Effect of hydroelectric dam operations on the freshwater productivity of a Columbia River fall Chinook salmon population

ARTICLE

Effect of hydroelectric dam operations on the freshwaterproductivity of a Columbia River fall Chinook salmonpopulationRyan A. Harnish, Rishi Sharma, Geoffrey A. McMichael, Russell B. Langshaw, and Todd N. Pearsons

Abstract: Altering the timing and magnitude of discharge fluctuations can minimize the adverse effects of operating hy-droelectric dams on the productivity of downstream salmon populations. Hydroelectric operations at Priest Rapids Dam duringthe mid-1970s resulted in dewatering of fall Chinook salmon (Oncorhynchus tshawytscha) redds, causing mortality of intragravel lifestages. Since then, a series of operational constraints have been implemented at Priest Rapids Dam to reduce the effects ofdischarge fluctuations on the population of fall Chinook salmon that spawns and rears downstream from the dam. Initialprotections that focused on preventing redd dewatering were subsequently increased to include postemergence life stages. Weused stock–recruit analyses to identify changes to the population’s freshwater productivity that occurred over a 30-year periodand coincided with changes to dam operations. We observed a 217% increase in productivity that corresponded with constraintsenacted to prevent redd dewatering and an additional 130% increase that coincided with enactment of constraints to limitstranding and entrapment of juveniles. The information gained from this study may be used to guide efforts elsewhere tomitigate the effects of hydroelectric dam operations on downstream fish populations.

Résumé : La modification du moment et de la magnitude des fluctuations des débits peut minimiser les effets néfastes del’exploitation de barrages hydroélectriques sur la productivité de populations de saumons en aval. Les activités hydroélectriquesau barrage de Priest Rapids, au milieu des années 1970, ont entraîné l’assèchement des nids de frai automnaux de saumonquinnat (Oncorhynchus tshawytscha), causant une mortalité parmi les stades de vie intragravier. Depuis, un ensemble de con-traintes opérationnelles a été mis en œuvre au barrage de Priest Rapids dans le but de réduire les effets de la fluctuation desdébits sur la population automnale de saumons quinnat qui fraie et grandit en aval du barrage. Les premières mesures deprotection, axées sur la prévention de l’assèchement des nids de frai, ont été élargies pour inclure la protection des stades de viepost-émergence. Nous avons utilisé des analyses stock–recrutement pour cerner les changements a la productivité en eau doucede la population qui ont eu lieu sur une période de 30 ans et qui ont coïncidé avec les modifications des activités d’exploitationdu barrage. Nous avons observé une augmentation de 217 % de la productivité correspondant aux contraintes mises en place pourprévenir l’assèchement des nids et une augmentation supplémentaire de 130 % coïncidant avec la mise en œuvre des contraintesvisant a limiter l’échouage et le piégeage des juvéniles. L’information obtenue dans le cadre de l’étude pourrait servir a orienterles efforts visant a atténuer les effets de l’exploitation d’autres barrages hydroélectriques sur les populations de poissons en aval.[Traduit par la Rédaction]

IntroductionOperation of hydroelectric dams changes the hydrologic char-

acteristics of the river environment, which affects the availabilityand quality of fish habitat (Pringle et al. 2000; Bunn and Arthington2002). Hydropeaking and load following operation modes, wherebypulses of water are released in rapid response to meet changes inelectrical demand, have the potential to alter the quantity, qual-ity, and accessibility of downstream habitats (Clarke et al. 2008;Fisk et al. 2013). Depending on the timing, frequency, duration,and magnitude, discharge fluctuations can have adverse effectson stream fishes (Young et al. 2011).

Streamflow alterations resulting from the operation of hy-droelectric dams have contributed to the decline of riverine fishpopulations in many regions of the world (Dudgeon 2000; Pringleet al. 2000; Bunn and Arthington 2002; Dauble et al. 2003). Dis-

charge fluctuations during the period of spawning may causeadults to abandon nests or alter spawning site selection (Chapmanet al. 1986; Auer 1996; Zhong and Power 1996; Geist et al. 2008).Fluctuations in discharge that occur shortly after the spawningperiod can dewater nests, resulting in mortality of eggs and larvalfish (Becker et al. 1982; McMichael et al. 2005; Fisk et al. 2013).Discharge fluctuations that occur during the early rearing stagecan strand juveniles along changing channel margins or entrapthem in isolated pockets of water (Cushman 1985; Halleraker et al.2003; Connor and Pflug 2004; Nagrodski et al. 2012). Repeated,rapid fluctuations in discharge may also negatively affect juvenilefishes indirectly by altering the density, biomass, and diversityof their food supply (Cushman 1985; Gislason 1985; Bunn andArthington 2002).

Received 22 May 2013. Accepted 10 December 2013.

Paper handled by Associate Editor Ray Hilborn.

R.A. Harnish and G.A. McMichael. Pacific Northwest National Laboratory, 902 Battelle Boulevard, P.O. Box 999, MSIN K6-85, Richland, WA 99352,USA.R. Sharma.* Columbia River Inter-Tribal Fish Commission, 729 NE Oregon Street, Suite 200, Portland, OR 97232, USA.R.B. Langshaw and T.N. Pearsons. Public Utility District No. 2 of Grant County, 30 C Street SW, Ephrata, WA 98823, USA.Corresponding author: Ryan A. Harnish (e-mail: [email protected]).*Present address: Indian Ocean Tuna Commission, P.O. Box 1011, Le Chantier Mall, Victoria, Seychelles.

602

Can. J. Fish. Aquat. Sci. 71: 602–615 (2014) dx.doi.org/10.1139/cjfas-2013-0276 Published at www.nrcresearchpress.com/cjfas on 25 February 2014.

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y B

AT

TE

LL

E P

AC

IFIC

NW

NA

TL

LA

B o

n 03

/31/

14Fo

r pe

rson

al u

se o

nly.

Reductions in spawning success, survival, and growth have thepotential to reduce the productivity of populations. However, theavailability of data sets that can be used to quantify the effectsof hydroelectric operations on the survival and productivity ofdownstream fish populations is limited. Therefore, few studieshave examined the effects of hydroelectric dam operations at thepopulation-level scale. The Hanford Reach fall Chinook salmon(Oncorhynchus tshawytscha) population that inhabits the ColumbiaRiver downstream from Priest Rapids Dam represented a uniqueopportunity for this type of analysis. Reliable escapement and recruitabundances could be estimated over a 30-year period, covering threedistinct dam operation regimes, each with adequate replication.

The objectives of our analyses were to (i) determine if modi-fications to Priest Rapids Dam operations altered the produc-tivity of the Hanford Reach fall Chinook salmon populationand (ii) identify specific dam operation variables that may haveaffected egg-to-presmolt survival of fall Chinook salmon in theHanford Reach.

Site descriptionHydroelectric development has inundated much of the spawn-

ing habitat historically used by fall Chinook salmon in the main-stem Columbia River (Dauble and Watson 1997). The HanfordReach, an 80 km stretch of river located between Priest RapidsDam and the head of Lake Wallula (i.e., McNary Reservoir) at thetown of Richland, Washington (Fig. 1), is the last segment of theColumbia River above Bonneville Dam that has not been dammed,dredged, or channelized (Whidden 1996) and is available to ana-

dromous fish. As such, the Hanford Reach contains the only re-maining substantial mainstem spawning area for fall Chinooksalmon in the Columbia River (Bauersfeld 1978; Chapman et al.1986; Dauble and Watson 1997).

Priest Rapids Dam marks the upstream boundary of the HanfordReach and is part of a seven-dam hydroelectric complex on themid-Columbia River that is typically operated under a load-following mode to meet electrical demand, which can result insubstantial hourly and daily fluctuations in discharge (Fig. 2;Langshaw and Pearsons 2010). Without operational constraints,these fluctuations could lead to widespread dewatering of fallChinook salmon redds and stranding or entrapment of rearingjuveniles in the Hanford Reach.

Hydroelectric operation of Priest Rapids Dam during the mid-1970s resulted in dewatered redds and high mortalities of alevinsand emergent fry on Vernita Bar (Chapman et al. 1983; Becker andNeitzel 1985). Concern over these losses led to several years ofstudy to determine if different minimum discharges affectedspawning success or embryo survival (Bauersfeld 1978; Chapmanet al. 1983, 1986). Based on results from these studies, the PublicUtility District No. 2 of Grant County (Grant PUD), which owns andoperates the Priest Rapids Project (Priest Rapids and Wanapumdams), implemented the interim Vernita Bar Settlement Agree-ment (VBSA) in 1984.

The VBSA included management constraints on discharges tominimize fall Chinook salmon spawning at higher elevations.Under the agreement, discharge is manipulated during the fall

Fig. 1. Map of the Hanford Reach of the Columbia River, which extends upstream from the town of Richland, Washington, to Priest RapidsDam. Escapement to the Hanford Reach is estimated by subtracting fish ladder counts at Priest Rapids (adjusted for fallback), Ice Harbor, andProsser dams; Hanford Reach and Yakima River sport and tribal harvest; returns to Priest Rapids and Ringold Springs hatcheries; andspawning escapement in the Priest Rapids Hatchery outflow channel and the lower Yakima River (downstream of Prosser Dam) from theladder count at McNary Dam. Little to no spawning occurs between McNary Dam and Richland or in the Snake River downstream fromIce Harbor Dam.

Harnish et al. 603

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y B

AT

TE

LL

E P

AC

IFIC

NW

NA

TL

LA

B o

n 03

/31/

14Fo

r pe

rson

al u

se o

nly.

Chinook salmon spawning period to limit Chinook salmon reddsite selection (which was thought to occur mainly during daylighthours) to lower elevations by reversing the normal load-followingpattern, providing low discharges during the day and higher dis-charges at night. Spawning activities are monitored to establishthe protection-level discharge, which is the minimum dischargerequired to keep redds fully watered through the incubation pe-riod (Fig. 2). The protection-level minimum discharge ceases afteremergence is estimated to be completed (from accumulated ther-mal units).

Although the VBSA was an important step in protecting fallChinook salmon redds from dewatering, adaptive managementled to development of the Hanford Reach Fall Chinook Protection

Program Agreement (HRFCPPA) to provide additional protectionsduring other life stages (Langshaw and Pearsons 2010). The in-terim HRFCPPA was enacted in 1999 and provided constraints bylimiting the magnitude of discharge fluctuations from Priest RapidsDam to limit stranding and entrapment of juveniles during theperiod of emergence and early rearing (Fig. 2).

Methods

River environmentWe characterized the river environment in the Hanford Reach

as it related to the egg-to-presmolt survival of fall Chinook salmonby quantifying environmental and dam operation variables for

Fig. 2. Hydrographs depicting the discharge (solid line) from Priest Rapids Dam near Locke Island in the Hanford Reach of the ColumbiaRiver during a representative year from the pre-Vernita Bar Settlement Agreement (Pre-VBSA; brood year (BY) 1975–1983), VBSA(BY 1984–1998), and Hanford Reach Fall Chinook Protection Program Agreement (HRFCPPA; BY 1999–2004) periods. Several variables used inthe analyses to identify dam operations that affected egg-to-presmolt survival are depicted in each hydrograph. The dashed line represents themean discharge during the spawning period (25 October – 25 November; SpAvgQ). The dotted line displays the minimum discharge during thespawning period (SpMinQ). The dash-dot-dot-dashed line represents the minimum discharge during the posthatch incubation period(PHMinQ), which is roughly equivalent to the protection level discharge of the VBSA and HRFCPPA. Several other dam operation variables,such as the ratio of PHMinQ to SpMinQ (PHMinQ:SpMinQ) and the difference between SpAvgQ and PHMinQ (SpAvgQ – PHMinQ) werecalculated from these variables. Dates are formatted as month/day/year.

604 Can. J. Fish. Aquat. Sci. Vol. 71, 2014

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y B

AT

TE

LL

E P

AC

IFIC

NW

NA

TL

LA

B o

n 03

/31/

14Fo

r pe

rson

al u

se o

nly.

each year from 1975 through 2005, a period that covered 30 broods offall Chinook salmon. We identified ten dam operation variablesthat were hypothesized to have the greatest influence on the egg-to-presmolt survival of fall Chinook salmon in the Hanford Reach(Table 1). The list included variables from each of the life stagesexpected to be directly affected by Priest Rapids Dam operationsin the Hanford Reach (spawning, incubation, and rearing).

The timing and duration of Hanford Reach fall Chinook salmonlife stages were estimated from spawning surveys, which identi-fied the initiation of spawning and the subsequent accumulationof degree-days. Results from more than 60 years of aerial reddsurveys suggest the initiation of fall Chinook salmon spawningoccurs around 24 October in the Hanford Reach and lasts untilabout the third week in November (Dauble and Watson 1997).Therefore, the period from 25 October through 25 November wasselected as the spawning period for each year in our analyses. Thetiming of incubation and nearshore rearing life stages wereestimated for each year by accumulating degree-days from thefirst and last day of spawning. A degree-day is defined as the meanwater temperature (°C) for a given day. Water temperatures of themid-Hanford Reach (near Locke Island; Fig. 1) were simulated foryears 1975–2005 using MASS1 (Modular Aquatic Simulation Sys-tem 1D), a one-dimensional, unsteady hydrodynamic model forriver systems (Tiffan et al. 2002; McMichael et al. 2005; Richmondand Perkins 2009). MASS1 utilizes cross-sectional-averaged valuesof hydraulics and temperature to provide single temperature anddischarge values at each designated cross-section. Cross-sectionswere generated every 0.40 km within the Hanford Reach using abathymetric surface of 1 m resolution (Coleman et al. 2010). Inflowboundary conditions for MASS1 included daily discharge fromUS Geological Survey gauge 12472800 on the Columbia Riverdownstream of Priest Rapids Dam; daily scroll case temperaturefor Priest Rapids and Rock Island dams; and hourly meteorologi-cal data (i.e., ambient air temperature, dew point temperature,wind speed, atmospheric pressure, and solar radiation) collectedfrom the Hanford Meteorological Station. Water temperatures

and discharges were simulated hourly, from which daily meanswere calculated.

The first day of incubation was estimated to be equivalent to thefirst day of spawning (i.e., 25 October). The first day of the post-hatch period was estimated as 530 degree-days from the first dayof spawning (Boyd et al. 2010). The last day of incubation (lastemergence) was estimated as 1000 degree-days from the last day ofspawning (Hoffarth et al. 2003; McMichael et al. 2005; Boyd et al.2010). The first day of rearing (first emergence) was estimated byaccumulating degree-days from the first day of spawning until1000 degree-days had been accumulated. The last day of nearshorerearing was estimated to occur 1400 degree-days from the last dayof spawning (Hoffarth et al. 2003). Once the timing and durationof life stages were estimated for each brood year (BY), dam opera-tion variables were calculated for each life stage.

Variables that considered available spawning habitat, the rela-tionship between spawning and incubation discharges, the extentand frequency of discharge fluctuations during rearing, and thevariability in incubation and rearing discharges were included inthe list of variables analyzed for their effect on egg-to-presmoltsurvival (Table 1). The mean spawning discharge (SpAvgQ) wasincluded as an indicator of available spawning habitat (Fig. 2).That is, higher SpAvgQ values would suggest fish were able tospawn at higher riverbed elevations in the Hanford Reach.

Five variables were calculated as indicators of redd dewateringpotential. Two variables related the minimum posthatch incuba-tion discharge to the spawning discharge. These included theratio of the minimum posthatch incubation discharge to theminimum spawning discharge (PHMinQ:SpMinQ) and the differ-ence between the mean spawning discharge and the minimumposthatch incubation discharge (SpAvgQ – PHMinQ). Small valuesof PHMinQ:SpMinQ and large values of SpAvgQ – PHMinQ indi-cate minimum posthatch incubation discharges were low relativeto spawning discharges and the potential for redd dewateringexisted (Fig. 2). Three additional incubation period variables wereincluded: the minimum posthatch incubation discharge (PHMinQ),

Table 1. Dam operation variables used to characterize the river environment in the Hanford Reach of the Columbia River for the three PriestRapids Dam operation periods (Pre-Vernita Bar Settlement Agreement (Pre-VBSA), Vernita Bar Settlement Agreement (VBSA), and Hanford ReachFall Chinook Protection Program Agreement (HRFCPPA)).

Variable code Variable descriptionShapiro–Wilk (P)

Levene(P)

Pre-VBSA VBSA HRFCPPA

ANOVA,Kruskal–Wallis (P)

PHCVQ Coefficient of variation of hourly discharge(Q, m3·s−1) during the posthatch incubationperiod

0.006* NA 0.26 0.25 0.26 0.665

PHMinQ Minimum posthatch incubation Q 0.254 0.182 1 188 1 793 1 812 <0.001*PHMinQ:SpMinQ Ratio of minimum posthatch incubation

Q to minimum spawning Q0.115 0.645 1.02 1.46 1.44 <0.001*

PHArea Cumulative area (ha) dewatered during theposthatch incubation period

0.332 0.868 54 805 44 492 45 656 0.127

Rear850 Proportion of nearshore rearing days thatmax. Q − min. Q was >850 m3·s−1

0.195 0.918 0.71 0.66 0.37 <0.001*

Rear1416 Proportion of nearshore rearing days thatmax. Q − min. Q was >1416 m3·s−1

0.187 0.267 0.32 0.33 0.11 0.001*

RearCVQ Coefficient of variation of hourly Qduring the period of nearshore rearing

0.041* NA 0.21 0.24 0.27 0.508

RearArea Cumulative area (ha) dewatered during theperiod of nearshore rearing

0.844 0.912 26 368 23 119 15 315 0.005*

SpAvgQ Mean Q during the spawning period 0.049* NA 2 719 2 792 2 783 0.534SpAvgQ–PHMinQ Difference between the mean spawning

Q and the minimum posthatchincubation Q

0.376 0.057 1 426 914 870 0.001*

Note: Results of the Shapiro–Wilk normality test and Levene median test for equal variance are displayed. Variables that failed these tests (P < 0.05) are indicatedby an asterisk (*), and median values were used to characterize the variable for each operation regime. Variables that were normally distributed with equal varianceare characterized by mean values for each operation regime. Results of analysis of variance (ANOVA) and Kruskal–Wallis tests are displayed. Variables that differedsignificantly (P < 0.05) among operation periods are indicated by an asterisk (*).

Harnish et al. 605

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y B

AT

TE

LL

E P

AC

IFIC

NW

NA

TL

LA

B o

n 03

/31/

14Fo

r pe

rson

al u

se o

nly.

the coefficient of variation (CV) of the posthatch incubation dis-charge (PHCVQ), and the cumulative area dewatered during post-hatch incubation (PHArea). The variable PHMinQ was included asan indicator of the effectiveness of the protection-level discharge,as implemented under the VBSA (Fig. 2). That is, if protection-leveldischarges were maintained above the level at which most reddswere constructed, we would expect high PHMinQ values. The vari-able PHCVQ characterized the variability in discharge during theposthatch incubation period. The variable PHArea measured thearea that was dewatered during incubation. To calculate PHArea,we first estimated the wetted area of 360 evenly spaced cross-sections of the Hanford Reach for each hour of the incubationperiod using MASS2, a two-dimensional hydrodynamic and trans-port model (Perkins et al. 2004; Perkins and Richmond 2007). Thedifference in wetted area was summed for each cross-section onlywhen a reduction in wetted area was observed from one hour tothe next. These values were then summed across all cross-sectionsto calculate PHArea. We assumed high values of PHArea wouldindicate a higher probability that redds were dewatered.

The final four variables (Rear850, Rear1416, RearCVQ, andRearArea) were included as indicators of flow fluctuations dur-ing the period of nearshore rearing that may have led to strand-ing and entrapment of juveniles. Two similar variables, Rear850and Rear1416, measured the proportion of rearing days in whichthe difference between the maximum and minimum dischargeswas >850 and >1416 m3·s−1, respectively. The variable RearCVQwas a measure of the variability of discharge during the rearingperiod and was calculated as the CV of discharge during the rear-ing period. The final variable, RearArea, measured the cumulativearea that was dewatered during the period of nearshore rearingand was calculated as described above for PHArea.

Dam operation variables were compared among three time pe-riods that coincided with the availability of data to estimate eggescapement and presmolt abundance, which spans from BY 1975through BY 2004, and changes to operations of Priest RapidsDam: (i) pre-VBSA (BY 1975–1983), (ii) VBSA (BY 1984–1998), and(iii) HRFCPPA (BY 1999–2004). The Shapiro–Wilk normality testwas used to determine whether or not the variables were nor-mally distributed (� = 0.05). Variables were also tested for equalvariance using the Levene median test (� = 0.05). If normally distrib-uted with equal variance, one-way analysis of variance (ANOVA) wasused to test for differences in variables among periods, and aStudent’s t test was used to make pairwise comparisons. For vari-ables that were not normally distributed or displayed unequalvariance, the nonparametric Kruskal–Wallis one-way ANOVA onranks was used to test for differences among periods. The signifi-cance level of pairwise comparisons was adjusted using the Bon-ferroni correction, whereby the critical � value of the wholefamily of tests (� = 0.05) was divided by the number of compari-sons.

Stock–recruit modeling

Stock estimationThe number of eggs in female fall Chinook salmon in the Hanford

Reach was estimated for each BY 1975–2004 by multiplying age-specific female escapement estimates by age-specific fecundityvalues. Hanford Reach fall Chinook salmon escapement estimateswere obtained from the Washington Department of Fish andWildlife (WDFW; Hoffarth 2010). Escapement of adult HanfordReach fall Chinook salmon spawners is based on the “subtrac-tion method” using dam fish ladder counts, sport and tribalharvest, and tributary spawning abundance. Using this method,ladder counts at Priest Rapids (adjusted for fallback), Ice Harbor,and Prosser dams; Hanford Reach and Yakima River sport andtribal harvest; returns to Priest Rapids and Ringold Springs hatch-eries; and spawning escapement in the Priest Rapids Hatcheryoutflow channel and the lower Yakima River (downstream of

Prosser Dam) are subtracted from the ladder count at McNaryDam (Fig. 1) to estimate adult fall Chinook salmon escapement tothe Hanford Reach. Escapement has been estimated using thismethod to successfully manage the population for over 30 years.

The number of adults of each age was estimated for each BY bymultiplying adult escapement estimates by the proportion of fishof each age. The proportion of each age was estimated from scalescollected during carcass surveys, which have been conducted byWDFW throughout the Hanford Reach annually since 1979. Foryears in which Hanford Reach fall Chinook salmon age data werenot available (BY 1975–1980, BY 1995), age structure data for theentire run of Columbia River upriver bright fall Chinook salmonwere used (Harlan et al. 1998).

Next, the number of females of each age was estimated for eachBY 1986–2004 by multiplying the age-specific escapement esti-mates by the proportion of females from each age group. Becausethere is typically a positive bias toward females associated withcarcass recovery (Hoffarth 2010), the proportion of females fromeach age group was estimated from data collected during WDFWHanford Reach sport fishery creel surveys. For BY 1975–1985, priorto the onset of WDFW creel surveys in the Hanford Reach, meanfemale proportions from BY 1986–2004 were used. Finally, thenumber of eggs in females was estimated by multiplying the num-ber of females of each age by the age-specific mean fecundity of441 hatchery fall Chinook salmon adult females that were sampledfor fecundity upon their return in 2011 to Priest Rapids Hatchery(Hoffarth and Pearsons 2012). The total escapement of eggs in fe-males was estimated for each BY by adding the age-specific esti-mates.

Recruit estimationNatural-origin fall Chinook salmon presmolts have been cap-

tured from multiple locations in the Hanford Reach using stickand beach seines each year since BY 1986. Captured presmolts thatmeasured 47–80 mm fork length were implanted with a codedwire tag (CWT) and marked with the removal of the adipose fin(Fryer 2005). From BY 1986 through BY 2004, the number of natural-origin presmolts implanted with CWTs ranged from 92 262 to258 726 fish, with a target goal of implanting 200 000 fish eachyear. Implanted fish serve as an indicator stock to the PacificSalmon Commission Chinook Technical Committee (PSC CTC) forestimating cohort-specific exploitation rates from recoveries ofCWTs in fisheries along the northwest coast of North America(Bernard and Clark 1996).

Seining effort data were not readily available to estimate catchper unit effort (CPUE) and an associated index of presmolt abun-dance. Additionally, estimates of seining CPUE may be biased ow-ing to annual variations in efficiency caused by environmentalconditions (e.g., discharge). Therefore, the number of presmoltsproduced each BY was estimated using a virtual population tech-nique (cohort reconstruction), similar to the approach used byCoronado and Hilborn (1998) and Sharma et al. (2013). The generalapproach of the cohort reconstruction began with WDFW escape-ment estimates, then worked backward in time to account forinterdam loss, in-river and ocean harvest, and natural ocean mor-tality to estimate the number of age-2 ocean recruits produced.This estimate was then divided by a CWT-based estimate of sur-vival from tagging to age-2 to estimate the number of presmolts inthe Hanford Reach at the time of tagging.

Specifically, the first step of the cohort reconstruction requiredestimating the number of hatchery fall Chinook salmon thatspawned in the Hanford Reach each year, which was then sub-tracted from the escapement estimates. The proportion of hatch-ery spawners was estimated from recoveries of CWT hatchery fishduring carcass surveys for run years 1979–2009. Mean hatcheryproportions from 1979 through 2009 were used for run years1975–1978 when no carcass surveys were conducted in the HanfordReach. After hatchery spawners were subtracted from the escape-

606 Can. J. Fish. Aquat. Sci. Vol. 71, 2014

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y B

AT

TE

LL

E P

AC

IFIC

NW

NA

TL

LA

B o

n 03

/31/

14Fo

r pe

rson

al u

se o

nly.

ment estimate, age-specific escapements were adjusted by adult-and jack-specific Bonneville Dam to McNary Dam conversion rates(C. LeFleur, US vs. Oregon Technical Advisory Committee, Vancou-ver, Washington, private communication, 2012) to account forinterdam loss, which is any mortality not accounted for duringupstream migration.

Next, the age-specific run to the mouth of the Columbia Riverwas constructed by dividing the interdam loss-adjusted estimatesby the in-river harvest rates obtained from CWT data (R. Sharma,PSC CTC, unpublished data). Priest Rapids Hatchery CWT groupswere used to estimate in-river harvest rates for years during which

no wild Hanford Reach CWT groups existed (BY 1975–1985) orwhen recoveries of wild Hanford Reach CWT fish were low.

Ocean recruits were then calculated for each age using thecohort analysis algorithm of the PSC CTC, ocean harvest ratesobtained from CWT recoveries (R. Sharma, PSC CTC, unpublisheddata), and conditional ocean survival rates currently used by thePSC CTC (S2 = 0.60, S3 = 0.70, S4 = 0.80, and S5 = 0.90). Priest RapidsHatchery CWT groups were used to estimate ocean exploitationrates for years during which no wild Hanford Reach CWT groupsexisted. For an individual CWT group, we began with the oldestage and worked back through successive ages until we reachedocean age-2. The following equations outline this approach:

(1) A5�(t � 5, t � 6) �({[E5(t � 5) � E6(t � 6)]/[I5(t � 5)]}/[1 � r5(t � 5)])/[1 � u5(t � 5)]

S5

(2) A4(t � 4) �[A5�(t � 5, t � 6)] � {({E4(t � 4)/[I4(t � 4)]}/[1 � r4(t � 4)])/[1 � u4(t � 4)]}

S4

(3) A3(t � 3) �[A4(t � 4)] � {({E3(t � 3)/[I3(t � 3)]}/[1 � r3(t � 3)])/[1 � u3(t � 3)]}

S3

(4) A2(t � 2) �[A3(t � 3)] � {({E2(t � 2)/[I2(t � 2)]}/[1 � r2(t � 2)])/[1 � u2(t � 2)]}

S2

where Ei(t) is the age-specific Hanford Reach natural-origin spawningescapement in year i = 2, 3, 4, 5, 6; Ii(t) is the interdam conversion ratein year t, calculated separately for jacks and adults as the number ofHanford Reach-destined fall Chinook salmon counted at McNaryDam as compared with Bonneville Dam; Si is the conditional oceannatural survival rate from age i to i + 1, given fish are not captured inocean fisheries at age i and do not mature at age i; ri(t) is the in-riverfishery exploitation rate at age i in year t; ui(t) is the ocean fisheryexploitation rate at age i in year t; and Ai(t) is the numbers of fish aliveat ocean age i in year t, prior to fishing or maturation.

Finally, the number of presmolts associated with each BY wasestimated by dividing the number of fish estimated to have reachedocean age-2 by a CWT-based estimate of survival from presmolt toocean age-2 (Sharma et al. 2013):

(5) P(t) �A2(t � 2)

S0(t)

where S0�t� is the survival from release to ocean age-2, prior tofishing or maturation, for CWT fish released from brood year t.

Fig. 3. Presmolt-to-age-2 survival for Priest Rapids Hatchery (solid circles and solid line) and wild Hanford Reach (open circles and dashed line)juveniles implanted with coded-wire tags, brood years 1986–2004.

Harnish et al. 607

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y B

AT

TE

LL

E P

AC

IFIC

NW

NA

TL

LA

B o

n 03

/31/

14Fo

r pe

rson

al u

se o

nly.

Because of the relatively strong correlation in survival to age-2of CWT Hanford Reach presmolts and Priest Rapids Hatchery smolts(Fig. 3), survival rates from presmolt to ocean age-2 of Priest RapidsHatchery CWT groups were used for BYs in which no wild HanfordReach presmolts were implanted with CWTs (BY 1975–1985).

Model fittingMultiple models, including the Ricker, Beverton–Holt, and

Smooth Hockey Stick, were fit to the presmolt/egg data. The linearform of the Ricker model — that is,

(6) ln(R/S) � ln� � �S � �

where R is the number of presmolts, S is the egg escapement, � isthe estimate of density-independent productivity, and 1/� is theestimate of spawners associated with maximum recruitment —was chosen as the model with which to proceed because it pro-vided the best fit to the data, it is flexible in fitting a wide varietyof data sets, management or policy parameters are readily calcu-lated from fitted Ricker model parameters (Ricker 1975), and be-cause it has been used previously for stock–recruit analyses of theColumbia River upriver bright fall Chinook salmon population(Bernard and Clark 1999; Peters et al. 1999; Langness and Reidinger2003).

Effect of operations on productivityWe tested for a change in density-independent mortality across

periods (pre-VBSA, VBSA, and HRFCPPA) to determine whetheralterations to dam operations affected the productivity of fallChinook salmon in the Hanford Reach. A temporal change indensity-independent mortality, such as that imposed by dewater-ing of redds or stranding and entrapment of juveniles would causenonstationary behavior in the recruitment function (Schaller et al.1999; Schaller and Petrosky 2007). Because productivity is measuredas the intercept, or � parameter, of the Ricker model, we wouldexpect this change to be reflected primarily in the � parameter ofthe regression (Schaller and Petrosky 2007). To account for non-stationarity in the recruitment functions, analysis of covariance(ANCOVA; JMP version 8.0, SAS Institute, Cary, North Carolina)was used to examine the differences in the intercepts (Ricker �) ofthe relationship of ln(R/S) versus S in the equation

(7) ln�Rij

Sij� � ln(�i � �) � �(Sij � S..) � �ij

where Rij is the number of presmolts, �i is the period effect, � is theoverall intercept, � is the overall slope, Sij is the egg escapement,S is the mean egg escapement for all 30 BYs, �ij is the normallydistributed residual, i is the period, and j is the BY.

First, the homogeneity of slopes was tested for significant inter-action between the treatment (period) and the covariate (egg es-capement) using a Student’s t test. An ANCOVA was then run to

estimate the period effect on ln(R/S), taking into account spawn-ing level. The measure of productivity by period was estimated as�i + � from the ANCOVA results, which is equivalent to the Ricker� parameter by period (assuming a common slope for each period;Schaller and Petrosky 2007). Pairwise comparisons were made

Fig. 4. Frequency plots of (A) the ratio of the minimum posthatchincubation discharge to the minimum spawning discharge(PHMinQ:SpawnMinQ), (B) the difference between the meanspawning discharge and the minimum posthatch incubationdischarge (SpAvgQ – PHMinQ), and (C) the minimum posthatchincubation discharge (PHMinQ) for fall Chinook salmon life stagesin the Hanford Reach of the Columbia River during the three PriestRapids Dam operation periods (Pre-Vernita Bar SettlementAgreement (Pre-VBSA), VBSA, and Hanford Reach Fall ChinookProtection Program Agreement (HRFCPPA)). The top and bottom ofeach box represent the 75th and 25th percentiles, respectively; thesolid line within the box represents the median; and the barsindicate the 90th and 10th percentiles.

608 Can. J. Fish. Aquat. Sci. Vol. 71, 2014

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y B

AT

TE

LL

E P

AC

IFIC

NW

NA

TL

LA

B o

n 03

/31/

14Fo

r pe

rson

al u

se o

nly.

using the Tukey–Kramer honestly significant difference (HSD) testif a significant difference in productivity was detected amongperiods (� = 0.05).

We also attempted to identify specific dam operation variablesthat affected egg-to-presmolt survival. Decision tree analyses wereused to identify which, if any, of the 10 selected variables werecorrelated with estimates of fall Chinook salmon egg-to-presmoltsurvival in the Hanford Reach. Regression trees, such as thosegenerated by decision tree analyses, are flexible and robust, ableto deal with nonlinear relationships and high-order interactions,yet easy to understand and interpret (De’ath and Fabricius 2000).We used the JMP Partition platform to perform our decision treeanalyses. Within this platform, the data were partitioned at eachsplit into two mutually exclusive groups, each of which was ashomogeneous as possible with regard to response (egg-to-presmoltsurvival estimates) and predictor (dam operation variable) values.The splitting procedure was then applied to each group sepa-rately. Partitioning was done according to a splitting “cut” valuefor the predictor variable. Splitting was based on maximizingthe LogWorth significance value, which is the negative log of theadjusted P value, for each split candidate (Sall 2002). We set theminimum group size to five and the adjusted P value to 0.05. Thatis, no fewer than five BYs could be split from the data to form ahomogeneous group, and the adjusted P value that resulted froma split had to be less than 0.05 to be significant. The objective wasto partition the response into homogeneous groups while alsokeeping the tree relatively small (De’ath and Fabricius 2000).

Linear and nonlinear bivariate relationships between each ofthe 10 variables and presmolt/egg estimates were also explored toidentify additional variables that may have affected egg-to-presmoltsurvival of fall Chinook salmon in the Hanford Reach. A falsediscovery rate of 0.10 was used to correct for multiple compari-sons, and significance of each linear relationship was determinedusing the methods of Benjamini and Hochberg (1995).

Results

Changes to the river environmentAlterations to Priest Rapids Dam operations resulted in substan-

tial changes to the Hanford Reach river environment during mul-tiple fall Chinook salmon life stages. We identified three variablesthat changed significantly as a result of alterations to incubationstage discharges as first implemented by the VBSA. The minimumposthatch incubation discharge (PHMinQ) increased from a meanvalue of 1188 m3·s−1 during the pre-VBSA period to 1793 and1812 m3·s−1 during the VBSA and HRFCPPA periods, respectively(Fig. 4). This difference was statistically significant (P < 0.001;Table 1). The ratio of minimum posthatch incubation discharge tothe minimum spawning discharge (PHMinQ:SpMinQ) was signifi-cantly lower during the pre-VBSA period (mean = 1.02) comparedwith all other periods (P < 0.001; Table 1; Fig. 4). In addition, thedifference between the mean spawning discharge and the mini-mum posthatch incubation discharge (SpAvgQ – PHMinQ) wassignificantly higher, on average, during the pre-VBSA period

(1426 m3·s−1) compared with the other two periods (P < 0.001;Table 1; Fig. 4).

We also identified three variables that changed significantly asa result of constraints placed on discharge fluctuations during the

Fig. 5. Frequency plots of (A) the proportion of fall Chinook salmonjuvenile rearing days in which the difference between the dailymaximum and minimum discharge values was greater than850 m3·s−1, (B) the proportion of fall Chinook salmon juvenilerearing days in which the difference between the daily maximumand minimum discharge values was greater than 1416 m3·s−1, and(C) the cumulative area dewatered during the period of nearshorerearing for the three Priest Rapids Dam operation periods (Pre-Vernita Bar Settlement Agreement (VBSA), VBSA, and Hanford ReachFall Chinook Protection Program Agreement (HRFCPPA)). The topand bottom of each box represent the 75th and 25th percentiles,respectively; the solid line within the box represents the median;and the bars indicate the 90th and 10th percentiles.

Harnish et al. 609

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y B

AT

TE

LL

E P

AC

IFIC

NW

NA

TL

LA

B o

n 03

/31/

14Fo

r pe

rson

al u

se o

nly.

period of nearshore rearing as implemented by the HRFCPPA. Theproportion of nearshore rearing days in which the difference be-tween the maximum and minimum discharges was greater than850 m3·s−1 (Rear850) and 1416 m3·s−1 (Rear1416) was significantlylower for the HRFCPPA period compared with the other periods(Table 1; Fig. 5). Additionally, the cumulative area dewatered dur-ing the rearing period (RearArea) was significantly lower duringthe HRFCPPA than it was during the pre-VBSA or VBSA periods(Table 1; Fig. 5).

Stock–recruit estimationThe pre-VBSA period was characterized by relatively low escape-

ment levels that averaged around 51 million eggs (Table 2). Thisperiod was followed by an increase in egg escapement during theearly portion of the VBSA period before declining to relatively lowlevels (50–100 million eggs) during the early 1990s. Escapementgenerally remained within the range of 50–100 million eggsthrough the latter portion of the VBSA and the first few years ofthe HRFCPPA before exceeding 150 million eggs each year from2002 through 2004.

We observed a positive trend over time in Hanford Reach fallChinook salmon presmolt abundance, which ranged from 2 millionto 78 million presmolts over the 30-year period from BY 1975–2004(Table 2). The lowest presmolt abundances were observed duringthe pre-VBSA period, which averaged 14 million presmolts. Themean presmolt production during the VBSA and HRFCPPA peri-ods was 39 million and 52 million, respectively. From the esti-mates of egg escapement and presmolt abundance, we observedan increase in the mean egg-to-presmolt survival probability from0.30 during the pre-VBSA period to 0.36 during the VBSA period to0.42 during the HRFCPPA period.

Changes in productivityResults from the ANCOVA indicated there was a significant dif-

ference in presmolt/egg productivity among periods (P = 0.01;Fig. 6; Table 3). The resulting Ricker � value equaled 0.29 for thepre-VBSA period, 0.63 for the VBSA period, and 0.82 for theHRFCPPA period. Pairwise comparisons revealed that presmolt/egg productivity was significantly higher during the VBSA andHRFCPPA periods compared with the pre-VBSA period (P < 0.05).The difference in productivity between the VBSA and HRFCPPAperiods was not significantly different (P = 0.58).

The slope of the Ricker model did not differ significantly amongperiods (P = 0.69; Table 3). A general density-dependent (i.e., neg-ative) trend was observed in the presmolt/egg stock–recruit data(Fig. 6). Although the density-dependent relationship was clearlypresent in the VBSA and HRFCPPA data, the relationship was weakfor the pre-VBSA period, which was characterized by high variabil-ity in egg-to-presmolt survival at low escapements and no data athigh escapements (Fig. 6). Egg-to-presmolt survival estimates forthree of the pre-VBSA BYs (1980–1982) follow the density-dependentrelationship observed for the other two time periods. However,the remaining six pre-VBSA BYs (1975–1979, 1983) deviated fromthis relationship, displaying below-average egg-to-presmolt sur-vival and density-independent mortality. The difference in egg-to-presmolt survival among periods was most notable at eggescapements less than 101 million eggs. Therefore, only the 20 broodyears with escapements less than 101 million eggs were in-cluded in the decision tree and bivariate regression analyses toidentify dam operation variables that may have resulted inbelow-average egg-to-presmolt survival and density-independentmortality.

The regression tree model produced PHMinQ:SpMinQ as thevariable that explained the greatest variability in egg-to-presmoltsurvival at low (<101 million eggs) escapements (Fig. 7). The sixpre-VBSA BYs that experienced below-average egg-to-presmoltsurvival (1975–1979, 1983) were grouped together in the regressiontree with PHMinQ:SpMinQ values less than 1.145 and mean egg-

to-presmolt survival probability of 0.149 presmolts/egg. PHMinQ:SpMinQ values less than or slightly greater than 1.0 indicate it islikely that some of the redds were dewatered because the mini-mum posthatch incubation discharge was less than or onlyslightly more than the minimum spawning discharge. The re-maining 14 brood years (1980–1982, 1984, 1991–1993, 1995–2001)had PHMinQ:SpMinQ values greater than or equal to 1.145 andwere associated with higher egg-to-presmolt survival probabilities( = 0.500 presmolts/egg). Further divisions of the data did notresult in significant differences between groups. Therefore, thefinal regression tree model included only PHMinQ:SpMinQ andhad a coefficient of determination (R2) of 0.669.

Two variables were correlated with egg-to-presmolt survival inthe bivariate regression analysis. The relationship between thedifference in mean spawning discharge and minimum posthatchincubation discharge (SpAvgQ – PHMinQ) and egg-to-presmoltsurvival was best fit with a simple linear regression model, whichindicated a relatively strong, negative correlation (P = 0.006; R2 =0.356; Fig. 8). Similar to the relationship modeled by the regres-sion tree, a three-parameter logistic function showed a dramaticincrease in egg-to-presmolt survival at PHMinQ:SpMinQ valuesgreater than 1.1 (P < 0.001; R2 = 0.597; Fig. 8).

Table 2. Egg escapement, presmolt abundance, and egg-to-presmoltsurvival probability estimated for Hanford Reach fall Chinook salmonduring each brood year (BY) of the pre-Vernita Bar Settlement Agree-ment (Pre-VBSA), VBSA, and Hanford Reach Fall Chinook ProtectionProgram Agreement (HRFCPPA) dam operation periods.

PeriodBroodyear

Eggescapement(millions)

Presmolts(millions)

Egg-to-presmoltsurvivalprobability

Pre-VBSA(BY 1975–1983)

1975 44.71 6.73 0.151976 34.87 5.12 0.151977 59.09 8.14 0.141978 47.54 2.27 0.051979 59.75 11.88 0.201980 53.51 26.94 0.501981 38.84 32.79 0.841982 43.21 18.21 0.421983 79.11 16.78 0.21Mean 51.18 14.32 0.30

VBSA(BY 1984–1998)

1984 96.91 42.45 0.441985 144.20 37.82 0.261986 167.86 33.00 0.201987 196.69 62.86 0.321988 191.99 39.14 0.201989 180.90 29.74 0.161990 125.58 26.37 0.211991 99.83 42.41 0.421992 72.94 45.56 0.621993 69.89 24.45 0.351994 139.43 38.89 0.281995 100.74 48.96 0.491996 81.79 43.94 0.541997 92.12 49.45 0.541998 73.18 24.36 0.33Mean 122.27 39.29 0.36

HRFCPPA(BY 1999–2004)

1999 59.11 21.00 0.362000 91.57 56.56 0.622001 93.72 49.49 0.532002 154.61 57.43 0.372003 220.41 78.03 0.352004 194.42 51.27 0.26Mean 135.64 52.30 0.42

Note: Mean egg escapement, presmolt abundance, and egg-to-presmolt sur-vival probability are also displayed for each period.

610 Can. J. Fish. Aquat. Sci. Vol. 71, 2014

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y B

AT

TE

LL

E P

AC

IFIC

NW

NA

TL

LA

B o

n 03

/31/

14Fo

r pe

rson

al u

se o

nly.

DiscussionDischarge fluctuations can have substantial effects on overall

fish production and system productivity (Hamilton and Buell1976; Cushman 1985; Clarke et al. 2008). Fluctuations from hy-droelectric dam operations have most often been associated withreductions in productivity of downstream reaches (Trotzky andGregory 1974; Cushman 1985; Burt and Mundie 1986). However, pos-itive or negative effects of discharge fluctuations on fish populationsand their habitat are dependent on the timing, duration, interval,and magnitude of the fluctuations (Cushman 1985; Bunn andArthington 2002; Korman et al. 2011a, 2011b). Therefore, manipu-lation of discharges from hydroelectric dams may be used as amanagement tool to increase the productivity of downstream fishpopulations if sufficient knowledge of habitat use and life stagephenology exists.

We observed a significant increase in freshwater productivity offall Chinook salmon in the Hanford Reach following manipula-tion of discharge fluctuations under the VBSA. The most notablechange to the Hanford Reach discharge regime that resulted fromthe VBSA was an increase in the minimum discharge during theperiod of intragravel development. This change appeared tohave the greatest effect on Hanford Reach fall Chinook salmonpresmolt/egg productivity. The observed relationships indicated egg-to-presmolt survival was highest when (i) spawning discharges weremanaged to reduce the elevation at which the majority of redds werebuilt and (ii) minimum incubation discharges were sufficient to keepredds watered during critical intragravel development stages. As

noted by Chapman et al. (1983), the elevation at which redds areconstructed in the Hanford Reach is dependent on the dischargelevel during the spawning period. Therefore, redds are likely to bedewatered when the minimum incubation discharge is low, relativeto the spawning discharge. By increasing the minimum incubationdischarge, the VBSA effectively reduced egg and alevin mortality thatresulted from redd dewatering.

We also found that constraints on discharge fluctuations duringthe period of nearshore rearing, as implemented by the HRFCPPA,further increased freshwater productivity of fall Chinook salmonin the Hanford Reach. It is likely that the HRFCPPA reducedstranding and entrapment of juvenile fall Chinook salmon bylimiting the magnitude of discharge fluctuations. However, thedifference in presmolt/egg productivity that resulted from theseadditional constraints was not statistically significant over thetime period examined. Thus, it appears the population-level effectof stranding- and entrapment-related mortality was relatively lowcompared with the effect of redd dewatering during the period ofstudy.

Management measures similar to those enacted at Priest RapidsDam have been implemented with success at other hydroelectricdams to increase salmon production. For example, constraints onmaximum discharges during spawning and minimum dischargesduring incubation, as well as limitations on the timing, rate, andmagnitude of discharge fluctuations during the rearing period,were implemented at the Skagit Hydroelectric Project in 1981(Connor and Pflug 2004). Increases in spawner abundances that

Fig. 6. Plot of log-transformed egg-to-presmolt survivial (ln[Presmolts/Egg]) ANCOVA for modeled Ricker recruitment functions using eggescapement as the covariate to compare productivity (y intercept) among pre-VBSA (BY 1975–1983; solid circles, solid line), VBSA(BY 1984–1998; open triangles, dash-dot-dot-dashed line), and HRFCPPA (BY 1999–2004; open circles, dotted line) periods for the HanfordReach fall Chinook salmon stock. The corresponding BY is displayed next to each point.

Table 3. Summary of presmolt/egg analysis of covariance (ANCOVA) period effect test results comparing productivity among pre-VBSA (BY 1975–1983), VBSA (BY 1984–1998), and HRFCPPA (BY 1999–2004) periods.

Interceptln(�1 + �),pre-VBSA

Interceptln(�2 + �),VBSA

Interceptln(�2 + �),HRFCPPA

Intercept H0:�1 = �2 = �3 (P) SE (�i) Slope �

Slope H0:� ≤ 0 (P) Adj. R2

Homogeneityof slope (P)

−1.431 −0.498 −0.244 0.001 0.234 −4.98e-9 0.024 0.341 0.692

Harnish et al. 611

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y B

AT

TE

LL

E P

AC

IFIC

NW

NA

TL

LA

B o

n 03

/31/

14Fo

r pe

rson

al u

se o

nly.

followed implementation of the discharge mitigation programwere attributed to reductions in both redd dewatering and frystranding in reaches of the Skagit River immediately downstreamfrom the dam (Connor and Pflug 2004). Similar to our results,Connor and Pflug (2004) found the difference between mean dailyspawning discharge and minimum incubation discharge in theSkagit River was negatively correlated with salmon production.The authors attributed the correlation to a reduction in redd de-watering and an increase in egg-to-fry survival as the differencebetween spawning and minimum incubation discharges becamesmaller.

The lethal effects of redd dewatering have been well documented(Becker et al. 1982, 1983; McMichael et al. 2005; Korman et al.2011a) and, as we have shown, can reduce the freshwater produc-tivity of salmon populations. However, the timing of dewateringevents is of critical importance to the survival of incubatingsalmon. Several studies have demonstrated that incubating fallChinook salmon eggs are more tolerant of dewatering than aresalmon alevins that have not yet emerged from the substrate(Becker et al. 1982, 1983; Neitzel and Becker 1985). Cleavage eggsand embryos can survive dewatering events for weeks if theyremain moist and are not subjected to extreme temperatures orpredation (Becker et al. 1982, 1983; Reiser and White 1983; Neitzeland Becker 1985). Conversely, survival of posthatch developmentphases can be reduced from relatively brief dewatering events,ranging from 1 to 10 h, and has been attributed to the formation offunctional gills (Reiser and White 1981; Becker et al. 1982).

Dewatering events that occurred in the Hanford Reach after egghatch but prior to emergence in March and April 1976 and 1977resulted in significant mortality of alevins and emergent fry(Chapman et al. 1983; Becker and Neitzel 1985). Consequently, thecorresponding broods (BY 1975 and BY 1976) experienced low egg-to-presmolt survival. Conversely, redd dewatering events that oc-curred prior to egg hatch had little effect on egg-to-presmolt survival.For example, in 1997, discharge ranged from 1600 to 4638 m3·s−1

during the majority of the spawning period (25 October to 22 Novem-ber). On 23 November, discharge was reduced to 1194 m3·s−1 andremained less than 1600 m3·s−1 for more than 8 h. Therefore, allredds constructed above the 1600 m3·s−1 discharge level prior to23 November were temporarily dewatered. However, it is likelythat little to no mortality resulted from this event because itoccurred prior to egg hatching. This conclusion is supported bythe low level of density-independent mortality and high egg-to-presmolt survival probability (0.537 presmolts/egg) we observedfor BY 1997 in the stock–recruit analysis.

Improvements in freshwater survival that occur as a result ofdischarge constraints (or some other management action) maylead to meaningful increases in adult returns and fishery yields.Results from our cohort reconstruction indicated that nearly two-thirds (65%) of the broods from 1975 through 2004 that displayedabove-average egg-to-presmolt survival also had above-averageadult/spawner production. Thus, Hanford Reach fall Chinooksalmon brood year strength appears to be largely determined byinterannual variation in freshwater survival, indicating the im-portance of the freshwater life phase to the overall productivity ofthe population.

Whereas some level of observation error is present in both es-capement and recruitment estimates, the data required to de-velop robust estimates of this uncertainty were not available.However, it is unlikely that the uncertainty of these estimateswould alter the observed relationships unless a substantive differ-ence in uncertainty existed among periods. For example, if es-capement was consistently overestimated or recruitment wasconsistently underestimated during one period, the egg-to-presmoltsurvival estimates for that period would be biased low. Conversely,underestimating escapement or overestimating recruitment wouldbias egg-to-presmolt survival estimates high. We do not believe thistype of bias was prevalent enough in the data to cause significant biaswithin any of the periods.

Fig. 7. Results of a regression tree analysis of egg-to-presmolt survival probabilities for the Hanford Reach fall Chinook salmon populationfor the 20 brood years between 1975 and 2004 that had egg escapements less than 101 million eggs. The mean survival probabilities, withstandard error in parentheses, and sample size (number of brood years) are shown for the 20 brood years in the rectangle and forhomogeneous groups in the ovals. The adjusted P value is displayed below the variable on which the split occurred (the ratio of the minimumposthatch incubation discharge to the minimum spawning discharge (PHMinQ:SpMinQ)), and the value of PHMinQ:SpMinQ that separatedgroups is displayed on each branch of the split.

612 Can. J. Fish. Aquat. Sci. Vol. 71, 2014

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y B

AT

TE

LL

E P

AC

IFIC

NW

NA

TL

LA

B o

n 03

/31/

14Fo

r pe

rson

al u

se o

nly.

Although we observed positive benefits of discharge constraintsto fall Chinook salmon, we did not explore how constraints placedon discharge fluctuations from Priest Rapids Dam may have af-fected other fish species in the Hanford Reach. For example, theeffect of the HRFCPPA on the spawning success of smallmouthbass (Micropterus dolomieu), a major predator of juvenile salmon(Fritts and Pearsons 2004, 2006), is of particular concern becausethe periods of fall Chinook salmon rearing and smallmouth bassspawning overlap in the Hanford Reach (Henderson and Foster1957; Becker et al. 1981). Several studies have demonstrated thatfluctuations in discharge can negatively affect the reproductivesuccess of smallmouth bass by flooding nests with cooler water,

depositing silt, driving away adult bass guarding nests, exposingeggs to desiccation, or stranding emerged fry (Henderson andFoster 1957; Becker et al. 1981; Lukas and Orth 1995). A study offactors that influence smallmouth bass production in the HanfordReach indicated fluctuations in discharge from hydroelectricpower generation at Priest Rapids Dam reduced productivity(Montgomery et al. 1980). Therefore, reducing flow fluctuations toprevent stranding and entrapment of juvenile salmon may havethe unintended consequence of increasing productivity of small-mouth bass in the Hanford Reach. However, the available datasuggests that any unintended benefits of flow stabilization to

Fig. 8. Bivariate regression relationships of river environment variables that were found to be correlated with Hanford Reach fall Chinooksalmon egg-to-presmolt survival estimates. Variables included the difference between mean spawning discharge and minimum posthatchincubation discharge (SpAvgQ – PHMinQ) and the ratio between the minimum posthatch incubation discharge and the minimum spawningdischarge (PHMinQ:SpMinQ).

Harnish et al. 613

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y B

AT

TE

LL

E P

AC

IFIC

NW

NA

TL

LA

B o

n 03

/31/

14Fo

r pe

rson

al u

se o

nly.

smallmouth bass have not outweighed the direct benefits toHanford Reach fall Chinook salmon.

Many hydroelectric dams are currently undergoing relicensingby the Federal Energy Regulatory Commission (Young et al. 2011).Several laws that affect hydropower relicensing require consider-ation or inclusion of conditions for the protection, mitigation, orenhancement of fish resources affected by dams. However, fewstudies have been conducted to aid resource managers, dam op-erators, or regulators in determining the direct effect of currentor proposed discharge regimes on fish populations located down-stream of hydroelectric dams. Most previous studies have focusedon changes to the area of suitable habitats over relatively shorttime periods or during specific life stages (Gibbins and Acornley2000; Geist et al. 2008; Hatten et al. 2009). Although this informa-tion is important, it does not directly relate dam operations topopulation-level productivity. Others have assessed the effect ofhydroelectric dam operations on downstream salmon popula-tions by using the abundance of returning adults as the responsemetric by which alterations to dam operations are measured(Connor and Pflug 2004). However, these results must be viewedwith caution because environmental factors that affect life stages(e.g., smolt, adult) outside of the dam’s hydraulic extent are man-ifest in the abundance of returning adults.

We examined more than 30 years of data to determine theeffects of discharge fluctuations from Priest Rapids Dam undermultiple operation regimes on the freshwater productivity of theHanford Reach fall Chinook salmon population. Our results indi-cate that altering the timing and magnitude of discharge fluctua-tions can minimize the adverse effects of operating hydroelectricdams on the productivity of downstream salmon populations. Theinformation gained from this study may be used to guide futureefforts to mitigate the effects of hydroelectric dam operations ondownstream fish populations in other areas. In addition, themethods we used proved to be an effective approach for assessingpopulation-level effects of flow management and may be used toassess future discharge management measures.

AcknowledgementsWe thank Paul Hoffarth, Jeffrey Fryer, Michelle Groesbeck, Sara

Niehus, Katie Klett, Jill Janak, Benjamin Miller, Robert Woodard,William Perkins, Henry Yuen, Cindy LeFleur, Marshall Richmond,and Michael Hughes for collecting and providing data. We alsothank Tracy Hillman, David Hankin, David Bernard, John Clark,Richard Bailey, Tom Cooney, Dennis Dauble, Richard Hinrichsen,Robert Kope, Scott McPherson, Gary Morishima, Charlie Petrosky,Howard Schaller, and Dell Simmons for reviewing the study de-sign and providing useful recommendations. Finally, we thankChris Vernon for graphics preparation and Andrea Currie andKathy Neiderhiser for editorial assistance.

ReferencesAuer, N.A. 1996. Response of spawning lake sturgeons to change in hydroelectric

facility operation. Trans. Am. Fish. Soc. 125: 66–77. doi:10.1577/1548-8659(1996)125<0066:ROSLST>2.3.CO;2.

Bauersfeld, K. 1978. The effect of daily flow fluctuations on spawning fall Chi-nook in the Columbia River. Technical Report No. 38. State of Washington,Department of Fisheries, Olympia, Wash.

Becker, C.D., and Neitzel, D.A. 1985. Assessment of intergravel conditions influ-encing egg and alevin survival during salmonid redd dewatering. Environ.Biol. Fish. 12: 33–46. doi:10.1007/BF00007708.

Becker, C.D., Fickeisen, D.H., and Montgomery, J.C. 1981. Assessment of impactsfrom water level fluctuations on fish in the Hanford Reach, Columbia River.PNL-3813, Pacific Northwest Laboratory, Richland, Wash.

Becker, C.D., Neitzel, D.A., and Fickeisen, D.H. 1982. Effects of dewatering onChinook salmon redds: tolerance of four developmental phases to daily de-waterings. Trans. Am. Fish. Soc. 111: 624–637. doi:10.1577/1548-8659(1982)111<624:EODOCS>2.0.CO;2.

Becker, C.D., Neitzel, D.A., and Abernethy, C.S. 1983. Effects of dewatering onChinook salmon redds: tolerance of four development phases to one-timedewatering. N. Am. J. Fish. Manage. 3: 373–382. doi:10.1577/1548-8659(1983)3<373:EODOCS>2.0.CO;2.

Benjamini, Y., and Hochberg, Y. 1995. Controlling the false discovery rate: apractical and powerful approach to multiple testing. J. R. Stat. Soc. Ser. BMethodol. 57: 289–300.

Bernard, D.R., and Clark, J.E. 1996. Estimating salmon harvest with coded-wiretags. Can. J. Fish. Aquat. Sci. 53(10): 2323–2332. doi:10.1139/f96-182.

Bernard, D.R., and Clark, J.H. 1999. Biological based escapement goals for fourpopulations of naturally spawning fall Chinook salmon in the ColumbiaRiver. Pacific Salmon Commission Chinook Technical Committee, Vancou-ver, B.C.

Boyd, J.W., Oldenburg, E.W., and McMichael, G.A. 2010. Color photographicindex of fall Chinook salmon embryonic development and accumulated ther-mal units. PLoS ONE, 5(7): e11877. doi:10.1371/journal.pone.0011877. PMID:20686709.

Bunn, S.E., and Arthington, A.H. 2002. Basic principles and ecological conse-quences of altered flow regimes for aquatic biodiversity. Environ. Manage.30: 492–507. doi:10.1007/s00267-002-2737-0. PMID:12481916.

Burt, D.W., and Mundie, J.H. 1986. Case histories of regulated stream flow and itseffects on salmonid populations. Can. Tech. Rep. Fish. Aquat. Sci. 1477.

Chapman, D.W., Weitkamp, D.E., Welsh, T.L., Dell, M.B., and Schadt, T.H. 1983.Effects of minimum flow regimes on fall Chinook spawning at Vernita Bar1978–1982. Report to the Public Utility District No. 2 of Grant County,Ephrata, Wash.

Chapman, D.W., Weitkamp, D.E., Welsh, T.L., Dell, M.B., and Schadt, T.H. 1986.Effects of river flow on the distribution of Chinook salmon redds. Trans. Am.Fish. Soc. 115: 537–547. doi:10.1577/1548-8659(1986)115<537:EORFOT>2.0.CO;2.

Clarke, K.D., Pratt, T.C., Randall, R.G., Scruton, D.A., and Smokorowski, K.E.2008. Validation of the flow management pathway: effects of altered flow onfish habitat and fishes downstream from a hydropower dam. Can. Tech. Rep.Fish. Aquat. Sci. 2784.

Coleman, A.M., Ward, D.L., Larson, K.B., and Lettrick, J.W. 2010. Development ofa high-resolution bathymetry dataset for the Columbia River through theHanford Reach. PNNL-19878, Pacific Northwest National Laboratory, Rich-land, Wash.

Connor, E.J., and Pflug, D.E. 2004. Changes in the distribution and density ofpink, chum, and Chinook salmon spawning in the upper Skagit River inresponse to flow management measures. N. Am. J. Fish. Manage. 24: 835–852.doi:10.1577/M03-066.1.

Coronado, C., and Hilborn, R. 1998. Spatial and temporal factors affecting sur-vival in coho salmon (Oncorhynchus kisutch) in the Pacific Northwest. Can. J.Fish. Aquat. Sci. 55(9): 2067–2077. doi:10.1139/f98-090.

Cushman, R.M. 1985. Review of ecological effects of rapidly varying flows down-stream from hydroelectric facilities. N. Am. J. Fish. Manage. 5: 330–339. doi:10.1577/1548-8659(1985)5<330:ROEEOR>2.0.CO;2.

Dauble, D.D., and Watson, D.G. 1997. Status of fall Chinook salmon populationsin the mid-Columbia River, 1948–1992. N. Am. J. Fish. Manage. 17: 283–300.doi:10.1577/1548-8675(1997)017<0283:SOFCSP>2.3.CO;2.

Dauble, D.D., Hanrahan, T.P., Geist, D.R., and Parsley, M.J. 2003. Impacts of theColumbia River hydroelectric system on main-stem habitats of fall Chinooksalmon. N. Am. J. Fish. Manage. 23: 641–659. doi:10.1577/M02-013.

De’ath, G., and Fabricius, K.E. 2000. Classification and regression trees: a pow-erful yet simple technique for ecological data analysis. Ecology, 81: 3178–3192. doi:10.1890/0012-9658(2000)081[3178:CARTAP]2.0.CO;2.

Dudgeon, D. 2000. Large-scale hydrological changes in tropical Asia: prospectsfor riverine biodiversity. BioScience, 50(9): 793–806. doi:10.1641/0006-3568(2000)050[0793:LSHCIT]2.0.CO;2.

Fisk, J.M., II, Kwak, T.J., Heise, R.J., and Sessions, F.W. 2013. Redd dewateringeffects on hatching and larval survival of the robust redhorse. River Res.Appl. 29: 574–581. doi:10.1002/rra.2561.

Fritts, A.L., and Pearsons, T.N. 2004. Smallmouth bass predation on hatchery andwild salmonids in the Yakima River, Washington. Trans. Am. Fish. Soc.133(4): 880–895. doi:10.1577/T03-003.1.

Fritts, A.L., and Pearsons, T.N. 2006. Effects of predation by nonnative small-mouth bass on native salmonid prey: The role of predator and prey size.Trans. Am. Fish. Soc. 135: 853–860. doi:10.1577/T05-014.1.

Fryer, J.K. 2005. Stock assessment and research for upper Columbia River fallChinook. Southeast Sustainable Salmon Fund final report for approved proj-ects. Project 45030, Columbia River Intertribal Fish Commission, Portland,Ore.

Geist, D.R., Murray, C.J., Hanrahan, T.P., and Xie, Y. 2008. A model of the effectsof flow fluctuations on fall Chinook salmon spawning habitat availability inthe Columbia River. N. Am. J. Fish. Manage. 28: 1894–1910. doi:10.1577/M07-074.1.

Gibbins, C.N., and Acornley, R.M. 2000. Salmonid habitat modelling studies andtheir contribution to the development of an ecologically acceptable releasepolicy for Kielder Reservoir, North-east England. Regul. Rivers Res. Manage.16: 203–224. doi:10.1002/(SICI)1099-1646(200005/06)16:3<203::AID-RRR579>3.0.CO;2-8.

Gislason, J.C. 1985. Aquatic insect abundance in a regulated stream under fluc-tuating and stable diel flow patterns. N. Am. J. Fish. Manage. 5: 39–46. doi:10.1577/1548-8659(1985)5<39:AIAIAR>2.0.CO;2.

Halleraker, J.H., Saltveit, S.J., Harby, A., Arnekleiv, J.V., Fjeldstad, H.-P., andKohler, B. 2003. Factors influencing stranding of wild juvenile brown trout

614 Can. J. Fish. Aquat. Sci. Vol. 71, 2014

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y B

AT

TE

LL

E P

AC

IFIC

NW

NA

TL

LA

B o

n 03

/31/

14Fo

r pe

rson

al u

se o

nly.

(Salmo trutta) during rapid and frequent flow decreases in an artificial stream.River Res. Appl. 19: 589–603. doi:10.1002/rra.752.

Hamilton, R., and Buell, J.W. 1976. Effects of modified hydrology on CampbellRiver salmonids. Technical Report Series No. PAC/T-76-20. Department of theEnvironment, Fisheries and Marine Service, Habitat Protection Directorate,Vancouver, B.C.

Harlan, K., Roler, R., and Pettit, R. 1998. Run size forecast of the return ofColumbia River fall Chinook salmon stocks in 1998. Columbia River ProgressReport 98-01, Washington Department of Fish and Wildlife, Vancouver,Wash.

Hatten, J.R., Tiffan, K.F., Anglin, D.R., Haeseker, S.L., Skalicky, J.J., andSchaller, H. 2009. A spatial model to assess the effects of hydropower opera-tions on Columbia River fall Chinook salmon spawning habitat. N. Am. J.Fish. Manage. 29: 1379–1405. doi:10.1577/M08-053.1.

Henderson, C., and Foster, R.F. 1957. Studies of smallmouth bass (Micropterusdolomieu) in the Columbia River near Richland, Washington. Trans. Am. Fish.Soc. 86: 112–127. doi:10.1577/1548-8659(1956)86[112:SOSBBM]2.0.CO;2.

Hoffarth, P. 2010. District 4 fish management annual report 2009. WashingtonDepartment of Fish and Wildlife, Olympia, Wash.

Hoffarth, P.A., and Pearsons, T.N. 2012. Priest Rapids Hatchery monitoring andevaluation annual report for 2011–2012. Washington Department of Fish andWildlife, Olympia, Wash.

Hoffarth, P., Fowler, A., and Brock, W. 2003. 2003 evaluation of juvenile fallChinook salmon stranding in the Hanford Reach of the Columbia River.Washington Department of Fish and Wildlife, Olympia, Wash.

Korman, J., Kaplinski, M., and Melis, T.S. 2011a. Effects of fluctuating flows and acontrolled flood on incubation success and early survival rates and growth ofage-0 rainbow trout in a large regulated river. Trans. Am. Fish. Soc. 140:487–505. doi:10.1080/00028487.2011.572015.

Korman, J., Martell, S.J.D., and Walters, C. 2011b. Describing population dynam-ics for early life stages of rainbow trout (Oncorhynchus mykiss) using a stocksynthesis model. Can. J. Fish. Aquat. Sci. 68(6): 1110–1123. doi:10.1139/f2011-046.

Langness, O.P., and Reidinger, K.F. 2003. Escapement goals for upriver bright(URB) fall Chinook salmon stocks of the Columbia River. Washington Depart-ment of Fish and Wildlife, Olympia, Wash.

Langshaw, R.B., and Pearsons, T.N. 2010. Final study plan and designs to deter-mine impacts on fall Chinook in the Hanford Reach under the Hanford ReachFall Chinook Protection Program – License Article 401(a)(14). Public UtilityDistrict No. 2 of Grant County, Ephrata, Wash.

Lukas, J.A., and Orth, D.J. 1995. Factors affecting nesting success of smallmouthbass in a regulated Virginia stream. Trans. Am. Fish. Soc. 124: 726–735. doi:10.1577/1548-8659(1995)124<0726:FANSOS>2.3.CO;2.

McMichael, G.A., Rakowski, C.L., James, B.B., and Lukas, J.A. 2005. Estimated fallChinook salmon survival to emergence in dewatered redds in a shallow sidechannel of the Columbia River. N. Am. J. Fish. Manage. 25: 876–884. doi:10.1577/M04-168.1.

Montgomery, J.C., Fickeisen, D.H., and Becker, C.D. 1980. Factors influencingsmallmouth bass production in the Hanford Area, Columbia River. North-west Sci. 54: 296–305.

Nagrodski, A., Raby, G.D., Hasler, C.T., Taylor, M.K., and Cooke, S.J. 2012. Fishstranding in freshwater systems: Sources, consequences, and mitigation.J. Environ. Manage. 103: 133–141. doi:10.1016/j.jenvman.2012.03.007. PMID:22481278.

Neitzel, D.A., and Becker, C.D. 1985. Tolerance of eggs, embryos, and alevins ofChinook salmon to temperature changes and reduced humidity in dewa-tered redds. Trans. Am. Fish. Soc. 114: 267–273. doi:10.1577/1548-8659(1985)114<267:TOEEAA>2.0.CO;2.

Perkins, W.A., and Richmond, M.C. 2007. MASS2, Modular Aquatic SimulationSystem in Two Dimensions, theory and numerical methods. PNNL-14820-1,Pacific Northwest National Laboratory, Richland, Wash.

Perkins, W.A., Richmond, M.C., and McMichael, G.A. 2004. Two-dimensionalmodeling of time-varying hydrodynamics and juvenile Chinook salmon hab-itat in the Hanford Reach of the Columbia River. In World Water and Envi-ronmental Resources Congress 2004, 27 June to 1 July 2004, Salt Lake City,Utah, USA. Edited by G. Sehlke, D.F. Hayes, and D.K. Stevens. American Societyof Civil Engineers, Reston, Va. p. 180. doi:10.1061/40737(2004)180.

Peters, C.N., Marmorek, D.R., and Parnell, I. (Editors). 1999. PATH decision anal-ysis report for Snake River fall Chinook. ESSA Technologies Ltd., Vancouver,B.C.

Pringle, C.M., Freeman, M.C., and Freeman, B.J. 2000. Regional effects of hy-drologic alterations on riverine macrobiota in the New World: tropical–temperate comparisons. BioScience, 50(9): 807–823. doi:10.1641/0006-3568(2000)050[0807:REOHAO]2.0.CO;2.

Reiser, D.W., and White, R.G. 1981. Incubation of steelhead trout and springChinook salmon eggs in a moist environment. Prog. Fish-Cult. 43: 131–134.doi:10.1577/1548-8659(1981)43[131:IOSTAS]2.0.CO;2.

Reiser, D.W., and White, R.G. 1983. Effects of complete redd dewatering onsalmonid egg-hatching success and development of juveniles. Trans. Am.Fish. Soc. 112: 532–540. doi:10.1577/1548-8659(1983)112<532:EOCRDO>2.0.CO;2.

Richmond, M.C., and Perkins, W.A. 2009. Efficient calculation of dewatered andentrapped areas using hydrodynamic modeling and GIS. Environ. Model.Softw. 24: 1447–1456. doi:10.1016/j.envsoft.2009.06.001.

Ricker, W.E. 1975. Computation and interpretation of biological statistics of fishpopulations. Bull. Fish. Res. Bd. Can. 191: 1–382.

Sall, J. 2002. Monte Carlo calibration of distributions of partition statistics [on-line]. SAS Institute. Available from http://www.jmp.com/software/whitepapers/pdfs/montecarlocal.pdf [accessed November 2013].

Schaller, H.A., and Petrosky, C.E. 2007. Assessing hydrosystem influence ondelayed mortality of Snake River stream-type Chinook salmon. N. Am. J. Fish.Manage. 27: 810–824. doi:10.1577/M06-083.1.

Schaller, H.A., Petrosky, C.E., and Langness, O.P. 1999. Contrasting patterns ofproductivity and survival rates for stream-type chinook salmon (Oncorhynchustshawytscha) populations of the Snake and Columbia rivers. Can. J. Fish.Aquat. Sci. 56(6): 1031–1045. doi:10.1139/f99-037.

Sharma, R., Vélez-Espino, L.A., Wertheimer, A.C., Mantua, N., and Francis, R.C.2013. Relating spatial and temporal scales of climate and ocean variability tosurvival of Pacific Northwest Chinook salmon (Oncorhynchus tshawytscha). Fish.Oceanogr. 22: 14–31. doi:10.1111/fog.12001.

Tiffan, K.F., Garland, R.D., and Rondorf, D.W. 2002. Quantifying flow-dependentchanges in subyearling fall Chinook salmon rearing habitat using two-dimensional spatially explicit modeling. N. Am. J. Fish. Manage. 22: 713–726.doi:10.1577/1548-8675(2002)022<0713:QFDCIS>2.0.CO;2.

Trotzky, H.M., and Gregory, R.W. 1974. The effects of water flow manipulationbelow a hydroelectric power dam on the bottom fauna of the upper Kenne-bec River, Maine. Trans. Am. Fish. Soc. 2: 318–324. doi:10.1577/1548-8659(1974)103<318:TEOWFM>2.0.CO;2.

Whidden, S.M. 1996. The Hanford Reach: protecting the Columbia’s last safehaven for salmon. Environ. Law, 26: 265–297.

Young, P.S., Cech, J.J., Jr., and Thompson, L.C. 2011. Hydropower-related pulsed-flow impacts on stream fishes: a brief review, conceptual model, knowledgegaps, and research needs. Rev. Fish Biol. Fish. 21: 713–731. doi:10.1007/s11160-011-9211-0.

Zhong, Y., and Power, G. 1996. Environmental impacts of hydroelectric projectson fish resources in China. Regul. Rivers Res. Manage. 12: 81–98. doi:10.1002/(SICI)1099-1646(199601)12:1<81::AID-RRR378>3.0.CO;2-9.

Harnish et al. 615

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y B

AT

TE

LL

E P

AC

IFIC

NW

NA

TL

LA

B o

n 03

/31/

14Fo

r pe

rson

al u

se o

nly.