to a British Columbia stream, enhanced
food availability enabled juvenile coho salmon to expand foraging
ranges into higher-velocity habitats where energy expenditures
were presumably greater (Rosenfeld et al. 2005). In another study,
Rosenfeld and Raeburn (2009) found that prey availability associ-
Fig. 6. Mortality of age-0+ coho salmon reared in experimental
enclosures as a function of maximum weekly maximum temperature
(MWMT). Points represent mean fish mortality in each replicate
enclosure. Shaded area represents the 95% confidence interval of the
model.
Table 3. Comparison of the temperature and null
model used to estimate mortality of juvenile coho
salmon.
Model df ⌬AIC
c
AIC
c
weight
P
Mort
⬃T2 0 0.959
P
Mort
⬃(Intercept only) 1 6.3 0.041
Note: P
Mort
is proportion mortality and Tis tempera-
ture (MWMT). df is the degrees of freedom, and ⌬AIC
c
is
the difference in AIC
c
between the model of interest and
the top-ranked model.
Fig. 5. Absolute growth rates of age-0+ coho salmon reared in
experimental enclosures as a function of overall invertebrate
density (prey availability). Invertebrate density represents the mean
invertebrate density at each enclosure study reach. Shaded area
represents the 95% confidence interval of the model.
Lusardi et al. 421
Published by NRC Research Press
Can. J. Fish. Aquat. Sci. Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Davis on 02/03/20
For personal use only.

ated with different habitats accounted for differences in juvenile
coho salmon growth rates. Our work builds on these studies and
demonstrates that coho salmon can grow rapidly at a mean Tof
16.6 °C (and MWMT of 21.1 °C) when prey resources are abundant.
We acknowledge that the results obtained in our experiment
were likely influenced by the use of enclosures, which prohibited
dispersal by juvenile coho salmon in response to stressful thermal
conditions (i.e., behavioral thermoregulation). Moreover, it is proba-
ble that elements of the thermal regime beyond mean Tand
MWMT are important determinants of habitat suitability for ju-
venile salmonids. Despite diel temperature fluctuation not play-
ing a prominent role as a predictor of age-0+ coho growth in our
study, such thermal variability is presumed to reduce heat stress
and metabolic demand. Chesney et al. (2009) described age-0+
coho salmon rearing at weekly maximum temperatures up to
22.4 °C in the Shasta River when sufficient diel thermal fluctua-
tions occurred (i.e., daily minimum temperatures of ⬃13 °C) and
speculated that rearing at elevated water temperatures was also
facilitated by abundant trophic resources. Substantial diel ther-
mal fluctuations (⌬Tup to 8.5 °C) occurred during our study, pos-
sibly enabling juvenile coho salmon to recover from prior thermal
stress and perhaps reducing overall study mortality, although
such fluctuations were not indicative of growth. The role of diel
thermal fluctuations on juvenile coho salmon growth in the wild
has not been extensively studied, but should be examined more
closely, particularly as it interacts with prey availability.
Our findings add to a growing literature showing juvenile coho
salmon can persist and grow in environments with elevated water
temperatures when sufficient food is available. While a broader
understanding of the interactions between water temperature
and food availability can provide context in understanding poten-
tial limiting factors on salmonid population viability, the results
presented here should not be misconstrued as an endorsement of
high water temperatures in lieu of functioning stream and ri-
parian habitat. Elevated water temperatures have been shown to
negatively affect obligate cold-water species through different path-
ways, both direct and indirect (Morita et al. 2015). However, there is
mounting evidence that fish often reside in locations that exceed
purported thermal optima when such locations (or nearby habitats)
exhibit thermal heterogeneity over time and (or) space (Brewitt et al.
2017). Movements between thermally heterogeneous environments
are often cyclical, with salmonids dispersing from physiologically
optimal cold-water habitat to warmer water to seek enhanced
foraging and growth opportunities (Brewitt et al. 2017;Osterback
et al. 2018). Conversely, juvenile salmonids rearing at more north-
ern latitudes have been shown to disperse from cooler to warmer
habitats to accelerate metabolic processes and growth (Armstrong
et al. 2013).
Watersheds are context-specific environments characterized by
a range of physical and biological factors that interact to influence
the growth, distribution, and fitness of salmonids. An improved
understanding of these interactions, and specifically the role that
food webs play in supporting salmonid rearing, is critical for
successful conservation and management of imperiled salmonid
populations. Our results suggest that enhanced food resources
may allow juvenile salmonids to persist in habitats that would be
deemed suboptimal based on temperature criteria alone. This may
be particularly true in watersheds with intrinsically high rates of
secondary production or in highly productive seasonal habitats
such as coastal freshwater lagoons (e.g., Bond et al. 2008;Hayes
et al. 2008;Osterback et al. 2018). Several recent studies have
called for a broader inclusion of prey availability into salmonid
habitat selection and suitability models (e.g., Wipfli and Baxter
2010;Weber et al. 2014), and the results presented here strongly
support this recommendation.
The potential for highly productive habitats to ameliorate the
negative effects of high stream temperature has important impli-
cations for salmonid conservation and management under cli-
mate change (Lusardi et al. 2016;Moyle et al. 2017). Elevated water
temperatures have been shown to strongly limit the distribution
and abundance of salmonids (Roper et al. 1994;Welsh et al. 2001;
Myrick and Cech 2004) with potentially important fitness conse-
quences (Angilletta et al. 2008). Climate change is predicted to
further increase stream water temperature (Morrill et al. 2005;
Isaak et al. 2012), affecting salmonids and other cold-water fishes
in numerous ways. Wenger et al. (2011) projected nearly a 50%
decline in cold-water trout habitat throughout the western United
States by 2080. Salmonids at the southern edge of their native
range are particularly vulnerable to climate change because water
temperatures regularly approach purported thermal tolerances in
many streams (Katz et al. 2013;Moyle et al. 2017). Our findings, there-
fore, suggests that food availability may offset negative effects of
increasing water temperature in highly productive ecosystems.
The climate refugia literature to date has largely focused on eco-
systems that provide cold-water habitats (e.g., Isaak et al. 2015). We
suggest that habitats with inherently high rates of secondary pro-
duction, such as spring-fed rivers, floodplains, seasonal lagoons,
and estuaries, will likely become increasingly important to the
viability of salmonids under future warming scenarios, and such
habitats should be emphasized in climate-adaptation strategies.
In general, however, these habitats are comparatively rare, under-
studied, and have often experienced considerable change due to
water development and urbanization, particularly in California
(Moyle et al. 2017). Further research is warranted to assess the
ability of such habitats to support juvenile salmonids and possibly
buffer the effects of climate warming on salmonid populations.
Acknowledgements
We thank Kyle Phillips, Mollie Ogaz, Eric Holmes, Nick Corline,
Drew Nichols, Ari Fingeroth, Derek Drolette, Nick Depsky, and
Kristen Hanson for their help building enclosures and collecting
data. We are indebted to Cynthia Kern for assistance processing
invertebrate samples, to Haven Kiers for assistance with figures,
and to Mike Deas for lodging during the experimental period. We
are also grateful to Ada Fowler, Chris Babcock, Andrew Braugh,
Curtis Knight, The Nature Conservancy, and California Trout for
assistance in the field. Finally, we thank Chris Adams, Bill Chesney,
and Caitlin Bean for help identifying study reaches and for sharing
their knowledge of the basin. Early drafts of this manuscript were
substantially improved with recommendations from Brian Spence
and two anonymous reviewers. Research using ESA-listed sal-
monids was authorized by National Marine Fisheries Service un-
der Section 4(d) permit No.17571. All animals were handled and
housed according to Institutional Animal Care and Use Protocols
(IACUC UC Davis, No. 18883).
References
Angilletta, M.J., Ashley Steel, E., Bartz, K.K., Kingsolver, J.G., Scheuerell, M.D.,
Beckman, B.R., and Crozier, L.G. 2008, Big dams and salmon evolution:
changes in thermal regimes and their potential evolutionary consequences.
Evol. Appl. 1: 286–299. doi:10.1111/j.1752-4571.2008.00032.x.
Armstrong, J.B., Schindler, D.E., Ruff, C.P., Brooks, G.T., Bentley, K.E., and
Torgersen, C.E. 2013. Diel horizontal migration in streams: Juvenile fish ex-
ploit spatial heterogeneity in thermal and trophic resources. Ecology, 94(9):
2066–2075. doi:10.1890/12-1200.1. PMID:24279277.
Bates, D., Mächler, M., Bolker, B., and Walker, S. 2015. Fitting linear mixed-
effects models using lme4. J. Stat. Softw. 67(1): 1–48. doi:10.18637/jss.v067.i01.
Becker, C., and Fujihara, M. 1978. The bacterial pathogen Flexibacter columnaris
and its epizootiology among Columbia River fish: a review and synthesis.
American Fisheries Society, Washington, D.C.
Becker, R.A., Chambers, J.M., and Wilks, A.R. 1988. The New S Language - a
programming environment for data analysis and graphics. 22(3): 785–795.
https://doi.org/10.1577/1548-8675(2002)022<0785:EODAVP>2.0.CO;2. Chap-
man and Hall, New York.
Bell, M.C. 1986. Fisheries handbook of engineering requirements and biological
criteria. US Army Corps of Engineers, Fish Passage Development and Evalu-
ation Program, North Pacific Division, Portland, Ore.
Bisson, P.A., Nielsen, J.L., and Ward, J.W. 1988. Summer production of coho
salmon stocked in Mount St. Helens streams 3-6 years after the 1980 eruption.
422 Can. J. Fish. Aquat. Sci. Vol. 77, 2020
Published by NRC Research Press
Can. J. Fish. Aquat. Sci. Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Davis on 02/03/20
For personal use only.

Trans. Am. Fish. Soc. 117(4): 322–335. doi:10.1577/1548-8659(1988)117<0322:
SPOCSS>2.3.CO;2.
Bolker, B.M. 2008. Ecological models and data in R. Princeton University Press,
Princeton, N.J.
Bolker, B. 2016. bbmle: tools for general maximum likelihood estimation.
R package version 1.0.18.
Bond, M.H., Hayes, S.A., Hanson, C.V., and MacFarlane, R.B. 2008. Marine sur-
vival of steelhead (Oncorhynchus mykiss) enhanced by a seasonally closed estu-
ary. Can. J. Fish. Aquat. Sci. 65(10): 2242–2252. doi:10.1139/F08-131.
Boughton, D.A., Gibson, M., Yedor, R., and Kelley, E. 2007. Stream temperature
and the potential growth and survival of juvenile Oncorhynchus mykiss in a
southern California Creek. Freshw. Biol. 52(7): 1353–1364. doi:10.1111/j.1365-
2427.2007.01772.x.
Brett, J.R. 1952. Temperature tolerance in young Pacific salmon, genus
Oncorhynchus. J. Fish. Res. Board Can. 9(6): 265–323. doi:10.1139/f52-016.
Brewitt, K.S., Danner, E.M., and Moore, J.W. 2017. Hot eats and cool creeks:
juvenile Pacific salmonids use mainstem prey while in thermal refuges. Can.
J. Fish. Aquat. Sci. 74(10): 1588–1602. doi:10.1139/cjfas-2016-0395.
Burnham, K.P., and Anderson, D.R. 2002. Model selection and multimodel
inference: a practical information-theoretic approach. Springer, New York.
Caton, L. 1991. Improved subsampling methods for the EPA “Rapid Bioassess-
ment” benthic protocols. Bull. N. Am. Benthol. Soc. 8(3): 317–319.
Chesney, W., Adams, C., Crombie, W., Langendorf, H., Stenhouse, S., and Kirkby, K.
2009. Shasta River juvenile coho habitat and migration study. Prepared for
US Bureau of Reclamation, Klamath Area Office by California Department of
Fish and Game, Yreka, Calif.
Cotton, C., Walker, R., and Recicar, T. 2003. Effects of temperature and salinity
on growth of juvenile black sea bass, with implications for aquaculture.
N. Am. J. Aquac. 65(4): 330–338. doi:10.1577/C02-037.
Coutant, C.C. 1977. Compilation of temperature preference data. J. Fish. Res.
Board Can. 34(5): 739–745. doi:10.1139/f77-115.
Fonds, M., Cronie, R., Vethaak, A.D., and Van Der Puyl, P. 1992. Metabolism, food
consumption and growth of plaice (Pleuronectes platessa) and flounder
(Platichthys flesus) in relation to fish size and temperature. Netherl. J. Sea Res.
29(1–3): 127–143. doi:10.1016/0077-7579(92)90014-6.
Goniea, T.M., Keefer, M.L., Bjornn, T.C., Peery, C.A., Bennett, D.H., and
Stuehrenberg, L.C. 2006. Behavioral thermoregulation and slowed migration
by adult fall Chinook salmon in response to high Columbia River water
temperatures. Trans. Am. Fish. Soc. 135(2): 408–419. doi:10.1577/T04-113.1.
Hammock, B.G., and Wetzel, W.C. 2013. The relative importance of drift causes
for stream insect herbivores across a canopy gradient. Oikos, 122(11): 1586–
1593. doi:10.1111/j.1600-0706.2013.00319.x.
Hanson, P.C., Johnson, T.B., Schindler, D.E., and Kitchell, J.F. 1997. Fish Bioener-
getics 3.0. Madison, Wisc.
Hayes, S.A., Bond, M.H., Hanson, C.V., Freund, E.V., Smith, J.J., Anderson, E.C.,
et al. 2008. Steelhead growth in a small central California watershed: up-
stream and estuarine rearing patterns. Trans. Am. Fish. Soc. 137(1): 114–128.
doi:10.1577/T07-043.1.
He, L.-M., and Marcinkevage, C. 2017. Incorporating thermal requirements into
flow regime development for multiple Pacific salmonid species in regulated
rivers. Ecol. Eng. 99: 141–158. doi:10.1016/j.ecoleng.2016.11.009.
Hicks, M. 2002. Evaluating standards for protecting aquatic life in Washington’s
surface water quality standards, Draft discussion paper and literature sum-
mary. Publ. No. 00-10-070 [online]. Washington State Department of Ecology,
Olympia, Wash. Available from https://fortress.wa.gov/ecy/publications/
documents/0010070.pdf.
Hildebrand, S.G. 1974. The relation of drift to benthos density and food level in
an artificial stream. Limnol. Oceanogr. 19(6): 951–957. doi:10.4319/lo.1974.19.
6.0951.
Isaak, D.J., Wollrab, S., Horan, D., and Chandler, G. 2012. Climate change effects
on stream and river temperatures across the northwest U.S. from 1980–2009
and implications for salmonid fishes. Clim. Change, 113(2): 499–524. doi:10.
1007/s10584-011-0326-z.
Isaak, D.J., Young, M.K., Nagel, D.E., Horan, D.L., and Groce, M.C. 2015. The
cold-water climate shield: delineating refugia for preserving salmonid fishes
through the 21st century. Glob. Change Biol. 21(7): 2540–2553. doi:10.1111/gcb.
12879.
Jeffres, C.A., and Moyle, P.B. 2012. When good fish make bad decisions: coho
salmon in an ecological trap. N. Am. J. Fish. Manage. 32(1): 87–92. doi:10.1080/
02755947.2012.661389.
Jeffres, C.A., Dahlgren, R.A., Deas, M.L., Kiernan, J.D., King, A.M., Lusardi, R.A.,
et al. 2009. Baseline assessment of physical and biological conditions within
waterways on Big Springs Ranch, Siskiyou County, California. Report pre-
pared for California State Water Resources Control Board [online]. Available
from https://watershed.ucdavis.edu/pdf/Jeffres-et-al-SWRCB-2009.pdf.
Katz, J., Moyle, P.B., Quiñones, R.M., Israel, J., and Purdy, S. 2013. Impending
extinction of salmon, steelhead, and trout (Salmonidae) in California. Envi-
ron. Biol. Fishes, 96(10–11): 1169–1186. doi:10.1007/s10641-012-9974-8.
Kennedy, T.A., Yackulic, C.B., Cross, W.F., Grams, P.E., Yard, M.D., and Copp, A.J.
2014. The relation between invertebrate drift and two primary controls, dis-
charge and benthic densities, in a large regulated river. Freshw. Biol. 59(3):
557–572. doi:10.1111/fwb.12285.
Konecki, J.T., Woody, C.A., and Quinn, T.P. 1995. Critical thermal maxima of
coho salmon (Oncorhynchus kisutch) fry under field and laboratory acclimation
regimes. Can. J. Zool. 73(5): 993–996. doi:10.1139/z95-117.
Kovach, R.P., Joyce, J.E., Echave, J.D., Lindberg, M.S., and Tallmon, D.A. 2013.
Earlier migration timing, decreasing phenotypic variation, and biocomplex-
ity in multiple salmonid species. PLoS ONE, 8(1): e53807. doi:10.1371/journal.
pone.0053807. PMID:23326513.
Lehman, B., Huff, D.D., Hayes, S.A., and Lindley, S.T. 2017. Relationships between
Chinook salmon swimming performance and water quality in the San
Joaquin River, California. Trans. Am. Fish. Soc. 146(2): 349–358. doi:10.1080/
00028487.2016.1271827.
Lusardi, R.A., Bogan, M.T., Moyle, P.B., and Dahlgren, R.A. 2016. Environment
shapes invertebrate assemblage structure differences between volcanic
spring-fed and runoff rivers in northern California. Freshw. Sci. 35(3): 1010–
1022. doi:10.1086/687114.
Lusardi, R.A., Jeffres, C.A., and Moyle, P.B. 2018. Stream macrophytes increase
invertebrate production and fish habitat utilization in a California stream.
River Res. Appl. 34(8): 1003–1012. doi:10.1002/rra.3331.
Mack, S. 1960. Geology and ground-water features of Shasta Valley, Siskiyou
County, California. US Geological Survey Water-Supply Paper 1484. US Gov-
ernment Printing Office, Washington, D.C.
Mantua, N., Tohver, I., and Hamlet, A. 2010. Climate change impacts on stream-
flow extremes and summertime stream temperature and their possible con-
sequences for freshwater salmon habitat in Washington State. Clim. Change,
102(1–2): 187–223. doi:10.1007/s10584-010-9845-2.
Martin, D.J., Wasserman, L.J., and Dale, V.H. 1986. Influence of riparian vegeta-
tion on posteruption survival of coho salmon fingerlings on the west-side
streams of Mount St. Helens, Washington. N. Am. J. Fish. Manage. 6(1): 1–8.
doi:10.1577/1548-8659(1986)6<1:IORVOP>2.0.CO;2.
Mauger, S., Shaftel, R., Leppi, J.C., and Rinella, D.J. 2017. Summer temperature
regimes in southcentral Alaska streams: watershed drivers of variation and
potential implications for Pacific salmon. Can. J. Fish. Aquat. Sci. 74(5): 702–
715. doi:10.1139/cjfas-2016-0076.
McCullough, D.A. 1999. A review and synthesis of the effects of alterations to the
water temperature regime on freshwater life stages of salmonids, with spe-
cial reference to Chinook salmon. US Environmental Protection Agency,
Seattle, Wash.
McElreath, R. 2016. Statistical rethinking: a Bayesian course with examples in
R and Stan. CRC Press, Boca Raton, Fla.
Miller, K.M., Teffer, A., Tucker, S., Li, S., Schulze, A.D., Trudel, M., et al. 2014.
Infectious disease, shifting climates, and opportunistic predators: cumula-
tive factors potentially impacting wild salmon declines. Evol. Appl. 7(7): 812–
855. doi:10.1111/eva.12164. PMID:25469162.
Moore, R.D., Nelitz, M., and Parkinson, E. 2013. Empirical modelling of maxi-
mum weekly average stream temperature in British Columbia, Canada, to
support assessment of fish habitat suitability. Can. Water Resour. J. 38(2):
135–147. doi:10.1080/07011784.2013.794992.
Morita, K., Ayumi, N., and Kikuchi, M. 2015. River temperature drives salmon
survivorship: is it determined prior to ocean entry? R. Soc. Open Sci. 2(1):
140312. doi:10.1098/rsos.140312. PMID:26064583.
Morrill, J.C., Bales, R.C., and Conklin, M.H. 2005. Estimating stream temperature
from air temperature: implications for future water quality. J. Environ. Eng.
131(1): 139–146. doi:10.1061/(ASCE)0733-9372(2005)131:1(139).
Moyle, P.B. 2002. Inland fishes of California. University of California Press,
Berkeley, Calif.
Moyle, P.B., Kiernan, J.D., Crain, P.K., and Quiñones, R.M. 2013. Climate change
vulnerability of native and alien freshwater fishes of California: a system-
atic assessment approach. PLoS ONE, 8(5): e63883. doi:10.1371/journal.pone.
0063883. PMID:23717503.
Moyle, P.B., Lusardi, R.A., Samuel, P.J., and Katz, J.V.E. 2017. State of the
salmonids: status of California’s emblematic fishes [online]. Center for Wa-
tershed Sciences, University of California, Davis, and California Trout, San
Francisco, Calif. Available from https://watershed.ucdavis.edu/files/content/
news/SOS%20II_Final.pdf.
Myrick, C.A., and Cech, J.J. 2004. Temperature effects on juvenile anadromous
salmonids in California’s central valley: what don’t we know? Rev. Fish Biol.
Fish. 14: 113–123. doi:10.1007/s11160-004-2739-5.
Myrvold, K.M., and Kennedy, B.P. 2015. Interactions between body mass and
water temperature cause energetic bottlenecks in juvenile steelhead. Ecol.
Freshw. Fish, 24(3): 373–383. doi:10.1111/eff.12151.
NMFS. 2014. Final recovery plan for the Southern Oregon/Northern California
Coast evolutionarily significant unit of coho salmon (Oncorhynchus kisutch).
National Marine Fisheries Service, Arcata, Calif.
NRC. 2004. Endangered and threatened fishes in the Klamath River basin. Na-
tional Research Council. National Academies Press, Washington, D.C. doi:10.
17226/10838.
Nichols, A.L., Willis, A.D., Jeffres, C.A., and Deas, M.L. 2013. Water temperature
patterns below large groundwater springs: management implications for
coho salmon in the Shasta River, California. River Res. Appl. 30(4): 442–455.
doi:10.1002/rra.2655.
Nielsen, J.L. 1992. Microhabitat-specific foraging behavior, diet, and growth of
juvenile coho salmon. Trans. Am. Fish. Soc. 121(5): 617–634. doi:10.1577/1548-
8659(1992)121<0617:MFBDAG>2.3.CO;2.
Null, S.E., Deas, M.L., and Lund, J.R. 2009. Flow and water temperature simula-
Lusardi et al. 423
Published by NRC Research Press
Can. J. Fish. Aquat. Sci. Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Davis on 02/03/20
For personal use only.

tion for habitat restoration in the Shasta River, California. River Res. Appl.
26(6): 663–681. doi:10.1002/rra.1288.
Obedzinski, M., Nossaman Pierce, S., Horton, G.E., and Deitch, M.J. 2018. Effects
of flow-related variables on oversummer survival of juvenile coho salmon in
intermittent streams. Trans. Am. Fish. Soc. 147(3): 588–605. doi:10.1002/tafs.
10057.
Osterback, A.-M.K., Kern, C.H., Kanawi, E.A., Perez, J.M., and Kiernan, J.D. 2018.
The effects of early sandbar formation on the abundance and ecology of coho
salmon (Oncorhynchus kisutch) and steelhead trout (Oncorhynchus mykiss)ina
central California coastal lagoon. Can. J. Fish. Aquat. Sci. 75(12): 2184–2197.
doi:10.1139/cjfas-2017-0455.
Pinheiro, J., Bates, D., DebRoy, S., Sarkar, D., and R Core Team. 2018. nlme: linear
and nonlinear mixed effects models. R package version 3.1-137 [online]. Avail-
able from https://cran.r-project.org/package=nlme [accessed 18 June 2018].
R Core Team. 2015. R: a language and environment for statistical computing.
R Foundation for Statistical Computing, Vienna.
Railsback, S.F., and Rose, K.A. 1999. Bioenergetics modeling of stream trout
growth: temperature and food consumption effects. Trans. Am. Fish. Soc.
128: 241–256. doi:10.1577/1548-8659(1999)128<0241:BMOSTG>2.0.CO;2.
Reeves, G.H., Everest, F.H., and Nickelson, T.E. 1989. Identification of physical
habitats limiting the production of coho salmon in western Oregon and
Washington. Gen. Tech. Rep. PNW-GTR-245. US Department of Agriculture,
Forest Service, Pacific Northwest Research Station, Portland, Ore.
Reiser, D.W., and Bjorn, T.J. 1979. Habitat requirements of anadromous sal-
monids. In Influence of forest and rangeland management on anadromous
fish habitat in the Western United States and Canada. Edited by W.R. Meehan.
US Forest Service Gen. Tech. Rep. PNW-96, Northwest. Forest Range Exp. Sta.,
Portland, Ore.
Roper, B.B., Scarnecchia, D.L., and La Marr, T.J. 1994. Summer distribution of and
habitat use by Chinook salmon and steelhead within a major basin of the
South Umpqua River, Oregon. Trans. Am. Fish. Soc. 123: 298–308. doi:10.1577/
1548-8659(1994)123<0298:SDOAHU>2.3.CO;2.
Rosenfeld, J.S., and Raeburn, E. 2009. Effects of habitat and internal prey
subsidies on juvenile coho salmon growth: implications for stream pro-
ductive capacity. Ecol. Freshw. Fish, 18(4): 572–584. doi:10.1111/j.1600-0633.
2009.00372.x.
Rosenfeld, J.S., Leiter, T., Lindner, G., and Rothman, L. 2005. Food abundance
and fish density alters habitat selection, growth, and habitat suitability
curves for juvenile coho salmon (Oncorhynchus kisutch). Can. J. Fish. Aquat. Sci.
62(8): 1691–1701. doi:10.1139/f05-072.
Spence, B.C., and Dick, E.J. 2014. Geographic variation in environmental factors
regulating outmigration timing of coho salmon (Oncorhynchus kisutch) smolts.
Can. J. Fish. Aquat. Sci. 71(1): 56–69. doi:10.1139/cjfas-2012-0479.
Stenhouse, S.A., Bean, C.E., Chesney, W.R., and Pisano, M.S. 2012. Water temper-
ature thresholds for coho salmon in a spring-fed river, Siskiyou County,
California. Calif. Fish Game, 98(1): 19–37.
EPA. 2003. EPA Region 10 Guidance for Pacific Northwest State and Tribal Tem-
perature Water Quality Standards Acknowledgments [online]. US Environ-
mental Protection Agency. Available from https://nepis.epa.gov/Exe/
ZyPDF.cgi/P1004IUI.PDF?Dockey=P1004IUI.PDF [accessed 15 June 2018].
Wales, J.H. 1951. The decline of the Shasta River king salmon run. In Inland
Fisheries Administrative Report 51-18. California Department of Fish and
Game, Sacramento, Calif.
Weber, N., Bouwes, N., and Jordan, C.E. 2014. Estimation of salmonid habitat
growth potential through measurements of invertebrate food abundance
and temperature. Can. J. Fish. Aquat. Sci. 71(8): 1158–1170. doi:10.1139/cjfas-
2013-0390.
Welsh, H.H., Hodgson, G.R., Harvey, B.C., and Roche, M.E. 2001. Distribution of
juvenile coho salmon in relation to water temperatures in tributaries of the
Mattole River, California. N. Am. J. Fish. Manage. 21: 464–470. doi:10.1577/
1548-8675(2001)021<0464:DOJCSI>2.0.CO;2.
Wenger, S.J., Isaak, D.J., Luce, C.H., Neville, H.M., Fausch, K.D., Dunham, J.B.,
et al. 2011. Flow regime, temperature, and biotic interactions drive differen-
tial declines of trout species under climate change. Proc. Natl. Acad. Sci.
108(34): 14175–14180. doi:10.1073/pnas.1103097108. PMID:21844354.
Wipfli, M.S., and Baxter, C.V. 2010. Linking ecosystems, food webs, and fish
production: subsidies in salmonid watersheds. Fisheries, 35(8): 373–387. doi:
10.1577/1548-8446-35.8.373.
Wurtsbaugh, W.A., and Davis, G.E. 1977. Effects of fish size and ration level on
the growth and food conversion efficiency of rainbow trout, Salmo gairdneri
Richardson. J. Fish Biol. 11(2): 99–104. doi:10.1111/j.1095-8649.1977.tb04102.x.
424 Can. J. Fish. Aquat. Sci. Vol. 77, 2020
Published by NRC Research Press
Can. J. Fish. Aquat. Sci. Downloaded from www.nrcresearchpress.com by Calif Dig Lib - Davis on 02/03/20
For personal use only.