The Columbia River Basin Fish and Wildlife Program calls for a regular system of independent and timely science reviews of the Fish Passage Center’s (FPC) analytical products. These reviews include evaluations of the Comparative Survival Study’s (CSS) draft annual reports. The ISAB has reviewed these reports annually beginning ten years ago with the evaluation of the CSS’s draft 2010 Annual Report and most recently the draft 2019 Annual Report (ISAB 2019-2). This ISAB review focuses on the 2019 Annual Report’s Chapter 2, Life Cycle Evaluations of Fish Passage Operations Alternatives from the Columbia River System Operations Environmental Impact Statement (CRSO-EIS), which was not available at the time the ISAB reviewed the draft 2019 Annual Report. Chapter 2 became available when the draft EIS (CRSO-DEIS) was released to the public on February 28, 2020. It is important to note that the ISAB is not reviewing the CRSO-DEIS but just the CSS analyses as reported in Chapter 2.
Chapter 2 of the 2019 CSS Annual Report is an important study for regional decision makers. Coordinated use of multiple models is a powerful approach answering difficult ecological questions. The use of a common hydrological input, management alternatives, and a common output predictor (smolt-to-adult return rate, SAR) potentially provide added confidence to modeling results and conclusions. The ISAB suggests improvements for description of model details, reporting of model results, and additional model development to better explore future conditions.
Model Development and Description
Chapter 2 implicitly assumes that the reader is familiar with the many years of the CSS reports and the CRSO-DEIS, and many key terms and data used in the chapter are not fully explained. Some methods and reporting styles used in Chapter 2 are likely a specific charge from the CRSO-DEIS steering committee, and the authors perhaps had less freedom to choose methodologies or the format for reporting and interpreting results. As such, it would be helpful for the authors to present the details of the charge they were given so readers can understand what was mandated for this chapter report and what was under control of the CSS.
Components of the cohort-specific (CSS-CS) model are not all based on the same years of data, and the years generally stop by about 2013, perhaps reflecting complete brood year records. However, partial brood year records are still useful for estimating juvenile survival. Why were data series truncated for the CSS-CS models? The CSS Grande Ronde life cycle (CSS-GR) model uses a longer record of data (1966 to 2010) than the CSS-CS models, but more complete brood-years appear to be available. How do the environmental conditions experienced by fish differ between the datasets used for model fitting? For example, if one set of data includes drought years while the other does not, the two models may perform differently. Reasons and implications for the choice of fitting data should be stated and discussed more fully.
The models examined effects of large-scale management actions, such as Snake River dam breaching, and various spill and flow scenarios that may incorporate other fish passage improvements, such as high capacity turbines or powerhouse surface passage. Model scenarios that include dam breaching and maximum spill result in the greatest SARs. Key models results, such as variability in projected SARs across years and associated measures of uncertainty for summary performance metrics, are not presented. Measures of process and population uncertainty around the point estimates will be difficult to develop because simulations do not include demographic stochasticity (i.e., the same set of inputs will always lead to the same number of fish surviving the hydrosystem, surviving the ocean, and returning to spawn). Variation in response metrics (e.g., SARs) would reveal differences among alternative management actions. A single point estimate of the average/median response without considering year-to-year variation across different flow regimes and without considering demographic stochasticity is much more difficult to interpret. Caveats about interpreting the modeling results should be stated more prominently.
SAR and marine survival may be sensitive to smolt body size, timing of emigration, and date of marine entry. Will these remain the same under all operational alternatives? Do other variables used in the model, partially or completely, capture these effects? The ocean is the source of important and highly variable mortality. How well have the models captured variation in ocean survival as a function of measurable variables that can be forecast into the future?
The life cycle models incorporate a rich model-generated dataset based on the modified flow record to identify processes that influence the variation and magnitude of SARs. Yet, only point estimates of the mean/median response are reported. A rationale for the sole use of means/medians needs to be included. The report could be improved with an analysis to provide deeper understanding and implication of WHY the scenarios performed as they did. The yearly outputs across scenarios should be sufficient to better understand why the models generated different SARs and relative abundances and why alternatives produced differences in SARs.
Future climate and environmental conditions
The CRSO-DEIS analyses and those reported in Chapter 2 are based on an 80-year modified flow dataset. Modified flows are defined as the historical streamflows that would have been observed if current irrigation depletions existed in the past and the effects of reservoir regulation were removed (CRSO DEIS, Appendix I, page I-4-1). The results from the mandated task of using the modified flow dataset to compare operational alternatives are useful, but the results may not be indicative of future benefits. The modified flow dataset does not reflect projections of flow and environmental conditions that may occur due to future climate change. This is noted by the authors at the end of the introduction of Chapter 2: “Finally, it is important to carefully consider the lower end of the predicted ranges of biological response metrics, as anticipated consequences of climate change suggest poor river or ocean conditions may occur more frequently, which would mean that the lower end of the predicted ranges is likely to occur more often.” This caution needs to be described more prominently in the report. Climate, ocean, and in-river conditions are unlikely to remain static in their current state or range of conditions.
The CRSO-DEIS contains an examination of the potential impacts of climate change (Chapter 4 and Appendix V) on the freshwater environment, ocean environment, and life histories of salmon and steelhead. These would provide a useful framework for future modeling and understanding the limitations of the current analysis. Chapter 2 should summarize these findings in more detail.
For such an important study, caveats of interpreting current findings as representing future benefits should be more fully presented in the chapter’s Discussion section. Many environmental, climate, operational, and social factors may change. A chart listing these other factors and their impact on the results would permit a side-by-side comparison and consideration of outcomes.
The ISAB recommends that:
Future projections of survival based on the modified flow dataset are likely to be overly optimistic. Recent years, such as 2015, have experienced very low summer flows and warm temperatures. The world’s five warmest years in the 1880 to 2019 record have all occurred since 2015 with nine of the 10 warmest years occurring since 2005 (NOAA Climate.gov). A sensitivity analysis needs to be performed to investigate the impact of climate change for potential future flow regimes, such as those described in Chapter 4 on Climate Change in the CRSO-DEIS. Such a sensitivity analysis will also need to account for changes in the maturation schedule, conversion probabilities due to warming water, and changes in the capacity in the Beverton-Holt spawner-recruit relationship for the CSS-GR model, habitat improvements, and other factors. Do relative rankings of the alternatives change if future climate scenarios (even if oversimplified) are represented?
A more detailed comparison of results between different types of flow years would be a useful first step toward meeting ISAB recommendation (1). Demographic and other stochasticity (80 years of hydrology is a great start) should be included in the models so that year-to-year variation in the output measures is more reflective of the response from different operations.
Both models do not incorporate the relationship of individual fish characteristics—such as body size, body mass, and condition factor, and date of ocean entry—to survival. The current literature is confusing (e.g., Faulkner et al. 2019 vs the rejoinder in Appendix G of the 2019 CSS Annual Report). It would be beneficial for both groups to collaborate on joint analyses and use a common data set to resolve this issue.