Environmental Biology of Fishes

, Volume 94, Issue 1, pp 87–100

Ecological risk assessment of multiple hatchery programs in the upper Columbia watershed using Delphi and modeling approaches

Authors

    • Public Utility District Number 2 of Grant County
  • Andrew R. Murdoch
    • Washington Department of Fish and Wildlife
  • Greg Mackey
    • Public Utility District Number 1 of Douglas County
  • Keely G. Murdoch
    • Yakama Nation
  • Tracy W. Hillman
    • BioAnalysts Inc
  • Matthew R. Cooper
    • United States Fish and Wildlife Service MCRFRO
  • Joseph L. MIller
    • Public Utility District Number 1 of Chelan County
Article

DOI: 10.1007/s10641-011-9884-1

Cite this article as:
Pearsons, T.N., Murdoch, A.R., Mackey, G. et al. Environ Biol Fish (2012) 94: 87. doi:10.1007/s10641-011-9884-1

Abstract

Ecological risks of Pacific salmon (spring, summer, and fall run Chinook, coho, and sockeye salmon) and steelhead trout hatchery programs operated between 2013 and 2023 in the Upper Columbia Watershed will be assessed using Delphi and modeling approaches. Committees composed of resource managers and public utility districts identified non-target taxa of concern (i.e., taxa that are not the target of supplementation), and acceptable hatchery impacts (i.e., change in population status) to those taxa. Biologists assembled information about hatchery programs, non-target taxa, and ecological interactions and this information will be provided to expert panelists in the Delphi process to facilitate assessment of risks and also used to populate the Predation, Competition, and Disease (PCD) Risk 1 model. Delphi panelists will independently estimate the proportion of a non-target taxa population that will be affected by each individual hatchery program. Estimates from each of the two approaches will be independently averaged, a measure of dispersion calculated (e.g., standard deviation), and subsequently compared to the acceptable hatchery impact levels that were determined previously by committees of resource managers and public utility districts. Measures of dispersion will be used to estimate the scientific uncertainty associated with risk estimates. Delphi and model results will be compared to evaluate the qualities of the two approaches. Furthermore, estimates of impacts from each hatchery program will be combined together to generate an estimate of cumulative impact to each non-target taxon.

Keywords

Ecological riskColumbia RiverHatcherySalmonSteelheadCompetitionPredationDiseaseRisk assessmentAdaptive management

Introduction

Salmon and steelhead (Oncorhynchus spp.) released from hatcheries in the upper Columbia Watershed may pose ecological risks to valued fish taxa that are not the target of a particular hatchery program (e.g., non-target taxa of concern, NTTOC). More than 20 million salmon and steelhead will be released annually from a variety of hatcheries in the upper Columbia Watershed between 2013 and 2023 (Fig. 1); most of these fish are subyearling or yearling smolts. The species and races that are propagated include spring, summer, and fall run Chinook salmon (O. tshawytscha), coho salmon (O. kisutch), sockeye salmon (O. nerka), and summer run steelhead trout (O. mykiss). These fish will be released into areas that contain species protected under the United States of America’s Endangered Species Act (ESA), or are highly valued for their commercial, recreational, or cultural attributes. Existing and proposed hatchery programs in the Columbia Basin are being modified to be consistent with Hatchery and Scientific Review Group (HSRG 2009) principles, and many new or expanded hatchery programs will be implemented within the coming decade. The HSRG evaluation addressed genetic and density-dependent interactions of the salmon and steelhead target species, but generally avoided ecological interactions between hatchery-produced fish and NTTOC. Furthermore, genetic interactions in the upper Columbia Watershed, the area of this evaluation, are being monitored as part of the public utility district funded hatchery programs (Murdoch and Peven 2005; Pearsons and Langshaw 2009).
https://static-content.springer.com/image/art%3A10.1007%2Fs10641-011-9884-1/MediaObjects/10641_2011_9884_Fig1_HTML.gif
Fig. 1

Map of release locations for anadromous salmoinds in the upper Columbia Watershed

Many species of fish inhabit the upper Columbia Watershed, but a small subset of these is of particular concern relative to ecological risks associated with hatcheries. These include spring and summer Chinook salmon (O. tshawytscha), summer steelhead trout (O. mykiss), westslope cutthroat trout (O. clarki), and Pacific lamprey (Entosphenus tridentatus). Upper Columbia spring Chinook salmon were listed under ESA as endangered in 1999 and Upper Columbia steelhead trout as endangered in 1997 and later reclassified to threatened in 2006. Spring Chinook salmon occur in the Wenatchee, Entiat, and Methow Watersheds: important components of these populations include White River, Nason Creek, Chiwawa River, Twisp River, Chewuch River and Methow River. Although spring Chinook salmon occurred in the Okanogan historically (Vedan 2002), they are currently extinct in the watershed.

Important populations of steelhead are found in the Wenatchee, Methow, and Okanogan rivers. Summer Chinook salmon populations are relatively healthy and large numbers of fish spawn in the Wenatchee, Methow, and Okanogan rivers. They support a large fishery in the ocean and main stem Columbia River. Sockeye salmon are found in Lake Wenatchee and Osoyoos Lake, are self-sustaining, and during some years very abundant. Cutthroat trout are widely distributed in the upper Columbia, particularly in high elevation tributaries where they provide for dispersed recreational angling opportunities. Pacific lamprey are relatively scarce, likely at a fraction of historical abundance, and are particularly valued by Native Americans for food.

Hatcheries in the upper Columbia River Watershed were built to mitigate for impacts associated with the construction and operation of hydro-electric facilities. Grant County Public Utility District (PUD) has hatchery mitigation for Priest Rapids and Wanapum dams built in 1959 and 1963 respectively; Chelan County PUD for Rock Island and Rocky Reach dams built in 1933 and 1961 respectively; Douglas County PUD for Wells Dam built in 1967, and the Federal Government for Grand Coulee and Chief Joseph dams built in 1942 and 1955 respectively. Hatchery mitigation generally includes losses associated with habitat inundation, dam operation, and blocked habitat. Hatchery mitigation by the public utility districts is overseen by committees of natural resource managers (e.g., representatives from the National Marine Fisheries Service, United States Fish and Wildlife Service, Washington State Department of Fish and Wildlife, Yakama Nation, and Colville Confederated Tribes) and public utility districts. These committees will be referred to as “resource committees.”

The hatchery programs in this study include existing and new programs that are anticipated to be operational between 2013 and 2023. In some cases, the specifics of a program were tenuous, but the best available information was used to describe the anticipated program. The PUD hatchery programs are designed to be adaptively managed as new information is generated from comprehensive monitoring and evaluation programs. One of the regional objectives in the public utility districts hatchery monitoring and evaluation plans is to assess ecological risks associated with their hatchery programs (Murdoch and Peven 2005; Pearsons and Langshaw 2009). In order to assess ecological risks from all hatchery programs in the upper Columbia Watershed, federally funded hatcheries were also included in the risk assessment.

The most common approaches that have been used to conduct ecological risk assessments of hatchery impacts are literature reviews, expert opinion (Delphi) and mathematical modeling (Ham and Pearsons 2001; Pearsons 2008, 2010). The latter two approaches are used in this assessment and were supplemented with literature reviews. More specifically, a modified version of an expert based approach described by Pearsons and Hopley (1999) and the Predation, Competition, and Disease (PCD) Risk 1 model (Busack et al. 2005; Pearsons and Busack 2011) will be used to assess risk. The PCD Risk 1 model is an individual based model for salmonid fishes that simulates competitory, predatory, and pathogenic interactions in freshwater.

The goals of this risk assessment are to: 1) estimate the impact of hatchery programs on NTTOC, individually and cumulatively; 2) estimate the scientific uncertainty of impacts; 3) compare the results from the two risk assessment approaches; and 4) compare the estimates of impact and uncertainty to containment objectives set by resource managers. Containment objectives are impact levels that alert managers of negative effects to NTTOC that cause concern and potentially require management action. This manuscript is limited to describing the approach that is being used in the upper Columbia Watershed and not the results of the risk assessment. Describing this approach is a worthwhile contribution because this ecological risk assessment is one of the largest of its kind that has been attempted, includes a broad array of hatchery programs and biological diversity, includes important lessons associated with the approach, and is uncommon because it includes and compares two different hatchery risk assessment approaches.

Methods

The information needed to conduct the expert-based approach (referred to here after as Delphi) and modeling (referred to as Predation, Competition, and Disease Risk 1 model (PCD Risk)) risk assessments was assembled by resource and technical committees, and by invited local experts. The process involves two panels of experts: a panel of local experts to help assemble information (referred to as local experts), and a panel of experts to participate in the Delphi approach (referred to as Delphi panelists). Local experts were identified as people who have the most information about particular hatcheries and NTTOC. The Delphi panelists will not necessarily have local expertise, but will represent expertise in various ecological interactions. Experts about ecological interactions will generally be recognized by having published in a peer-reviewed journal. The information that was needed to conduct the risk assessments has been described for the Delphi approach by Pearsons and Hopley (1999) and the modeling approach by Busack et al. (2005) and Pearsons and Busack (2011). The general approach included: 1) identifying NTTOC, containment objectives, and regional experts, 2) assembling information relevant to conducting risk assessments, 3) estimating impacts with associated uncertainty using Delphi and modeling approaches, and 4) comparing impacts and uncertainty to containment objectives. Steps 1 and 2 have been completed and steps 3 and 4 will be completed in the future. The results section only presents information about steps 1 and 2 and the discussion section describes future work associated with steps 3 and 4. A list of tasks that have been completed and those that are yet to be completed are presented in Fig. 2. In addition, the groups that work on each of the tasks are also identified in Fig. 2.
https://static-content.springer.com/image/art%3A10.1007%2Fs10641-011-9884-1/MediaObjects/10641_2011_9884_Fig2_HTML.gif
Fig. 2

Past and future tasks associated with the risk assessment. The different groups that perform each task are also represented by the different shape encompassing each task. The ellipse represents the tasks of the resource committee, the rectangle represents the tasks of the local experts, and the wavy figure represents the task of the Delphi Panel

Resource committees consisting of resource managers and public utility districts identified NTTOC (Fig. 2). They developed containment objectives within which hatchery effects (i.e., change in baseline status) to those taxa would be acceptable, and identified experts to be invited to participate in the risk assessment panel. Status was generally defined as the abundance, distribution, and size structure of the population prior to 2013. Bull trout were not included in the assessment because they are included in a separate ESA assessment conducted by the United States Fish and Wildlife Service. The resource committees attempted to produce a large and balanced panel of scientific experts to minimize bias in the Delphi process.

Local experts assembled standardized information into three templates that contain the information needed for the risk assessment: 1) hatchery program size and biological data, 2) NTTOC population and biological data, and 3) data describing ecological interactions between NTTOC and hatchery fish. The information collected was the input for the PCD Risk 1 model and for use by the Delphi panelists (Tables 1, 2 and 3). The local experts developed rules, equations, and standards to populate templates (Tables 1, 2 and 3). We used in order of priority: 1) best available data, 2) literature values if local data were not available, and 3) best professional judgment to populate data templates. Sources of data were recorded on each of the templates and included published values, unpublished data, and expert opinion. Assumptions were also documented. Where necessary, the minimum, most probable, and maximum values were estimated. This provided a description of the uncertainty associated with a variable, as well as annual variability.
Table 1

Definitions of variables based on PCD Risk model inputs that were used to create the hatchery templates. These variables will also be provided to the Delphi panel

Variable

Definition

Methods

Hatchery program

Name of the hatchery program.

Identified by the resource managers

Species and race

Species and race (e.g. Spring Chinook) of the program.

Identified by the resource managers

Release location(s)

Locations where fish are released to the natural environment.

Empirical information

Release date(s)

Dates fish are typically released.

Empirical information

Number of hatchery fish

Number of hatchery-origin fish released into the natural environment.

Empirical information

Mean, minimum, CV fork length

Mean fork length (mm), minimum fork length (mm), and coefficient of variation (CV) of fork length of hatchery-origin fish (SD/Mean). CV is expressed as a proportion.

Estimates from monitoring and evaluation programs

Hatchery fish survival rate

Survival rate of hatchery-origin fish in freshwater from release to an area below natural-origin fish interactions (e.g., mouth of river).

Estimates from monitoring and evaluation programs. Based primarily on PIT-tag data.

Hatchery fish residence time (days)

Average number of days interactions will occur.

Estimates from monitoring and evaluation programs. Calculated using travel time information (PIT-tag data) and the proportion of fish that residualize.

% residuals

Percentage of hatchery fish that do not migrate and take up residency.

Used data as available and best professional judgment based on other observations of similar situations.

Table 2

Definitions of variables based on PCD Risk model inputs that were used to create the NTTOC templates. These variables will also be provided to the Delphi panel

Variable

Definition

Methods

NTTOC

Non-target taxa of concern (NTTOC) population as defined by species, race and geographic location.

Identified by the resource managers

Number (minimum, mean, maximum)

Population estimates that indicate the likely range and mean of the NTTOC population at the time of hatchery fish release.

Estimates from monitoring and evaluation programs and life cycle models developed based on empirical data and values from literature

Age class, proportion in age class

Proportion of total fish abundance in each age class.

Estimates from monitoring and evaluation programs and life cycle models developed based on empirical data and values from literature

Mean, minimum, and CV fork length

Mean fork length (mm), minimum fork length (mm), and coefficient of variation (CV) of fork length of NTTOC fish (SD/Mean). CV is expressed as a proportion.

Estimates from monitoring and evaluation programs

Table 3

Definitions of variables based on PCD Risk inputs that were used to create the ecological interactions templates. These variables will also be provided to the Delphi panel

Variable

Definition

Methods

Percentage population overlap

Percentage of natural-origin population available for interaction because of spatial or temporal overlap. For example, 100% overlap occurs if hatchery fish are released above and before emigration of wild occurs and they totally overlap the entire wild fish distribution. Greater overlap increases opportunity for ecological interactions.

The sum of population overlap across life stages (fry, parr, smolt): ∑p*s*t; where p = the proportion of each life stage in the population, s = spatial overlap of hatchery and wild fish, and t = temporal overlap of hatchery and wild fish.

Percentage habitat complexity

Percentage of natural-origin population in overlap protected by habitat from competitive interactions (i.e., visual isolation). An environment with high visual isolation has many physical features (e.g., wood, rocks) that preclude fish from seeing and competing with each other. Greater complexity can reduce the opportunity for ecological interactions.

Estimate percentage for the pairing(s) using the best available information and incorporating best professional judgment. Based on these estimates, percentages were adjusted for pairings with less information.

Percentage habitat segregation

Percentage of hatchery-origin population in overlap that are excluded from competitive interactions because they occupy different habitats (e.g., deeper and faster water). Habitat segregation can reduce the opportunity for ecological interactions.

Estimate percentage for each type of NTTOC-to-hatchery program pairing(s) based on body size and habitat preference.

Probability dominance results in body weight loss

Probability that a fish that is dominated will have a ‘Body Weight Loss’ that is equal to 1 day of no feeding (e.g., the proportion of daily body weight loss that occurs from being dominated). Dominated fish may have reduced growth or be more susceptible to disease.

Hypothetical range developed using best professional judgment.

Dominance mode

Dominance mode for hatchery-origin fish competing with natural-origin fish. Dominance is a function of relative size of hatchery- and natural-origin fish and the relative behavioral dominance when they are the same size (e.g., aggression, prior residence). Defines the likely outcome (mode) of competitive interactions.

Dominance modes were selected based on literature review. Dominance modes 4 (minimum); 3 (most likely) and 2 (maximum) were used for all interactions.

Percentage of body weight loss that results in death

Percentage of body mass lost due to competitive encounters that will cause death. Threshold for death resulting from competitive interactions.

Based on a range from 11 fish species presented in Letcher et al. 1996.

Maximum daily encounters per hatchery fish

Maximum number of encounters, predatory or competitive, a hatchery-origin fish is allowed to have with natural-origin fish in 1 day (excluding fish that are protected or segregated by habitat). Encounters will increase as the capacity of the environment is filled. Increasing number of encountered will increase the opportunity for ecological interactions.

Calculated by: \( {\text{pHK}} * {\text{pWK}} * 10 \); where pHK = proportion of river carrying capacity used by hatchery fish (in the absence of wild fish), and pWK = proportion of the river carrying capacity used by wild fish (in the absence of hatchery fish) (from Busack et al. 2005).

Piscivory rate

Proportion of hatchery-origin fish that will feed on NTTOC fish. This proportion will be allowed to eat to capacity and the rest will not eat at all. (e.g., gut fullness of average hatchery-origin fish). Higher rate of piscivory may result in a greater mortality of NTTOC.

Based primarily on literature values, with best professional judgment used to fill data gaps.

Temperature

Average water temperature during the duration of interaction (°C). Used for bioenergetics input in the PCD Risk model, and available for the Delphi panel. In general, higher water temperatures will increase fish metabolism.

Mean, min, and max temperatures, derived from the nearest water temperature gauge to a release site, for the time period starting with release date and spanning the average residence time for the hatchery fish.

Disease mortality rate for fish with no dominance encounters

Hatchery induced disease mortality rate of natural-origin fish that have not been competitively dominated by hatchery-origin fish. Factors that influence mortality rate are pathogen density, pathogen virulence, and susceptibility of natural-origin fish. Mortality is likely to be delayed. Greater disease mortality rate may result in greater mortality of NTTOC.

Hypothetical range: 0 (minimum), 0 (most likely), and 0.0001 (maximum) developed using best professional judgment based on Busack et al. 2005.

Disease mortality rate for fish with max dominance

Hatchery induced disease mortality rate of natural-origin fish that have been competitively dominated up to the point of death by hatchery-origin fish. Factors that influence mortality rate are pathogen density, pathogen virulence, and susceptibility of natural-origin fish. Mortality is likely to be delayed. Greater disease mortality rate may result in greater mortality of NTTOC.

Hypothetical range: 0 (minimum), 0.01 (most likely), and 1.0 (maximum) developed using best professional judgment based on Busack et al. 2005.

The context of the risk assessment was biologically, spatially, and temporally defined to increase precision of and minimize confusion regarding the assessments. The assessment was constrained to include only the naturally produced component of the NTTOC. The spatial context of fish released in the tributaries to the Columbia River (e.g., Wenatchee, Methow, and Okanogan) included the tributary to the river mouth. Fish released into the Columbia River included the Columbia River from the point of release downstream to McNary Dam. This spatial context was selected because: it was relevant to the PCD Risk model (e.g., freshwater), it had the most information available, was subject to future testing, was most likely influenced by the hatchery programs of interest, and avoided areas where other hatchery programs would confound the risk assessment. Other potentially important ecological risks that might occur in portions of the migration corridor, estuary, and ocean were not considered in this assessment. The time frame of the risk assessment was constrained between 2013 and 2023. These years were selected because hatchery fish abundances are adjusted every decade as part of the public utility districts’ mitigation and this time period is the main focus of adaptive management decisions. Furthermore, most of the new or expanded hatchery programs will be implemented during this period.

Information was generated so that ecological risks could be assessed for each hatchery program and NTTOC interaction. This approach allows for the possibility to combine risk assessments for multiple hatchery programs. For instance, risks could be evaluated by hatchery species (e.g., all spring Chinook salmon hatchery programs combined), by funding entity (e.g., all programs funded by Grant PUD), by tributary (e.g., Wenatchee), by NTTOC (e.g., all programs that affect Methow cutthroat), or all programs combined. One of the uncertainties associated with assessing cumulative risks is how to combine risks from multiple programs. The Delphi panelists will be asked whether cumulative risks should be estimated through a) adding risk scores of individual programs, b) using the risk score of the highest individual program within a group of programs, c) multiplying each individual score by a weighting factor and then adding the scores together, or d) using some other approach. The panel will also be asked to recommend an approach for estimating uncertainty (e.g., variance) associated with estimating cumulative risks.

Results

Resource committees identified five species of NTTOC among 25 geographically distinct areas with containment objectives for each species (Fig. 2; Table 4). The NTTOC in the geographically distinct areas were often, but not always, genetically distinct from other populations. Acceptable impacts were qualitatively described as Very Low (<5%), Low (<10%), and Moderate (<41%) with respect to reduction in NTTOC status. Status was defined as abundance, spatial distribution, and size structure prior to 2013. An affect was deemed unacceptable if it exceeded the NTTOC threshold for any of the three status metrics (e.g., abundance, spatial distribution, and size structure). All of the species that are federally protected under ESA had very low acceptable impacts (spring Chinook salmon and steelhead trout). The only other species that had very low acceptable impact was Pacific lamprey. This species was afforded the most conservative impact level because of its importance to native tribes. Summer Chinook and sockeye salmon were assigned a low acceptable impact because they are relatively more abundant than other salmon. Cutthroat trout had the most liberal acceptable impact because they are relatively abundant, widely distributed, and provide lower fishery benefits than salmon.
Table 4

List of NTTOC species and containment objectives associated with ecological risk assessments of hatcheries in the Upper Columbia Watershed

NTTOC

Number of geographic locations

Containment objective (%)

Spring Chinook salmon

7

<5

Summer Chinook salmon

3

<10

Steelhead trout

8

<5

Sockeye salmon

2

<10

Cutthroat trout

2

<41

Pacific lamprey

3

<5

Data pertaining to the hatchery release template and NTTOC template are presented in Tables 5 and 6 respectively. Only means are presented in the tables to conserve space. Please contact the authors to obtain minimum and maximum values.
Table 5

Summary of hatchery release information for programs anticipated in 2013–2023 (means are presented, does not include minima or maxima to conserve space)

Watershed

Location

Hatchery

Species

Age

Date

Number

Fork length (mm)

CV

Survival

Residence time (days)

% residuals

Columbia

Mainstem

Priest Rapids

FAC

sub-yearling

9 June–26 June

7,700,000

96

6.88

0.76

10

1

Columbia

Mainstem

Chief Joseph

SPC

yearling

15 April–15 May

650,000

153

11

0.63

30

1

Columbia

Mainstem

Wells

STH

yearling

15 April–15 May

200,000

184

10.5

0.53

11

5

Columbia

Mainstem

Chief Joseph

SUC

sub-yearling

June

400,000

110

11

0.31

40

1

Columbia

Mainstem

Wells

SUC

sub-yearling

15 April–15 May

438,680

112

7

0.31

36

1

Columbia

Mainstem

Chief Joseph

SUC

yearling

15 April–15 May

500,000

176

11

0.63

30

1

Columbia

Mainstem

Chelan Falls

SUC

yearling

15 April–15 May

600,000

163

17

0.63

25

1

Columbia

Mainstem

Wells

SUC

yearling

15 April–15 May

301,056

158

8.3

0.63

27

1

Methow

Beaver

Winthrop, Willard, Cascade

COH

yearling

1 May–31 May

27,000

130

7

0.97

3

<1

Methow

Chewuch

Winthrop, Willard, Cascade

COH

yearling

1 May–31 May

163,000

130

7

0.97

3

<1

Methow

Chewuch

Methow

SPC

yearling

15 April–15 May

225,000

132

9.4

0.97

4

2

Methow

Methow

Winthrop, Willard, Cascade

COH

yearling

1 May–31 May

254,000

130

7

0.97

3

<1

Methow

Methow

Methow

SPC

yearling

15 April–15 May

225,000

132

9.4

0.97

4

2

Methow

Methow

Winthrop

SPC

Yearling

11 April–20 April

400,000

120

7.5

0.97

7

4

Methow

Methow

Wells

STH

Yearling

15 April–15 May

100,000

184

10.5

0.95

1

5

Methow

Methow

Eastbank, Carlton Pond

SUC

Yearling

15 April–15 May

374,398

148

14.8

0.97

13

2

Methow

Methow-Winthrop

Winthrop

STH

2 year-old

22 April–7 May

50,000

180

15.2

0.96

4

3

Methow

Methow-Winthrop

Winthrop

STH

Yearling

22 April–7 May

50,000

158

15.2

0.85

4

5

Methow

Twisp

Winthrop, Willard, Cascade

COH

Yearling

1 May–31 May

142,857

130

7

0.97

3

<1

Methow

Twisp

Methow

SPC

Yearling

15 April–15 May

100,000

133

7.9

0.97

4

2

Methow

Twisp

Wells

STH

Yearling

15 April–15 May

47,571

178

12.5

0.95

4

5

Okanogan

Okanogan

Shuswap

SOK

Fry

Late May–Early June

4,550,000

49

8.8

0.23

365

<1

Okanogan

Okanogan

Chief Joseph

SPC

Yearling

15 April–15 May

200,000

153

11

0.97

5

2

Okanogan

Okanogan

Wells

STH

Yearling

15 April–15 May

80,000

184

10.50

0.95

2

5

Okanogan

Okanogan

Chief Joseph

SUC

sub-yearling

June

300,000

103

11

0.97

5

2

Okanogan

Okanogan

Chief Joseph, Omak Pond

SUC

Yearling

15 April–15 May

400,000

176

11

0.97

5

2

Okanogan

Okanogan

Chief Joseph, Omak Pond

SUC

Yearling

15 April–15 May

400,000

176

11

0.97

5

2

Okanogan

Okanogan

Chief Joseph, Riverside Pond

SUC

Yearling

15 April–15 May

200,000

129

11

0.97

6

2

Okanogan

Omak

Chief Joseph

SPC

Yearling

15 April–15 May

50,000

153

11

0.97

5

2

Okanogan

Omak

Wells

STH

Yearling

15 April–15 May

20,000

184

10.50

0.95

2

5

Okanogan

Okanogan

Eastbank, Similkameen Pond

SUC

Yearling

15 April–15 May

555,494

129

11

0.97

13

2

Okanogan

Okanogan

Eastbank, Bonaparte Pond

SUC

Yearling

15 April–15 May

79,156

129

11

0.97

6

2

Okanogan

Salmon

Chief Joseph

SPC

Yearling

15 April–15 May

50,000

153

11

0.97

5

2

Okanogan

Salmon

Wells

STH

Yearling

15 April–15 May

20,000

184

10.50

0.95

2

5

Wenatchee

Chiwawa

Willard, Cascade

COH

Yearling

1 May–31 May

190,000

130

7

0.65

3

<1

Wenatchee

Chiwawa

Eastbank, Chiwawa Pond

SPC

yearling

15 April–15 May

298,000

140

6.9

0.65

3

2

Wenatchee

Chiwawa

Eastbank

STH

yearling

15 April–15 May

37,200

192

13.1

0.90

6

10

Wenatchee

Chumstick

Willard, Cascade

COH

yearling

1 May–31 May

35,000

130

7

0.97

3

<1

Wenatchee

Icicle

Willard, Cascade

COH

yearling

15–Apr

126,000

130

7

0.97

3

<1

Wenatchee

Icicle

Leavenworth

SPC

yearling

15 April–28 April

1,200,000

128

12.3

0.97

2

2

Wenatchee

Little Wenatchee

Willard, Cascade

COH

yearling

1 May–31 May

65,143

130

7

0.65

3

<1

Wenatchee

Nason

Willard, Cascade

COH

yearling

1 May–31 May

185,000

130

7

0.65

3

<1

Wenatchee

Nason

Eastbank, Nason Pond

SPC

yearling

15 April–15 May

250,000

140

6.9

0.65

3

2

Wenatchee

Nason

Eastbank

STH

yearling

15 April–15 May

69, 800

192

13.1

0.90

4

10

Wenatchee

Wenatchee

Wenatchee Netpens

SOK

yearling

31-Oct

281,000

145

8.5

0.59

192

5

Wenatchee

Wenatchee (lower)

Eastbank

STH

yearling

15 April–15 May

93,000

192

13.1

0.9

6

10

Wenatchee

Wenatchee (upper)

Eastbank

STH

yearling

15 April–15 May

200,000

192

13.1

0.9

6

10

Wenatchee

Wenatchee

Eastbank

SUC

yearling

15 April–15 May

719,667

149

16.7

0.97

2

2

Wenatchee

White

Willard, Cascade

COH

yearling

1 May–31 May

114,000

130

7

0.65

3

<1

Wenatchee

White

Eastbank, White Pond

SPC

yearling

15 April–15 May

150,000

140

6.9

0.65

3

2

COH Coho salmon, FAC Fall Chinook salmon, SOK Sockeye salmon, SPC Spring Chinook salmon, STH Summer Steelhead, SUC Summer Chinook salmon

Table 6

Summary of NTTOC information. Data are means and do not include minima or maxima do to space constraints

Species

Location

Population estimate

L1

L2

L3

L4

L5

C1

C2

C3

C4

C5

P1

P2

P3

P4

P5

CUT

Methow

ND

144

174

218

393

437

9.8

11.7

13.7

10.8

10.0

ND

ND

ND

ND

ND

CUT

Wenatchee

ND

144

174

218

393

437

9.8

11.7

13.7

10.8

10.0

ND

ND

ND

ND

ND

LAM

Methow

ND

123

135

   

13.5

8.8

   

0.968

0.031

   

LAM

Okanogan

ND

123

135

   

13.5

8.8

   

0.968

0.031

   

LAM

Wenatchee

ND

123

135

   

13.5

8.8

   

0.968

0.031

   

SOK

Okanogan

4,599,270

31

139

   

11.4

13.5

   

0.551

0.484

   

SOK

Wenatchee

3,019,789

31

86

128

150

 

11.4

7.9

6.2

8.0

 

0.497

0.476

0.024

0.003

 

SPC

Chewuch- Methow

264,811

38

96

   

10.4

8.6

   

0.968

0.032

   

SPC

Methow

571,608

38

100

   

10.4

9.1

   

0.969

0.031

   

SPC

Twisp-Methow

139,831

38

96

   

10.4

8.6

   

0.884

0.116

   

SPC

Chiwawa-Wenatchee

1,022,141

38

93.0

   

7.4

7.7

   

0.919

0.082

   

SPC

Little Wenatchee-Wenatchee

93,132

45

104

   

15.9

8.6

   

0.916

0.082

   

SPC

Nason-Wenatchee

450,132

47

93

   

16.0

8.0

   

0.914

0.086

   

SPC

White-Wenatchee

111,402

43

100

   

14.0

7.0

   

0.914

0.086

   

SUC

Methow

378,973

41

96

   

16.2

11.9

   

0.995

0.005

   

SUC

Okanogan

1,841,855

41

98

   

5.0

9.4

   

0.998

0.002

   

SUC

Wenatchee

3,910,598

41

98

   

5.0

9.4

   

0.999

0.001

   

STH

Chewuch- Methow

34,730

105

161

178

179

 

22.2

9.7

11.2

7.3

 

0.599

0.340

0.061

0.001

 

STH

Methow

198,791

126

170

179

172

 

18.8

11.7

13.1

10.4

 

0.599

0.340

0.061

0.001

 

STH

Twisp-Methow

71,410

105

161

178

179

 

22.2

9.7

11.2

7.3

 

0.599

0.340

0.061

0.001

 

STH

Okanogan

236,016

128

166

192

223

 

13.1

4.9

3.8

4.8

 

0.599

0.340

0.061

0.001

 

STH

Omak- Okanogan

14,357

128

166

192

223

 

13.1

4.9

3.8

4.8

 

0.599

0.340

0.061

0.001

 

STH

Chiwawa-Wenatchee

15,156

123

163

192

242

 

12.8

5.2

4.4

10.8

 

0.599

0.340

0.060

0.001

 

STH

Nason-Wenatchee

51,150

128

166

192

223

 

13.1

4.9

3.8

4.8

 

0.599

0.340

0.060

0.001

 

STH

Wenatchee

74,049

128

166

192

223

 

13.1

4.9

3.8

4.8

 

0.599

0.340

0.060

0.001

 

CUT cutthroat trout, LAM Pacific lamprey, SOK Sockeye salmon, SPC Spring Chinook salmon, SUC Summer Chinook salmon, and STH Summer Steelhead. L1 stands for mean fork length of age class 1, C1 stands for the coefficient of variation of length of age class 1, and P1 stands for the proportion of the NTTOC abundance that are in age class 1. Age classes range from 1 to 5. ND No Data for an NTTOC

Interactions between a hatchery program and an NTTOC are described by a suite of variables that addresses competition, predation, bioenergetics, and disease. Given the large number of programs (N = 50) and NTTOC (N = 25) within the distinct geographic areas covered by this study, a suite of 527 interactions are being evaluated in this analysis. Interactions include NTTOC of different species, as well as conspecific NTTOC that exist in locations that are not the target of a given hatchery program. Ecological interactions between hatchery and naturally produced fish of the target species are not the subject of this risk assessment because a rigorous monitoring program is currently being implemented to determine impacts of the hatchery program. Examples of interactions data (defined in Table 3) that range across three watersheds and different NTTOC and hatchery program species are presented in Table 7. Similar interactions data sets will be used for each of the hatchery target and NTTOC pairings. Interactions data include variables with the minimum, maximum, and most likely values to describe the interaction of habitat, competition, predation, disease and physiology on the survival of NTTOC.
Table 7

Examples of NTTOC and hatchery program interaction data for the Delphi and PCD Risk modeling processes. Definitions for variables are in Table 3

NTTOC:

Okanogan summer steelhead

Methow spring Chinook

Wenatchee summer steelhead

Program:

Similkameen summer chinook

Methow summer steelhead

Wenatchee sockeye

Variables

Min

Most likely

Max

Min

Most likely

Max

Min

Most likely

Max

Percentage habitat complexity

5

15

30

15

25

35

7

17

27

Percentage population overlap

44

0.1

54

Percentage habitat segregation

30

50

70

30

60

90

30

60

90

Probability dominance results in body weight loss

0.00

0.05

0.10

0.00

0.05

0.10

0.00

0.05

0.10

Dominance mode

4

3

2

4

3

2

4

3

2

Percentage of body weight loss that results in death

46

50

74

46

50

74

46

50

74

Maximum daily encounters per hatchery fish

0.07

0.11

0.17

0.02

0.04

0.08

0.10

0.22

0.47

Piscivory rate

0.0000

0.0000

0.0001

0.0000

0.0023

0.0200

0.0000

0.0023

0.0200

Temperature

6.8

13.0

23.8

5.0

7.1

10.3

0.1

6.6

20.3

Disease mortality rate for fish with no dominance encounters

0.0000

0.0000

0.0001

0.0000

0.0000

0.0001

0.0000

0.0000

0.0001

Disease mortality rate for fish with max dominance

0.00

0.01

1.00

0.00

0.01

1.00

0.00

0.01

1.00

Discussion

The approach that we are using in the upper Columbia Watershed provides a flexible approach to ecological risk assessment of hatchery programs. It has the advantages of being standardized (e.g., use of rules, equations, and standards), transparent (e.g., data available to all, documentation of assumptions), comparative, and hierarchical. However, the approach is labor intensive and subject to many assumptions. The many assumptions that were required to help inform the risk assessment were uncomfortable for many of the participants, as is often the case in assessments when empirical data are not available. We found that the work required to populate risk assessment templates increased dramatically with larger numbers of hatchery programs, NTTOC, experts, and amount of unconsolidated data. When relevant quantitative data were scarce, then most of the information needed to generate the risk assessment was from local experts. This process can often occur relatively quickly. In contrast, when rich, unconsolidated data sets existed, substantial work was required to assemble data and determine how to manipulate it in a way that can inform a risk assessment. We found that development of simple equations or rules could improve standardization of model inputs and also decrease the time necessary to populate information templates (e.g., Regan et al. 2004).

There was a limited amount of quantitative information that was available to populate risk assessment templates about cutthroat trout and Pacific lamprey. Most previous data collection efforts in the upper Columbia Watershed focused on salmon and steelhead. The PCD Risk 1 model was designed for salmonids, so Pacific lamprey cannot be included in the modeled risk assessment. Furthermore, the lack of information may limit or preclude including cutthroat in the modeling process. However, for both species we will ask the Delphi panel to perform risk assessments on these NTTOC to the extent possible, and make recommendations for future risk assessment of these species. This is likely to be the case in other locations too, which means that risk assessments on these or other data-poor species will likely depend heavily upon expert opinion. However, the exercise of discovering data gaps can help to prioritize future data collection efforts by multiple organizations that may be interested in obtaining the information.

The risk assessment approach described in this paper limited the spatial, temporal, taxonomic, and ecological interaction factors that could occur from hatchery programs. As such, some risks such as indirect predation to NTTOC that may occur in the estuary or risks to other species such as minnows, suckers, and sculpins will not be addressed in this assessment. Furthermore, the PCD Risk 1 model was designed to assess affects associated with predation, competition, and disease in freshwater; and does not include mechanisms such as indirect predation nor does it specifically address exploitative competition.

Based on filling out data templates with local experts, we found that it was important to carefully define spatial, temporal, and biological context to obtain cohesive data and minimize variation between information providers. Hatchery programs can be perceived to interact with NTTOC across wide ranges of time, space, and life stages. Without the context defined, participants would generate information across various spatial, temporal, and biological scales. For instance, one participant would generate information for a hatchery program from a release location downstream to a tributary mouth, whereas another would extend the information into the Columbia River. Another example may occur when a participant generates information for current hatchery programs while others generate information for future hatchery programs. These differences could have substantial influences on risk assessment results and interpretation for management decisions; therefore, it is wise to define the context very early in the process.

Next steps

Experts that have been identified by the resource committees will be invited to participate in a Delphi approach to assess risks (Fig. 2). These Delphi panelists will be provided with the information templates populated by the local experts so that all of the panelists in the Delphi process have access to the same, most relevant information. In addition, the opportunity to ask clarifying questions will be provided. Delphi panelists will independently estimate the proportion of a non-target taxa population that will be affected by each individual hatchery program (Pearsons and Hopley 1999). Impacts will be described as the impact to abundance, size at age, and spatial distribution of an NTTOC. These estimates will be averaged, a measure of dispersion calculated, and subsequently compared to the acceptable hatchery impact levels that were determined previously by resource committees. Measures of dispersion will be used to estimate the scientific uncertainty associated with risk estimates. It is expected that pairings of hatchery target fish and NTTOC for which little data exist will produce high levels of dispersion. Furthermore, estimates of impacts from each hatchery program will be combined together to generate an estimate of cumulative impact to each non-target taxa.

The PCD Risk 1 model will also be populated with the same information templates that will be provided to the Delphi panelists. Modeled results will provide the opportunity to compare risk assessments between expert opinion and the model such as has been described by McCarthy et al. (2004). If results are correlated, then future changes to programs or improvement in data templates could be assessed using the model instead of reconvening Delphi panelists. In addition, risks could also be compared to PCD Risk 1 model results that have been conducted on hatcheries in western Washington and perhaps evaluate bias associated with expert opinion (McCarthy et al. 2004). Furthermore, the model allows for opportunities to assess various risk reduction strategies by conducting multiple model runs with different inputs (Pearsons and Busack 2011).

After the risk assessment is complete, results will be used by managers to reduce risks if necessary, modify monitoring and evaluation plans, and adaptively manage programs (Fig. 2). For example, if risks are unacceptably high and the scientific uncertainty is acceptable; then modification of the hatchery program might occur (Pearsons and Hopley 1999; Ham and Pearsons 2001). Alternatively, if risks appear sufficiently high, but scientific uncertainty is high, then additional monitoring and evaluation or studies may be necessary to assess risk at the desired level of certainty.

Acknowledgements

We thank the members of the Wells Hatchery Committee, Rocky Reach Hatchery Committee, and Priest Rapids Coordinating Committees Hatchery Sub-Committee for their work on selection of NTTOC, containment objectives, and regional experts. We also thank the many local experts who helped with providing data and opinions about interactions. These experts included: John Arterburn, Charles Snow, John Crandall, Kirk Truscott, and David Hopkins. Ali Wick and Carmen Andonaegui were instrumental in helping facilitate discussions as well as recording and compiling information. Chad Herring created Fig. 1. Public Utility District Number 1 of Chelan County, Public Utility District Number 1 of Douglas County, and Public Utility District Number 2 of Grant County helped fund this effort.

Copyright information

© Springer Science+Business Media B.V. 2011