This project continues to make steady progress in developing state-of-the-art procedures for monitoring genetic diversity and cost-effective applications of Genetic Stock Identification (GSI) and Parental Based Tagging (PBT). It contributes practical advice for fisheries management and addresses a number of critical uncertainties. Application of PBT and GSI has improved the accuracy of stock-specific estimates of harvests throughout the Columbia River Basin and of abundance at multiple dams. Project results are used extensively by the U.S. v. Oregon Technical Advisory Committee for in-season harvest monitoring as well as post-season run reconstruction for multiple species of salmon and steelhead. The record of primary publications and annual reports is excellent.

ISRP suggestions and comments in the sections below should be considered in future proposals.

Q1: Clearly defined objectives and outcomes

Three genetic assessment projects reviewed separately in 2018 have been merged into a single project with six objectives. The first four objectives listed in the Progress to Date section of the proposal correspond to previous research under BPA project 200890700: 1) discover and evaluate SNP markers (for multiple salmonid species and white sturgeon), 2) expand and create genetic baselines (to support PBT and GSI analyses), 3) implement genetic monitoring (PBT/GSI) programs for mainstem Chinook salmon, sockeye salmon and steelhead fisheries, and 4) apply genetic monitoring (PBT/GSI) to Chinook, sockeye and coho salmon and steelhead passing Bonneville Dam. The fifth objective, to characterize adaptive genetic variation associated with environment, landscape, and phenotypic traits, corresponds to research previously undertaken in BPA project 200900500. The sixth objective, to characterize genetic diversity and structure of white sturgeon in the Columbia River Basin, corresponds to research previously undertaken by BPA project 200850400.

Section 3 of the proposal lists additional quantitative objectives, implementation objectives, research questions and predictions under each of the six objective headings. We found these distinctions to be vague, repetitious, and insufficiently quantitative to clearly specify the desired outcomes for the proposal’s five-year time period or to evaluate future progress. In short, these objectives do not meet SMART criteria and should be revised in future proposals to clearly document the project’s objectives and hypotheses. That said, we understand the challenge of documenting quantitative objectives for this continuing project given its complex mix of exploratory and applied science. One suggestion for meeting the SMART criteria is to create summary plots to show past trends and future projections for quantitative deliverables such as the number of single nucleotide polymorphism (SNP) markers developed and/or selected for application, number of populations incorporated in baselines, number of GSI reporting groups, number of adaptive traits mapped to genotypes identifiable with SNPs, specific fisheries and dams monitored, etc. These summary plots could be based on data provided in various sections of the annual report for 2019.

No specific milestones or timelines are associated with any of the objectives, but it is clear from the proposal and annual report for 2019 that the project is expected to continue for the foreseeable future. PBT requires annual collection and genotyping of tissue samples from hatchery broodstock, and GSI of salmonids harvested in mainstem fisheries or passing Bonneville Dam requires annual collection and analysis of mixture samples. The genomic research programs on adaptive traits and white sturgeon are making steady progress, but they are technically challenging and will likely take many years to deliver their full potential. In any case, it is important to indicate the expected yearly (or perhaps 5-year) quantitative outcomes for each objective that will be used for evaluating progress in the project’s internal adaptive management cycle.

Q2: Methods

The proposal provides only a brief overview of the approaches being followed for each objective. Much more detail is provided in the lengthy annual report for 2019 (289 pages with many links and references). It has not been possible for the ISRP to examine these methods in detail during the current review process given the scope and complexity of this aggregate project, which spans many years, objectives, covers multiple species, populations, and fisheries, and employs technically sophisticated laboratory and analytical procedures.

A major strength of this project is the successful development of cost-effective methods (e.g., GT-seq) to screen genotypes at hundreds of SNPs for application to PBT and GSI. Another important innovation of this project is the combined application of PBT and GSI to improve estimates of stock composition of fish passing Bonneville Dam and harvested in mainstem fisheries. Sampling and analytical protocols are clearly documented in the annual report for 2019. Standard methods are being used for statistical analysis, and confidence intervals or significance test probabilities are provided to support most conclusions. The proponents also use state-of-the-art techniques that were peer-reviewed in their primary publications, and hence, are scientifically appropriate.

The proponents have been diligent and creative in finding ways to minimize bias that can arise in analyzing stock composition of mixtures if weighted stratified random subsamples are improperly weighted. They developed, published, and applied a bias-correction procedure for incorporating new data on PBT detections of hatchery fish that are not adipose-clipped and previously had been assumed to be wild (about 20% of Chinook and 8% of steelhead from 2014-2018, Hargrove et al. 2021), while simultaneously adjusting the overall proportions of hatchery and wild populations based on the GSI results for wild fish. Presumably, a larger proportion (e.g., 0-18% at Bonneville Dam in the fall of 2019) of hatchery fish are “missed” by PBT because of incomplete PBT baseline sampling, but they are not counted as wild fish because adipose clips indicate their hatchery provenance.

White sturgeon present special challenges for genetic analysis because of their polyploid ancestry. Despite this, the proponents are making steady progress in identifying SNPs, demonstrating functional tetraploidy and Mendelian inheritance patterns, and in developing ways to modify analytical procedures when using software programs designed to analyze diploid species.

The proponents have fully addressed previous ISRP requests (2018) to justify the choice of assignment thresholds (they no longer set assignment thresholds), and to compare the relative accuracy of counting individual fish assignments versus estimating mixing proportion parameters without attempting to assign individuals.

Q3: Provisions for M&E

The proponents use section 6 of the proposal (Potential Confounding Factors) to emphasize how this project is uniquely positioned to document widespread influences of climate change on abundance, run-timing, and genetic diversity of salmonids of the Columbia River. As PBT and GSI baselines are updated and expanded, they provide successive “snap shots” of biodiversity within a number of target species. These data could be used to augment High Level Indicators for monitoring the state of biodiversity within the basin. The proponents also plan to routinely sample populations in extreme temperature sites and expect to use SNP markers in combination with genomic and physiological analysis to predict and track the populations’ adaptability to climate and landscape changes.

Section 6, however, is primarily intended to elicit discussion of factors that might hamper progress in meeting objectives. One such factor is the difficulty of getting sufficiently representative samples of various stocks passing Bonneville Dam due to restrictions placed on trap operations there by USACE and NMFS. Sampling opportunities at the Bonneville Dam trap may become increasingly restrictive due to rising temperatures and increasing numbers of shad.

In future proposals, the ISRP would like to see more explanation of the practical limits to GSI resolution with SNP baselines. We understand that GSI resolution will be constrained by the amount and stability of differentiation in allele frequencies among populations (“population structure”) determined by the historical balance of genetic drift and gene flow. What then is the practical limit to GSI resolution that can be reasonably expected despite increasing the number of SNPs or microhaplotypes examined, and regularly updating baselines given observed levels of year-to-year variability associated with random demographic effects (e.g., genetic drift within natural spawning subpopulations and baseline sampling errors)? How are factors that reduce gene flow among populations, and thus maintain differentiation among populations in neutral traits, expected to change in the face widespread hatchery propagation (more straying and transplantation) versus reduced abundance and fragmentation of natural metapopulations, and range contractions due to climate change? Are adaptive traits expected to be more robust to the confounding effects of sampling variation, genetic drift, and gene flow, so that GSI baselines might need be updated less often?

Section 5 (Project Evaluation and Adjustment Process) of the proposal is too brief. In future proposals, the ISRP would like more explanation of the process by which the proponents allocate effort and resources among objectives such as finding new SNPs and updating and expanding baselines. Is there a regularly scheduled decision process to determine, for example, when more SNPs need to be added to panels, which SNPs to include, which stocks to include in baselines, how often to resample them, and how large the samples should be?

Q4: Results - benefits to fish and wildlife

Objectives 1-4: This project has developed the genetic baselines (i.e., SNP) and technical capacity to routinely provide timely estimates of stock composition of mixed stock harvests of hatchery and natural origin Chinook salmon, sockeye salmon and steelhead in the Columbia River, and to estimate the run timing and abundance of genetic stocks passing Bonneville Dam. The practical management application of the project results is impressive. SNP panels are also being developed to identify the origin of coho salmon, white sturgeon, and several lamprey species. The SNP panels have been especially useful for PBT and measuring reproductive success in other pedigree studies.

Sample rates for Chinook salmon, sockeye salmon, and steelhead are often lower than desired due to restrictions imposed by USACE and NMFS on sampling at the Bonneville trapping facility. Another continuing challenge is that GSI reporting groups based on the genetic differentiation of populations, which provide the most accurate estimate of stock composition, are not identical to the management units of most interest to fisheries managers. The proponents are continuing to work with fisheries managers to explore how to best incorporate genetic monitoring results with more traditional monitoring and tagging programs.

In principle, PBT and GSI based on SNP markers are now sufficiently developed that they could replace some (but not all) functions of the coded wire tag (CWT) program for hatchery fish. However, replacing the CWT program would require continued annual genotyping of hatchery broodstock, fish passing Bonneville Dam, and harvested fish.

Objective 5: The project has made steady progress in developing techniques to identify and monitor adaptive divergence among specific environments in a number of species within the Columbia Basin. Recent results indicate that precipitation, elevation, and temperature are among the most important environmental factors driving adaptive divergence in salmonids. Multiple studies are underway to investigate the genetic basis for run-timing, age-at-maturity, disease resistance, and thermal adaptation. Candidate genes for several of these traits have been identified in both steelhead and Chinook. SNP markers from these regions are now being incorporated into standard genotyping panels to cost-effectively scan large numbers of individual fish for genetic variation in adaptive traits. For example, genotypes for temperature tolerance could be screened to identify broodstock for reintroduction efforts. Understanding the genetic mechanisms and potential for adaptation to changes in precipitation, temperature, and other environmental factors will help to guide long-term conservation policies for salmonid populations in the face of climate change.

Objective 6: Genetic analyses of SNP variation in white sturgeon have corroborated earlier surveys of variation in microsatellite markers, with both analyses showing little genetic differentiation among white sturgeon inhabiting different impoundments of the Middle Columbia region. Broodstocks from the Yakama hatchery that are used for supplementation in the Middle Columbia River were genetically similar to the wild populations from which they are derived, which justifies their continued use. The recent increase in SNP markers from 117 to 325 will improve capabilities for resolving population structure and PBT. The new genotyping approach provides an estimate of ploidy level and enables evaluation of the rate of spontaneous autopolyploidy. A large number of samples representing various age classes of white sturgeon from all sections of the Columbia and Snake rivers are currently being genotyped to investigate genetic structure and the frequency of ploidy levels within each section. The project is also making steady progress towards assembling a draft genome for white sturgeon, as a first step in discovering a genetic marker to non-lethally distinguish males from females at any age.

1. Objectives

The previous proposal in Taurus lists three objectives with corresponding hypotheses: (a) evaluate population differentiation and migration (gene flow) among reservoirs; (b) determine relatedness, mean productivity, and number of spawners within each reservoir; and (c) characterize broodstock by identifying origins (i.e., reservoir or population) and degree of relatedness among candidate broodstock fish. More recently, the proponents have developed two additional objectives involving whole genome sequencing to: (d) quantify adaptive variation within and among the sampled populations, and (e) search for a sex-linked marker that can be used to sex sturgeon at any age. Hypotheses were not developed for the two new objectives.

Overall, the goals and hypotheses are well defined, measurable, testable, and relevant to the Fish and Wildlife Program. No specific milestones or timelines are associated with any of the objectives, but it is clear from the narrative and project reports that steady progress is being achieved.

2. Methods

White sturgeon present challenges for genetic analysis because of their tetraploid ancestry, longevity, and the difficulty of obtaining adequate sample sizes. The proponents are to be commended for their skills in developing cost-effective methods (e.g., GT-seq) to screen genotypes at hundreds of DNA loci (SNPs) for application in parent-based tagging and other tools of genetic analysis. Sampling and analytical protocols seem appropriate and are clearly documented in Annual Reports, which include useful links to onlineresourcemonitoring.org websites.

Standard methods are being used and statistically reliable results are being produced. Measures of effect size (e.g., confidence intervals) are provided to support most conclusions. An exception is Figure 3 (i.e., comparing haplotype frequencies between pooled samples representing Snake and Lower Columbia stocks) and the AMOVA results (p.9) in the 2017 Annual Report that lack statistical tests and sample sizes, so the strength of the evidence for differences among stocks is unclear. Questions also remain regarding sample sizes needed to adequately characterize populations in individual reservoirs, and to clarify the influence of sturgeon movements among reservoirs, especially the downstream movement of fry and young-of-year fish.

Two very recent papers by C.D. Waters and colleagues seem highly relevant to identifying and quantifying adaptive variation among white sturgeon populations (objective d). In case these papers are unfamiliar to the proponents, the references are:

Waters C.D., Hard J.J., Brieuc M.S.O., Fast D.E., Warheit K.I., Knudsen C.M., Bosch W.J., and Naish K.A. 2018. Genomewide association analyses of fitness traits in captive-reared Chinook salmon:         Applications in evaluating conservation strategies. Evolutionary Applications 11: 853-868.

Brieuc, M.S.O., Waters C.D., Drinan D.P., and Naish K.A. 2018. A practical introduction to Random Forest for genetic association studies in ecology and evolution. Molecular Ecology Resources 18: 755-766.

3. Results

A major strength of this project is the successful development of a novel suite of 117 SNP markers and the use of the GT-seq high throughput sequencing method. Prior to 2017, genetic monitoring of white sturgeon had been investigated with a small published set of 13 microsatellite markers. The set of SNP markers developed in 2017 is currently being expanded, which will likely improve our knowledge of population structure and facilitate the application of parent-based tagging and other genetic tools.

It is reassuring that preliminary analyses of population structure based on the new SNP markers are corroborating previous results based on microsatellite loci. The metapopulation structure of white sturgeon is characterized by low genetic diversity and small effective population size. The low diversity and relatively small individual pairwise genetic distances likely stem from small numbers of families and/or few effective spawners within several impoundments. Broodstocks from the Yakama hatchery that are used for supplementation in the Middle Columbia River are genetically similar to the wild populations from which they are derived, which justifies their continued use. Gene flow has been highly restricted by barriers in the hydrosystem. The strong pattern of isolation by distance evident in genetic data supports the hypothesis that white sturgeon movement is largely downstream through the hydrosystem. In contrast, white sturgeon in the Lower Fraser and lower Columbia appear genetically similar; these populations have unimpeded access to the Pacific Ocean and may continue to intermix to some degree.

The proponents are now undertaking to sequence the genome of a single female white sturgeon, primarily to develop a practical genetic technique for non-lethally differentiating males and females at any age. The project is also making significant progress toward developing a parentage database that would allow juvenile fish to be assigned to parents spawning in the wild or to identify hatchery-origin fish without the need for other types of identifying marks.

Development of a reliable suite of genetic markers for white sturgeon is a critical step in understanding ecological and evolutionary dynamics of populations in the Columbia Basin and for future genetic monitoring of stocks in the basin. The ultimate utility of this program rests on its ability to track a population of long-lived fish over many years to fully ascertain some key population parameters. Translocation (i.e., stocking) of hatchery juveniles into upper Columbia River areas might be a useful strategy to compensate for the lack of recruitment or loss of productivity and diversity in blocked areas. But first, it will be necessary to examine the genetic population structure of white sturgeon in the impoundments between Priest Rapids Dam and Grand Coulee Dam and to investigate the conservation implications of possible introgression between remnant wild populations and hatchery fish derived from non-local broodstocks.

Collaboration and sharing of information among partners appear excellent. Annual reporting has been timely with satisfactory detail. However, in future reports, the ISRP would like to see more interpretation of results relative to stated objectives and hypotheses, and more discussion of how data from this project will be used in adaptive management strategies for population management and supplemental stocking programs.

4. 2017 Research Plan uncertainties validation

There is a mismatch between the list/numbering in the Research Narrative versus the Uncertainties Database or 2017 Research Plan. The ISRP concludes that this project indirectly (rather than directly) addresses the three critical uncertainties (CUs) listed by the proponent.

The Uncertainties Database indicates that six questions are being directly addressed by the project, but of these, it seems that only two (F1.4 and F3.2) are directly addressed, and the others are indirectly addressed.

As stated in the 2015 CUs review, "it is unclear how results obtained thus far and those anticipated in the future will be used help make decisions needed for sturgeon hatcheries and stock recovery. A critical uncertainty for these fish is: are observed genetic differences among fish from different reservoir pools indicative of adaptations to specific areas or just the result of recent segregation of putative stocks and thus not of great importance to managers?"

This is a very well-written and well-justified proposal that meets scientific criteria, though some minor design and method details are absent. The publication track record and overall participation in communication of results from this Principal Investigator and team is very commendable and bodes well for the project being able to achieve its stated objectives and reach a wide audience in the basin. The ISRP recommends to the sponsor: • Formal and informal communication of results and analyses – in spite of the Principal Investigator’s track record, communication of the data (and standardization) with other labs/programs in the Columbia River Basin should be planned and articulated. • Expansion on design details to include number and location of sampling sites, as well as intensity of sampling within a site for sufficient analytical power. • Control/reference approach for calibrating and verifying GSI estimates for runs passing Bonneville Dam.

An adequate response will address several ISRP recommendations found in Sections 1 and 2 and listed below. 1. Provide a more robust technical justification for the proposed genetic analyses, including: (a) a review of past and ongoing genetic assessments of white sturgeon in the Columbia River Basin, as well as relevant studies in other rivers outside the Basin, (b) justification for use of microsatellite DNA versus other techniques (SNPs, mitochondrial DNA), (c) an evaluation of potential pitfalls in meaningful interpretation of the results, for example, if migration of individuals interferes with distinguishing fish from specific reservoirs, (d) the rationale for analysis of fish of different ages, and (e) a discussion of the sample sizes required to make interpretations under various assumptions about the breeding population sizes in particular reservoirs. 2. Provide specific details on the relations and coordination between the proposed project and other white sturgeon projects in the Basin: (a) #1986-050-000 (types of data collected, number of suitable fish, and how data for the proponent’s Objective (2) will complement the current effort); (b) the Yakima Sturgeon Management Project (# 2008-455-00), and (c) other projects (e.g. Kootenai Tribe’s genetics work) or additional avenues available to collect sturgeon samples. 3. Include the necessary first step that is missing from the proposal, that is, optimizing the amplification of DNA, genotyping the microsatellite DNA in Columbia River white sturgeon, and confirming that sufficient genetic variation can be detected. 4. For both Objectives (1) and (2), provide a better description of the samples available for analysis and provide evidence that the sampling of fish is consistent with a robust analysis of the genotypic data. 5. For Objective (1), clarify how sampling of various fish will provide complete coverage of potential contributing populations and that the approach to the analysis will be able to sort out migrant individuals. 6. For Objective (2), the experimental design (parentage analysis) and statistical analysis (relatedness analysis) appear to be for two different objectives. Provide an experimental design and statistical analysis for both. Discuss potential limitations of the proposed designs and analyses, for example, in the case that analyses are based only on young-of-the-year genotypes. 7. Provide justification that proposed sample sizes (up to 1000 fish per year for ten years) will be sufficient to yield useful results and interpretations of results. 8. To each of the objectives, add a sufficiently detailed description of potential outcomes of uses of project data that will result in measurable benefits to Columbia Basin fish and wildlife, more specifically white sturgeon. This project has a reasonable likelihood of aiding management and conservation of white sturgeon in these lower Columbia River reservoirs. The Sturgeon Strategic and Hatchery Master Plan document that is currently being prepared should guide this project. It is somewhat surprising that the proponents did not make note of the sturgeon workshop conducted under the auspices of Project #2007-155-00, scheduled to take place the first week of December 2009. The work conducted for the current project needs to support the Master Plan and the analysis conducted under this MOA should reflect uncertainties that are documented in the Master Plan. Once the Master Plan is completed and preliminary data are available on genetic diversity of sturgeon in the mid-Columbia, more robust experimental designs for both fish collections and data analysis should be developed and peer reviewed.

The proposal is insufficient for technical review. The landscape genetics and QTL investigations have the potential to be useful contributions to understanding how and why the diverse life-history variation in Chinook salmon and steelhead/resident rainbow trout is distributed in the Columbia River Basin. This information could lead to improved planning in the face of climate change and human population growth. This, however, is an omnibus proposal, with two distinct parts. It is not clear how they fit together. The proposal should be rewritten to address the criticisms identified below in the ISRP Comments.

Responses:

1. Previously generated genotype data are already available in shared databases known as GAPS and SPAN.  New genotype data can be made available if a public database is identified.  When possible, new baseline genotype data will be submitted to GAPS and SPAN databases.  Also, genotype data that is part of publications for this project will be submitted to the "Dryad" database.  Dryad is an international repository of data underlying peer-reviewed articles in the basic and applied biosciences at http://datadryad.org

2. There is currently no public database for gene expression data that may be generated in this study.  However, we will attempt to submit published data to Dryad.