Evaluation of Reconnection Options for White Sturgeon in the Snake...

American Fisheries Society Symposium 56:319–335, 2007© Copyright by the American Fisheries Society 2007

319

Evaluation of Reconnection Options forWhite Sturgeon in the Snake River Using a

Population Viability ModelHENRIETTE I. JAGER* AND MARK S. BEVELHIMER,

Environmental Sciences Division, Oak Ridge National LaboratoryOak Ridge, Tennessee, 37831, USA

KEN B. LEPLA AND JAMES A. CHANDLER

Environmental Affairs, Idaho Power Company, Boise, Idaho, 83702, USA

WEBB VAN WINKLE

Van Winkle Environmental Consulting, Boise, Idaho, Idaho, 83714, USA

Abstract.— This paper describes a simulation study of reconnection options for whitesturgeon Acipenser transmontanus subpopulations in adjacent river segments aboveand below CJ Strike Dam on the Snake River, Idaho, USA. In contrast to the downstreamriver segment, the upstream river segment is long and has areas that are suitable forspawning during normal and wet hydrologic conditions. We evaluated demographicand genetic consequences of upstream and downstream passage using different modelassumptions about trashrack spacing and density-dependent effects on the spawninginterval. Our genetic results predict that, although reconnection would introduce newalleles to the upstream subpopulation, it would also preserve alleles from the down-stream subpopulation by propagating them in the larger subpopulation above the dam.Our demographic results predict that halving the space between trashracks would havelarge and unequivocal benefits, whereas the predicted effects of reconnection weresmaller and more sensitive to model assumptions. Simulated upstream passage tendedto benefit both subpopulations only in the absence of density-dependent limitation. Inthe presence of density dependence, the combination of halved trashrack spacing andupstream and downstream passage produced the best results. Narrower trashracks keptspawning adults in the upstream segment with spawning habitat, while allowing theirprogeny to migrate downstream. Screening appears to be the best option for such aspecies in this configuration of a long river segment acting as a demographic sourceabove a short one acting as a demographic sink.

* Corresponding author: [email protected]

Introduction

This book describes the research and statusesof anadromous sturgeon species worldwide.Loss of habitat threatens many fishes. This istrue whether the habitat is actually destroyedor simply lost to the population because barri-ers prevent access to it. The fact that sturgeons

require multiple habitats, potentially separatedby long distances, makes this group particu-larly vulnerable to habitat loss, followed bypopulation decline and extinction. This sym-posium highlighted two significant habitatthreats facing sturgeon: degradation of estuar-ies and development of large rivers (Munro2007, this volume). All sturgeons spawn infreshwater (Bemis and Kynard 1997). Conse-quently, habitat fragmentation by dams is a

320 JAGER ET AL.

significant concern during movement to fresh-water, for these and other anadromous species(Peter 1998). From a population perspective,losing access to a required habitat has the sameeffect as destroying it.

Our paper focuses on far-inland subpopu-lations of white sturgeon that inhabit river seg-ments between dams in the Snake River.Historically, these subpopulations were likelypotamodromous (migrating within the Colum-bia and Snake rivers) and semianadromous(migrating to the Columbia River estuary).Today, a primary concern is that dams are hav-ing an adverse effect by isolating subpopula-tions. But, it is not entirely accurate to saythat dams isolate subpopulations. Dams havetwo main effects: first, they restrict access foreach subpopulation to a subset of the types oflarge-river habitat that are available in the riveras a whole. Second, as semipermeable barri-ers, dams cause asymmetric migration, withconsiderably more downstream than upstreammovement. We envision these two effects aspathways leading to higher extinction risk un-less efforts are made to avoid them (Figure1).

Restricted access creates a problem if habi-tats needed during different seasons or lifestages are not available. We believe that free-flowing habitat is important for sturgeonspawning and as refuge from episodes of poorwater quality, if and when they occur, in res-ervoirs. Simulation results suggest that extinc-tion risk for individual subpopulations ofsturgeon is high in river segments with toolittle free-flowing habitat (Jager et al. 2002).However, white sturgeon in reservoirs tend tohave higher condition factors (Lepla and Chan-dler 1995), suggesting that access to both ad-equate free-flowing habitat for spawning andgood-quality reservoir habitat might be anasset.

Although spawning habitat is undoubtedlyimportant, it is entirely possible that sturgeonrequire other riverine features, not yet identi-fied. According to Humphries and Lake(2000), attention to flow-related habitat forspawning has obscured the larger role of in-

cubation habitat for native species in Austra-lian rivers. The larvae of most riverine fishspecies require shallow, slow-water habitatswhere productivity of small prey is higher,riparian vegetation provides structure, andpredators are excluded (Bowen et al. 2003).It remains unclear to what extent incubationhabitat contributes to recruitment in whitesturgeon (Coutant 2004). Perrin et al. (2003)proposed that productive areas differ in regu-lated and unregulated rivers and in confinedand unconfined channels. They recovered eggs

Riverfragmentation

Asymmetricmigration

Decline ofupstream

subpopulation

Allee effects

Morevulnerable

to catastrophes

Fewerspawners

Lessfrequent

recruitment

Loss of free-flowing river

Spawning habitatLarval exportUpstream mig

No rescue effectRefuge

FIGURE 1. Schematic of extinction risksassociated with river fragmentation.

EVALUATION OF RECONNECTION OPTIONS FOR WHITE STURGEON IN THE SNAKE RIVER USING A POPULATION VIABILITY MODEL 321

and larvae mainly in side channels andconfluence areas of the unregulated FraserRiver. Habitat diversity alone could be impor-tant by offering protection from episodic, lo-cal threats and by providing prey habitat.Habitat characteristics (e.g., slackwater, habi-tat diversity) that ensure a plentiful supply ofprey (e.g., salmon or shad carcasses, musselbeds, and other benthic prey) for older stur-geon may be important to enhance overwintersurvival and increase the frequency of femalespawning (Bevelhimer 2002).

Asymmetric migration (Figure 1) first re-duces population sizes and, then, prevents im-migration that might “rescue” those that reachlocal extinction (Jager et al. 2001). Dams thatlack upstream passage prevent upstream migra-tion. Downstream migration can easily be toohigh to sustain the upstream population (Fig-ure 1). Sturgeon small enough to become en-trained into the turbines move downstream(whether or not they survive is another mat-ter). In addition, larvae may be exported andlost from upstream subpopulations in shorterriver segments with spawning areas too closeto the dam. Larval export is a problem sharedby other native species of riverine fishes in frag-mented rivers (e.g., Plantania and Altenbach1998). Asymmetric migration is a more impor-tant concern for upstream subpopulations thanfor downstream subpopulations that receive im-migrants from upstream.

Once they become small as a result of fac-tors described above, populations are always athigher risk due to Allee effects—a tendency forsmall populations to decline (bottom of Figure1). Mechanisms for extinction attributed tochance events, demographic stochasticity, in-clude years with catastrophes (e.g., fish kills)and years with skewed sex ratios during whichspawners fail to mate (Engen et al. 2003). Fe-male sturgeons spawn less frequently than mostfishes, and are therefore more likely to experi-ence spawner limitation.

Because upstream movements betweenriver segments are not currently possible in theSnake River, there is a danger that upstreamsubpopulations will be lost. For this reason,

reconnection of subpopulations between dams,via fish ladders, translocation, or stocking, isbeing considered. The hope is that reconnectionwould eliminate risks associated with the leftpathway in Figure 1.

Connecting subpopulations has potentialrisks as well as benefits. In the Snake River,there is no clear evidence that a growing sub-population exists that can withstand removalof individuals to other segments. Human-caused problems that prevent subpopulationsin some river segments from growing may actas demographic sinks when connections withother segments are established. Fixing theselocal problems first may ensure that recon-nection options are more successful. Recon-necting subpopulations could also have adversegenetic consequences. Species with a strongstock structure that evolve in isolation becomeadapted to the local environment and subse-quently lose fitness if they interbreed withstocks adapted to conditions in different partsof the river.

Case Study

In this study, we use a subpopulation viabilitymodel to quantify the costs and benefits to thewhite sturgeon metapopulation of reconnect-ing populations in two river segments in theSnake River: Bliss Dam to CJ Strike Dam andCJ Strike Dam to Swan Falls Dam (Figure 2).The river segment above CJ Strike Dam hasconditions more favorable for spawning andincubation than the segment below the dam.Consequently, restoring upstream linkages be-tween these two segments could potentiallyprovide spawning habitat for adults in thedownstream segment. We use a simulation ap-proach to evaluate upstream passage as a meansof restoring upstream connection between thetwo subpopulations. The main question we ad-dress here is: “What reconnection options (up-stream and/or downstream passage) result inlarger upstream and downstream subpopula-tions than those that would exist withoutreconnection?” In addition, we evaluate theeffect of model assumptions on our results.

322 JAGER ET AL.

River Segment above CJ Strike Dam

The Snake River below Bliss Dam runs 106km to CJ Strike Dam (Figure 2). Waters re-leased from Bliss Dam flow through a narrowupper section with large rapids and deep, tur-bulent run-and-pool habitats. The river trans-forms into a steep canyon, then meandersslowly through braided channels in a low-gra-dient section and slows to join the 39-km res-ervoir pooled behind CJ Strike Dam (Leplaand Chandler 1995). The white sturgeon sub-population above CJ Strike Dam is the larg-est subpopulation in the Snake River aboveHells Canyon Dam and one of two remain-ing naturally reproducing subpopulations inthe Snake River. Lepla and Chandler (1995)estimated a population size of between 1,938

and 4,445 individuals greater than 80 cm forklength.

River Segment below CJ Strike Dam

The Snake River below CJ Strike Dam extends57.9 km to Swan Falls Dam (Figure 2). Thefree-flowing river extends for 17.3 km beforeentering Swan Falls Reservoir (IPC 2003). Thissegment, referred to as CJ-SF, has a low gradi-ent and is dominated by shallow runs and is-land complexes, with few deep pools (Anglinet al. 1992). A 1994–1996 survey estimated asubpopulation of between 473 and 1,565 indi-viduals (>90 cm fork length) below CJ StrikeDam (Lepla and Chandler 1997). In contrast tothe subpopulation above the dam, this riversegment shows little evidence of recruitment.

PacificOcean

Washington

IdahoOregonBon

nevil

le Dam

The D

alles

John

Day

McNary

Dam Hells Canyon Dam

Oxbow DamBrownlee Dam

Shoshone Falls (natural upper limit for migration)

Lower

Granite

Little

Goo

se

Ice H

arbor

Columbia RiverSnake

River

Salmon River

Lower

Monum

ental

Salmon

Fall

s Dam

Bliss D

am

Lower

& U

pper

Swan Falls DamCJ Strike Dam

Snake River

FIGURE 2. These simulations represented the Snake River, Idaho, between ShoshoneFalls Dam and Swan Falls Dam. The particular segments of interest here are the down-stream segment between Swan Falls Dam and CJ Strike Dam and the upstream seg-ment between CJ Strike Dam and Bliss Dam.


Other River Segments

Simulations also include four other river seg-ments (Figure 2):

• Shoshone Falls (the historic upstream limitof white sturgeon distribution in the SnakeRiver) to Upper Salmon Falls Dam,

• Upper Salmon Falls Dam to Lower SalmonFalls Dam,

• Lower Salmon Falls Dam to Bliss Dam, and• Below Swan Falls Dam.

The segment between Shoshone Falls andupper Salmon Falls Dam contains both free-flowing river and a current population of morethan 700 white sturgeon, but the next two down-stream are short, reservoir segments with fewwhite sturgeon. The segment below Swan Fallsis included in these simulations simply to pro-vide realistic downstream boundary conditions.

Methods

We conducted simulation experiments with amodel designed for population viability analy-sis (PVA). The model simulates population dy-namics and predicts the future viability of whitesturgeon subpopulations in the Snake River.Below, we review the structure of the model andprovide references that give more details, wedescribe aspects of the model that deal with simu-lating reconnection in detail, and we describeour simulation experiments.

The PVA model is a stochastic model thatsimulates a sample of replicate subpopulationsinto the future using an annual time step.Subannual processes are represented statistically.Spatially, the model tracks individuals in eachriver segment between dams as members of thesame subpopulation. The primary demographicprediction of the model presented here is finalpopulation size. Two genetic predictions, thefinal number of original and introduced alle-les, are described below. Downstream migra-tion between adjacent segments is influencedby hydrologic year, turbine type, and individualgirth, as is described further below.

We simulate future environmental condi-tions in the Snake River as stochastic sequences

of hydrologic-year types (i.e., dry, normal, orwet). A Markov model simulates transitionfrom one type to another, with probabilitiesestimated from the historical record between1912 and 1995. We defined normal years asthose with annual inflows to Brownlee Reser-voir between 4,616 m3/s and 7,730 m3/s. Wetand dry years fall on either end of this range.

In these simulations, age-0 fish were ex-posed to the following risks: baseline mortal-ity, mortality due to extreme temperaturesduring incubation, larval export, and mortalitydue to poor water quality in summer. Sourcesof mortality applied to older fish includedbaseline mortality, harvest and hooking, poorwater quality in summer, and turbine mortal-ity. Simulation of these factors is described inJager et al. (2002).

Below, we focus on describing aspects ofthe model that pertain specifically to thesesimulation experiments. Other aspects are de-scribed in Jager et al. (2002). The first sec-tion below explains two alternative assump-tions about hydrologic conditions required forspawning below CJ Strike Dam used in thesimulations. The next three sections describeaspects of the model specifically related toreconnection: upstream passage, turbine mor-tality, and screening and downstream passage.Next, we describe initial conditions for themodel, followed by a description of the ge-netic and demographic response variables re-ported for these simulations. The final sectionoutlines the design of our simulation experi-ments.

Spawning

High spring flows are considered important forthe spawning of white sturgeon (Parsley andBeckman 1994). Although the PVA model isdesigned to represent recruitment processesmechanistically in response to average springflows and weighted usable area for spawningand rearing during different hydrologic yeartypes, the simulations performed for this paperuse a simpler alternative. For each river seg-ment, we designate which hydrologic year types

324 JAGER ET AL.

(wet, normal, or dry) provide suitable condi-tions for spawning.

Above CJ Strike Dam, we assume thatspawning can occur in wet and normal years.Variation in the percentage of juvenile whitesturgeon surveyed between 1979 and 2000 sug-gest that higher recruitment occurred in yearswith normal and above-normal spring flows(IPC 2003). Below CJ Strike, we consider twoassumptions: (1) habitat for spawning is neverpresent and (2) habitat for spawning is onlypresent in wet years. Although habitat analysissuggests that spawning habitat should be avail-able in wet years below CJ Strike Dam, em-pirical surveys found little evidence ofrecruitment during five consecutive high-flowyears from 1982 to 1986 or during four con-secutive high-flow years from 1996 to 1999(IPC 2003). The reconnection scenarios listedin Table 1 allow us to consider both cases.

Under favorable hydrologic conditions, thePVA model evaluates reproduction annually foreach subpopulation as described by Jager(2001). Eligible spawners form a mating ag-gregation that participates in broadcast spawn-ing. Each egg produced by females inheritsgenetic material (alleles) from the female par-ent and from a male parent selected at randomfrom the aggregation.

We consider two possibilities for determin-ing the spawning interval for females. In halfof the scenarios, we simulate a fixed spawninginterval of 6 years for females. In the otherhalf, the interval, I, depends on the density ofspawners, where D

t is the density of spawners

in year t, Dopt

= 5 adults/km, Imin

= 3 years, andI

lag = 10 years:

I D II D

D Dt

t

t

( ) = ++min

lag

opt2

2

2

Upstream Passage

We simulate upstream passage by setting therate of upstream movement equal to the maxi-mum downstream rate. In other words, eachindividual experiences the same annual prob-ability of moving upstream and the number of

sturgeon that move upstream increases in pro-portion to the size of the downstream subpopu-lation. We assume that sturgeon ages 1 and olderare equally likely to move upstream when up-stream passage is available.

Turbine Mortality

Each individual sturgeon older than age-0simulated in the model has a chance of attempt-ing downstream migration each year (Jager etal. 2002). This chance reaches a maximum of0.02 per year when sturgeon densities exceedthe capacity of the river habitat. Once an in-dividual “decides” to move downstream, thereare two options: it may spill over the dam orentrain through the turbines. We make the sim-plifying assumption that the likelihood of en-trainment for sturgeon migrating downstreamin a particular hydrologic year type is the sameas the proportion of flow entrained. Thus, en-trainment is less likely in wet years becausethe proportion of flow spilled over the dam ishigher in wet years than in dry years. We as-sume that individuals spilled over the damsurvive.

Individuals that are slated to move down-stream, but that do not spill over the dam, mayavoid entrainment if they are broader in girththan the distance between bars in the trashrack.As the name implies, trash racks are designed toscreen out logs and other large debris that mightdamage the turbines. In some scenarios, we ma-nipulate the trashrack spacing by reducing thespacing by half. Given that an individual stur-geon is small enough to be entrained, its likeli-hood of being struck and killed by a turbinedepends on the size of the fish (Figure 3). Therisk of blade strike increases linearly with in-creasing sturgeon size, and the slope of this re-lationship differs for different types of turbines.CJ Strike Dam has three vertical fixed-blade tur-bines (FERC 2004). Those individuals that sur-vive passage through the turbines are relocatedto the next downstream segment. Narrower spac-ing between trashracks screens out a size rangeof sturgeon that are currently small enough topass through trashracks, but large enough to be


excluded by the new spacing, and prevents themfrom becoming entrained into turbines and risk-ing exposure to blade-strike mortality

Screening and Downstream Passage

Because fish that move upstream are subse-quently subject to entrainment and turbinemortality at CJ Strike Dam, our simulationexperiments evaluate the effects of trashrackspacing and downstream passage to seewhether they influence the efficacy of up-stream passage.

If a particular individual chooses thedownstream segment as its desired destination,it may or may not succeed. The sturgeon mayattempt to move downstream via the spillway(i.e., over the dam, particularly in wet years),

via entrainment with flow through the turbines,or through a downstream passage facility (Fig-ure 3). We estimated the sturgeon’s diameterfrom empirical data from fork length [FL] as0.2765/π FL1.07. If the sturgeon is too large topass through screens, we assume that it is un-able to move downstream unless there is adownstream bypass facility. Simulations withdownstream passage differ from others in tworespects: (1) larger (excluded) sturgeon maymove downstream without turbine mortality,and (2) the size-structure of down-migratingfish more-closely mirrors that of the upstreamsubpopulation.

We simulated a number of scenarios withdownstream passage (Table 1). To simulatedownsteam passage, we assume that sturgeon

Fork length (cm)

0 20 40 60 80 100 120 140 160 180

Ris

k of

ent

rain

men

t an

d bl

ade

strik

e

0.0

0.2

0.4

0.6

0.8

1.0

Current(22.2 cm)

Half(11.1 cm)

DryNormalWet

FIGURE 3. The risk of turbine strike for the ~2% of white sturgeon that attempt tomigrate downstream annually results from two sequential events: Entrainment of thefish through turbines (more likely in dry years than wet years), and the turbine bladestriking the fish (more likely for larger fish). Size thresholds for sturgeon excluded bycurrent and halved trashrack spacing are indicated.

326 JAGER ET AL.

large enough to be excluded by trashracks findthe bypass facility and travel downstream withno risk of mortality. This is a reasonable assump-tion for sturgeon, which grow quickly and, uponentering the downstream segment, might expe-rience a much lower level of predation risk thansalmon would.

Initial Conditions and Calibration

Simulations were initialized with recent esti-mates of the initial densities of sturgeon greaterthan 43 cm. For example, in one replicatesimulation, this resulted in 772 fish belowShoshone Falls, 0 in Upper and 38 in Lower

Salmon Falls reservoirs, 2,662 below BlissDam, and 726 below CJ Strike Dam. We alsospecified initial size distributions (percent ofindividuals less than 95 cm, between 95 and170.5 cm, and greater than 170.5 cm) basedon recent stock assessments in these river seg-ments (Lepla and Chandler 1995, 1997; Leplaet al. 2002, 2004).

We used a previously calibrated value ofage-0 survival, S

y = 0.0002. This value was ob-

tained using simulations with current trashrackspacing, no passage, and a density dependentspawning interval and gave reasonable agree-ment to current stock assessments in historicalsimulations from presumed conditions taken

TABLE 1. The 24 options simulated for CJ Strike Dam vary model assumptions (col-umns two and three) and reconnection scenarios (columns four and five). Model as-sumptions varied concern the potential for spawning below CJ Strike Dam, whether ornot density influences spawning interval. Alternating regular and italic fonts are usedbelow to indicate changes in model assumptions.

Model assumptions Reconnection scenario

Spawning below Density-dependent Trashrack Passage, up andOption CJ Strike Dam spawning? spacing downstream

1 Never No Current None2 Never No Current Up3 Never No Current Both4 Never No Half None5 Never No Half Up6 Never No Half Both7 Never Yes Current None8 Never Yes Current Up9 Never Yes Current Both10 Never Yes Half None11 Never Yes Half Up12 Never Yes Half Both13 Wet years No Current None14 Wet years No Current Up15 Wet years No Current Both16 Wet years No Half None17 Wet years No Half Up18 Wet years No Half Both19 Wet years Yes Current None20 Wet years Yes Current Up21 Wet years Yes Current Both22 Wet years Yes Half None23 Wet years Yes Half Up24 Wet years Yes Half Both


from those below Hell’s Canyon Dam topresent.

Genetic Responses

We quantify the potential for genetic introgres-sion by simulating private alleles, defined asalleles present in one subpopulation but not theother. In the simulations described here, wesimulated a single locus with four private alle-les and four shared alleles initially present ineach subpopulation. We report the final num-ber of original alleles preserved in each sub-population, where the maximum number iseight (neglecting mutation). We also report thefinal number of introduced alleles (private al-leles that originated in the other river segment)found in each subpopulation. Because we alsosimulated three subpopulations in river segmentsupstream of the two segments of interest, thenumber of alleles can be as high as 24 (fourprivate alleles per segment times five segmentsplus four shared alleles).

Demographic ResponsesWe consider upstream and/or downstream pas-sage most beneficial in win–win situations: thosefor which passage increases average final popu-lation sizes (Nf) both above and below CJ StrikeDam compared with those of the correspondingbaseline scenario without passage. However, al-ternative definitions of a successful reconnectionare possible. Because the downstream popula-tion is smaller and at greater risk of extinction,we might be interested in reconnection strate-gies that increase the population below CJ Strikewithout too much negative impact on the popu-lation above CJ Strike. Success along these linesis measured by the index NT500, defined as thesum of final population sizes capped at an arbi-trary size of 500 individuals, which prevents thesum from being dominated by one healthy popu-lation at the expense of others.

Model ScenariosReconnection scenarios.—To evaluate the

potential benefits of providing upstream pas-sage at CJ Strike Dam, we simulated 24 sce-

narios (Table 1). These represent all eight com-binations of assumptions about hydrologic con-ditions for recruitment below the dam (only inwet years versus never), density dependence (ornot), and screening (current versus halvedtrashrack spacing). For each set of assumptions,we simulate three passage options: none, down-stream passage, and both upstream and down-stream passage. The above simulations assumethat spawning is possible during wet and nor-mal years in the segment above CJ Strike Dam.These scenarios address the left risk pathwayin Figure 1.

Predam scenarios.—In addition, we simu-lated two predam scenarios with no dam be-tween Bliss and Swan Falls dams, one withand one without density dependence. We in-cluded these scenarios because the otherreconnection options (those described above)do not change the relative amounts of free-flowing river and reservoir habitat available,whereas removing the dam increases the lengthof free-flowing river by 46%. These scenariosremove the right risk pathway depicted in Fig-ure 1.

For each scenario, we ran 100 replicatesimulations over 500 years. Statistics reportedinclude final subpopulation size, the numberof introduced alleles acquired, and the num-ber of original alleles remaining. For each ofthree response variable, we report subpopula-tion averages and standard errors for the rep-licate simulations. We compare baselineresponses with those of each reconnection op-tion (Table 1).

Results

Demographic

Upstream passage increased the simulated fi-nal sizes of both subpopulations relative to theno-passage baseline only in scenarios that as-sumed no density dependence (Appendix A).These win–win scenarios are illustrated by ra-tios that exceed one in Figure 4. Upstream pas-sage was beneficial to both subpopulations intwo scenarios with current trashrack spacing

328 JAGER ET AL.

FIGURE 4. Bar height indicates the ratio of final population size, Nf, for the reconnectionscenario indicated on the x-axis to that of the corresponding baseline scenario with nopassage. Scenarios combine current (Current) or half trashrack spacing (Half) withupstream passage (Up) or both upstream and downstream passage (Both). The hori-zontal line at one is the line of no net effect. The four bars represent different modelassumptions about density dependence (DD) and the potential for spawning below CJStrike Dam. Simulated results are shown for the populations (a) above, and (b) belowCJ Strike Dam. The final population size, Nf, is predicted to be more than 6,000 aboveCJ Strike Dam for all scenarios (not shown), but Nf below CJ Strike Dam varies, asshown by the line graph in (b).


(solid white and black bars in Figure 4, “Cur-rent, Up”), and in the scenario with halvedtrashrack spacing that assumed spawning wasnot possible below CJ Strike (white bar inFigure 4, “Half, Up”). Adding downstreampassage was not beneficial in scenarios thatassumed no density dependence (solid bars in“Current, Both” and “Half, Both” in Figure4).

When we assumed that spawning intervalswould increase at high density (striped andcrosshatched bars in Figure 4), there were nowin–win scenarios that increased subpopula-tions both above and below the dam. However,halved trashrack spacing with both upstreamand downstream passage resulted in significantincreases in the downstream population, withthe upstream subpopulation remaining large(Figure 4). The NT500 index also favors thisoption (Appendix). Thus, adding downstreamto upstream passage benefited the downstreamsubpopulation only when density dependencewas simulated (Figure 4b), and at the expenseof the subpopulation above the dam.

Halved trashrack spacing increased the twosubpopulations in all scenarios, sometimessubstantially, by increasing the subpopulationabove the dam and providing trickle-downbenefits downstream. This was true with orwithout upstream passage. Returning sturgeonupstream via upstream passage was beneficialif the possibility of spawning downstream wasnil and no density dependent impacts weresimulated upstream (open bar in Figure 4b,“Half, Up” scenario).

We expected the reference scenarios with-out CJ Strike Dam, and therefore more free-flowing habitat, to grow a larger singlepopulation than the two combined subpopula-tions of other reconnection options. As expected,the predam scenario produced a larger averagefinal population than the combined sizes of thetwo segments in other reconnection scenariosin simulations with density dependence, (solidbars in Figure 5). In the absence of densitydependence, the combined population size pro-duced by scenarios with halved trashrack spac-ing and no downstream passage exceeded that

of the corresponding predam scenario (stripedbars in Figure 5).

Genetic

Providing upstream and/or downstream pas-sage increased the retention of original alle-les below the dam (Figure 6). This wasparticularly true when spawning was possibleduring wet years. The pattern for introducedalleles acquired and maintained in the popu-lation above the dam was more complex.Recall that these alleles could have originatedin one of the three populations above BlissDam, or they might have been acquired fromthe population below CJ Strike via upstreampassage.

First consider the baseline scenarios withno passage (Figure 6). The number of intro-duced alleles acquired from upstream was verylow in baseline simulations with current spac-ing but higher in simulations with both halvedtrashrack spacing and no density dependence.In the former case, few of the 12 alleles pri-vate to the three upstream populations mi-grated to the Bliss–CJ Strike population, or ifthey did, nearly all were lost to genetic drift.In contrast, the frequencies of alleles acquiredfrom upstream increased in simulations withhalved trashrack spacing and no density de-pendence (Figure 6). We attribute this resultto fact that the population above the damreached much larger final sizes (~9,000 indi-viduals) in these scenarios.

Next, we consider the effect of passageon the acquisition and retention of introducedalleles. In all scenarios, upstream passage in-creased the final number of introduced alleles(Figure 7). The final number of introducedalleles was especially high in scenarios thatassumed no density dependent effects on thespawning interval and those with halvedtrashrack spacing—the same scenarios that re-sulted in large final populations. Scenarioswith both upstream and downstream passagealso maintained higher numbers of introducedalleles, but not quite as many as scenarios withupstream passage alone (Figure 7).

330 JAGER ET AL.

FIGURE 5. For postdam reconnection scenarios we assumed recruitment was possiblebelow the dam in wet years. Bar height indicates the sum of average final populationsizes above and below the dam. For the predam scenario, bar height is the final popu-lation size of the single population. Results are shown for simulations that assumeddensity dependence (black bars) and those that did not (striped bars). Scenarios (x-axis) combine current (Current) or half trashrack spacing (Half) with upstream passage(Up) or both upstream and downstream passage (Both).

Discussion

The simulated benefits of reconnection dependon model assumptions, whereas those of reducedtrashrack spacing appear to be robust. Our simu-lation results suggest that upstream passage ismore likely to benefit both populations whendensity dependence is weak and downstreampassage is not provided. If density dependenceis strong, the best we can hope for is thatreconnection will permit the upstream popula-tion to bolster the downstream subpopulation,at the cost of a smaller, but self-sustaining, up-stream subpopulation. The scenarios that accom-plished this result required halved trashrackspacing and downstream passage, in addition toupstream passage. Density dependence played an

important role in this situation1 because the po-tential for increasing the larger, upstream popu-lation is much higher when it is absent. However,the population sizes attained by the upstreampopulation in the absence of density dependencein some scenarios appear unrealistically high tous, suggesting that density dependence is notcompletely absent in the real world.

Given this uncertainty, perhaps the bestcourse of action would be to implement upstreamtranslocation as a means of determining the re-sponses of the two subpopulations, before in-

1 This situation is a river configuration with a largersturgeon subpopulation in a longer river segment abovea smaller subpopulation that has little or no potential forrecruitment in a shorter segment.


vesting in passage structures. Translocation (cap-turing sturgeon below the dam and moving themabove the dam) is one alternative method forproviding upstream passage. In a preliminaryanalysis2, we compared upstream passage with

upstream translocation (both with halvedtrashrack spacing). We found that simulated up-stream translocation was generally more benefi-cial than upstream passage because we assumedall sizes of sturgeon could take advantage ofupstream passage, whereas the size-selectivity ofgear caused translocation to focus on larger in-dividuals (i.e., those closer to breeding age).

FIGURE 6. Bar height indicates the number of original alleles in the final populationsimulated below CJ Strike Dam for the reconnection scenario indicated on the x-axis.Scenarios combine current or half trashrack spacing with three passage options (none,upstream passage, or both upstream and downstream passage). The four bars repre-sent different model assumptions about density dependence (DD) and the potential forspawning below CJ Strike Dam. Simulated values above the dam are all equal to seven.

2 These simulations assumed a density dependentincrease in spawning interval.

332 JAGER ET AL.

Unlike those of passage, the simulated ben-efits of halved trashrack spacing were unequivo-cal: Halved trashrack spacing always increasedboth the upstream and downstream subpopula-tions. We imagine that intermediate-sized stur-geon that are excluded by trash bars continuedto spawn in the upstream reach, and that some

fraction of their offspring move downstreamto support the population below the dam. Re-turning sturgeon upstream via upstream pas-sage was beneficial only when there was noopportunity for spawning downstream and whenincreased upstream densities did not lead tolonger female spawning intervals.

FIGURE 7. Bar height indicates the number of introduced alleles in the final populationsimulated above CJ Strike Dam for the reconnection scenario indicated on the x-axis.Scenarios combine current or half trashrack spacing with three passage options (none,upstream passage, or both upstream and downstream passage). The four bars repre-sent different model assumptions about density dependence (DD) and the potential forspawning below CJ Strike Dam. Simulated values below the dam did not vary much,with a range between 4.0 and 5.5 alleles.


Not surprisingly, predam simulations, (i.e.,those with one large population, no dam, andmore free-flowing habitat) usually producedbetter results than simulations with artificialreconnection. This confirms the idea that theamount of free-flowing habitat can be as impor-tant as reconnection for species that require free-flowing river habitat. Scenarios with reducedtrashrack spacing were the only exception: theseproduced final populations almost as large, orlarger than, the corresponding predam scenarios.

This case study confirms the results of amore-general examination of connection in dif-ferent river configurations that also used thismodel (Jager 2006a). In that study, reducedtrashrack spacing without upstream passage wasfound to be the best option for the metapop-ulation as a whole in configurations with a longriver segment above a short river segment (Jager2006a), but not for short segments above longsegments. In the case of translocation, closertrashrack spacing caused the recipient segmentto benefit from subsidy by a long upstream seg-ment (Jager 2006b).

The genetic results of this study showed thatthe number of alleles introduced from other sub-populations increased with upstream recon-nection, but the number of original allelespreserved in the downstream segment also in-creased. By sharing its private alleles with thedemographically stronger upstream subpopula-tion, alleles that originally occurred only in thesmaller downstream subpopulation were effec-tively protected. On the other hand, we observedthat introduced alleles were also more likely tobe maintained when introduced to a growingupstream population. Given the proximity of thetwo subpopulations and the benefit of retainedalleles that is suggested by this study, recon-nection in the situation described here would bemore likely to have a positive than a negativegenetic impact on these subpopulations.

We conclude that designing a reconnectionstrategy to help restore subpopulations is notnecessarily as simple as one might think. Thisstudy highlights the importance of unknown fac-tors, such as density dependence, that are pre-dicted to influence the response. Another study

found that response to reconnection depends onriver configuration (Jager 2006a). In a thirdanalysis, we evaluated the effects of reconnect-ing subpopulations in the larger system that addsthree river segments above Bliss Dam, none ofwhich is believed to be self-sustaining, to thetwo segments considered here. We found thatconnecting pairs of subpopulations in short up-stream river segments via upstream passage wasnot beneficial. Only options that linked series ofsmaller segments to a self-sustaining downstreamsource population between Bliss and CJ Strikedams were beneficial. Together, these studies areforming a picture of the benefits and costs ofpassage for sturgeon, and the uncertainties thatneed to be resolved.

Providing fish passage through dams is of-ten required during relicensing of hydropowerfacilities. The goal of these policies is to increasefish populations, particularly those that haveexperienced declines following impoundment.Because passage structures are costly, it is re-markable how few studies have been done todetermine under what conditions recon-nectionhas actually restored populations. We suggest aneed for field studies designed with enoughpower (enough dams and years/dam) to detectchanges in population size before and after imple-mentation of passage options. Such studies wouldprovide a scientific basis for policy decisions,which would help the hydropower industry toallocate resources in a way that is most likely tobenefit the fish populations involved.

AcknowledgmentsThis research was sponsored in part by IdahoPower Company under U.S. Department ofEnergy (DOE) contract No. ERD-99–1813.ORNL is managed by UT-Battelle, LLC, for theDOE under contract DE-AC05–00OR22725. Weappreciate reviews by two anonymous review-ers. We especially thank Jean Munro for orga-nizing the symposium.

References

Anglin, D. R., T. R. Cummings, and A.E. Ecklund.1992. Swan Falls Instream Flow Study. U.S.

334 JAGER ET AL.

Fish and Wildlife Service and Idaho PowerCo., Report No. AFF1-FRO-92–14, Boise.

Bemis, W. E., and B. Kynard. 1997. Sturgeon riv-ers: an introduction to acipenseriform bioge-ography and life history. EnvironmentalBiology of Fishes 48:167–183.

Bevelhimer, M. S. 2002. A bioenergetics modelfor white sturgeon Acipenser transmontanus:assessing differences in growth and reproduc-tion among Snake River reaches. Journal ofApplied Ichthyology 18:550–556.

Bowen, Z. H., K. D. Bovee, and T. J. Waddle. 2003.Effects of flow regulation on shallow-waterhabitat dynamics and floodplain connectiv-ity. Transactions of the American FisheriesSociety 132:809–823.

Coutant, C. C. 2004. A riparian habitat hypoth-esis for successful reproduction of white stur-geon. Reviews in Fisheries Science 12:23–73.

Engen, S., R. Lande, and B.-E. Saether. 2003.Demographic stochasticity and Allee effectsin populations with two sexes. Ecology84:2378–2386.

FERC (Federal Energy Regulatory Commission).2004. Order Issuing New License for ProjectNo. 2055–011 (C.J. Strike Project).

Humphries, P., and P. S. Lake. 2000. Fish larvaeand the management of regulated rivers.Regulated Rivers: Research and Management16:421–432.

IPC (Idaho Power Company). 2003. Snake RiverWhite Sturgeon Conservation Plan. IdahoPower Company, Boise.

Jager, H. I. 2001. Individual variation in life his-tory characteristics can influence subpopula-tion extinction risk. Ecological Modelling144(1):59–74.

Jager, H. I. 2006a. Chutes and ladders, and othergames we play with rivers: I. Upstream pas-sage. Canadian Journal of Fisheries andAquatic Sciences 63(1):165–175.

Jager, H. I. 2006b. Chutes and ladders, and othergames we play with rivers: II. Translocation.Canadian Journal of Fisheries and AquaticSciences 63(1):176–184.

Jager, H. I., W. Van Winkle, K. B. Lepla, and J. A.Chandler. 2001. A theoretical study of riverfragmentation by dams and its effects onwhite sturgeon populations. EnvironmentalBiology of Fishes 60:347–361.

Jager, H. I., W. Van Winkle, K. B. Lepla, J. A. Chan-dler, P. Bates, and T. Counihan. 2002. Factors

controlling white sturgeon recruitment in theSnake River: a simulation study. Pages 127–150 in W. Van Winkle, P. Anders, and D. H.Secor, editors. Biology, management, and pro-tection of sturgeon, American Fisheries Soci-ety, Bethesda, Maryland.

Lepla, K., P. Bates, and B. Bentz. 2004. Stockassessments of Snake River white sturgeonbelow Upper and Lower Salmon Falls Dams,Idaho. Power Company, Boise.

Lepla, K. B., J. Anderson, and P. Bates. 2002. Anassessment of white sturgeon in the ShoshoneFalls-Upper Salmon Falls Reach of the SnakeRiver, Idaho. Power Company, Boise.

Lepla, K. B., and J. A. Chandler. 1995. A survey ofwhite sturgeon in the Bliss Reach of theMiddle Snake River, Idaho. Idaho PowerCompany, Technical Report Appendix E. 3.1-E, Boise.

Lepla, K. B., and J. A. Chandler. 1997. Status ofwhite sturgeon in the C.J. Strike Reach of themiddle Snake River, Idaho. Idaho Power Com-pany, Technical Report Appendix E. 3.1-B,Boise.

Munro, J, R. E. Edwards, and A. W. Kahnle. 2007.Anadromous sturgeons: habitats, threats, andmanagement. Summary and synthesis. Pages1–15 in J. Munro, D. Hatin, J. E. Hightower,K. McKown, K. J. Sulak, A. W. Kahnle, and F.Caron, editors. Anadromous sturgeons: habi-tats, threats, and management. American Fish-eries Society, Symposium 56, Bethesda,Maryland.

Parsley, M. J., and L. G. Beckman. 1994. Whitesturgeon spawning and rearing habitat in theLower Columbia River. North American Jour-nal of Fisheries Management 14:812–827.

Perrin, C. J., L. L. Rempel, and M. L. Rosenau.2003. White sturgeon spawning habitat in anunregulated river: Fraser River, Canada.Transactions of the American Fisheries Soci-ety 132:154–165.

Peter, A. 1998. Interruption of the river con-tinuum by barriers and the consequences formigratory fish. Pages 99–112 in M. Jung-wirth, S. Schmutz and S. Weiss , editors. Fishmigration and fish bypasses. Blackwell Sci-entific Publications Inc., Malden, Massachu-setts.

Plantania, S. P., and C. S. Altenbach. 1998. Repro-ductive strategies and egg types of seven RioGrande Basin cyprinids. Copeia 3:559–569.


APP

END

IX A

. Si

mul

atio

n re

sults

for s

ix o

ptio

ns re

pres

entin

g al

l com

bina

tions

of r

econ

nect

ion

scen

ario

s an

d m

odel

ass

umpt

ions

.A

vera

ge a

nd S

E fin

al p

opul

atio

n si

ze (

Nf)

is r

epor

ted

for

subp

opul

atio

ns u

pstr

eam

and

dow

nstr

eam

of

CJ

Strik

e D

am.

T-te

stst

atis

tics

com

pare

Nf

with

the

app

ropr

iate

no-

pass

age

scen

ario

. Th

e la

st t

wo

colu

mns

rep

ort

the

sum

of

aver

age

Nf

and

the

NT5

00 i

ndex

, th

e su

m o

f m

in(N

f, 5

00)

for

the

two

subp

opul

atio

ns c

ombi

ned.

Alte

rnat

ing

regu

lar

and

italic

fon

ts a

re u

sed

belo

w t

o in

dica

te c

hang

es in

mod

el a

ssum

ptio

ns.

T-s

tatis

ticR

econ

nect

ion

Bli

ss D

am to

CJ

Str

ike

Dam

to

com

pare

d to

Mod

el a

ssum

ptio

nsc

enar

ioC

J S

trik

e D

amS

wan

Fal

ls D

amP

assa

ge =

non

e

Spaw

ning

Den

sity

Tra

sh-

belo

w C

Jde

pen-

rack

Com

bine

dN

T50

0S

trik

e D

amde

nce

spac

ing

Pass

age

Avg

. N

fS

E N

fA

vg.

Nf

SE

Nf

Abo

veB

elow

Nf

avg.

inde

x

neve

rno

curr

ent

none

7,75

9.4

160.

928

7.3

5.5

0.0

0.0

8,04

6.7

787.

3ne

ver

nocu

rren

tup

10,9

42.9

339.

137

0.9

10.5

12.7

10.5

11,3

13.8

870.

9ne

ver

nocu

rren

tbo

th5,

906.

111

5.2

216.

23.

7–1

3.4

–15.

56,

122.

371

6.2

neve

rno

half

none

594,

778.

018

,702

.717

,028

.147

8.0

0.0

0.0

611,

806.

11,

000.

0ne

ver

noha

lfup

808,

365.

013

,159

.819

,207

.652

4.6

13.4

4.3

827,

572.

61,

000.

0ne

ver

noha

lfbo

th10

,168

.927

4.7

538.

512

.6–6

1.6

–67.

210

,707

.41,

000.

0ne

ver

yes

curr

ent

none

6,67

5.8

99.8

250.

92.

70.

00.

06,

926.

775

0.9

neve

rye

scu

rren

tup

6,88

1.7

97.9

239.

13.

02.

1–4

.17,

120.

873

9.1

neve

rye

scu

rren

tbo

th6,

404.

196

.123

4.2

2.6

–2.8

–6.2

6,63

8.3

734.

2ne

ver

yes

half

none

9,08

9.4

142.

232

8.6

3.8

0.0

0.0

9,41

8.0

828.

6ne

ver

yes

half

up9,

426.

413

4.0

318.

23.

62.

4–2

.89,

744.

681

8.2

neve

rye

sha

lfbo

th6,

889.

496

.837

1.3

3.3

–18.

412

.07,

260.

787

1.3

wet

yea

rsno

curr

ent

none

7,74

8.5

185.

61,

461.

679

.90.

00.

09,

210.

11,

000.

0w

et y

ears

nocu

rren

tup

19,8

62.5

613.

61,

677.

278

.630

.32.

721

,539

.71,

000.

0w

et y

ears

nocu

rren

tbo

th10

,223

.926

0.9

1,40

7.6

79.2

11.1

–0.7

11,6

31.5

1,00

0.0

wet

yea

rsno

half

none

577,

999.

017

,769

.447

,025

.82,

252.

30.

00.

062

5,02

4.8

1,00

0.0

wet

yea

rsno

half

up83

7,40

0.0

6,31

4.9

39,3

22.4

1,61

8.1

21.5

–4.0

876,

722.

41,

000.

0w

et y

ears

noha

lfbo

th37

,890

.21,

131.

78,

707.

453

3.8

–57.

2–2

7.5

46,5

97.6

1,00

0.0

wet

yea

rsye

scu

rren

tno

ne6,

758.

691

.61,

567.

881

.90.

00.

08,

326.

41,

000.

0w

et y

ears

yes

curr

ent

up7,

384.

810

6.8

999.

852

.26.

3–8

.58,

384.

61,

000.

0w

et y

ears

yes

curr

ent

both

7,07

2.8

99.8

1,30

0.8

68.1

3.3

–3.6

8,37

3.6

1,00

0.0

wet

yea

rsye

sha

lfno

ne9,

028.

313

2.1

1,82

4.1

93.2

0.0

0.0

10,8

52.4

1,00

0.0

wet

yea

rsye

sha

lfup

10,3

38.4

147.

01,

210.

556

.49.

4–8

.211

,548

.91,

000.

0w

et y

ears

yes

half

both

7,95

1.0

116.

02,

233.

611

9.1

–8.7

3.9

10,1

84.5

1,00

0.0

Evaluation of Reconnection Options for White Sturgeon in the Snake...

Documents

Transcript of Evaluation of Reconnection Options for White Sturgeon in the Snake...