Executive Summary: ISAB Review of Fish Passage Analyses and Life-Cycle Models for the U.S. Army Corps of Engineers’ Willamette Valley Project
November 3, 2014
In August 2014, the ISAB completed a two-part review of salmonid passage and life-cycle modeling associated with the Willamette Project, a system operated by the U.S. Army Corps of Engineers that comprises 10 high-head federal dams and reservoirs, three run-of-river dams that function as re-regulating projects, and 42 revetments located in Willamette River tributaries. Bonneville Power Administration (BPA) is responsible for marketing and transmitting power generated from 8 dams, with the remaining dams being non-power producing facilities. The U.S. Bureau of Reclamation (BOR) administers a water marketing program to agricultural users for water stored in Corps’ reservoirs.
The models reviewed by the ISAB were developed by the Corps, NOAA Fisheries, and other cooperators to inform the Corps’ Configuration and Operations Plan, which is to be completed by December 2014. The ISAB first reviewed the Fish Benefits Workbook (FBW), which was designed as a tool to help evaluate alternative approaches to improving downstream passage efficiency and survival of salmonids through dams and associated reservoirs in the Willamette River Basin (ISAB 2014-3; June 23, 2014). The ISAB then reviewed NOAA Fisheries’ Viable Salmonid Population (VSP) life-cycle modeling of Willamette River spring Chinook populations (ISAB 2014-4; August 1, 2014), which depends on inputs from the FBW and other data sources to predict salmonid viability under alternative structural or operational changes to the dams. The ISAB reviews are intended to provide constructive feedback to the Corps, NOAA Fisheries, and their cooperators as they complete analyses supporting the Configuration and Operations Plan.
This executive summary covers both ISAB reviews and is intended to convey the ISAB’s primary findings and recommendations to decision makers and a general audience. The ISAB recognized the need for a short high-level summary during discussions with the Corps and model developers after the more detailed reviews had been released.
Valid Approach: The approach for estimating VSP scores based on outputs from the FBW and VSP life-cycle models was judged to be conceptually and technically valid. The ISAB recognizes this was a sophisticated effort to grapple with a very difficult task. VSP scores for diversity and spatial structure parameters are based on qualitative assessments, but this approach seems reasonable given that data are insufficient to conduct a more quantitative analysis. Similarly, modeling genetic impacts of hatchery fish as a reduction in wild egg-to-fry survival based on proportionate natural influence is somewhat speculative, but the approach seems plausible and consistent with current theory. The ISAB sees merit in the effort to consider genetic impacts of hatcheries within the model.
Limited Data: Developing a life-cycle model of this complexity for Willamette Chinook populations is particularly challenging because insufficient biological information for many parts of the models is available to model the life cycle of individual populations with confidence. In short, the modeling effort is far ahead of the empirical data needed to fit and verify the models. Faced with an absence of location-specific data, the authors have made a commendable effort to parameterize models with ranges derived from expert opinion generated through formal procedures at workshops, rather than arbitrary guesswork. Moreover, the models are flexible enough to incorporate additional information as it becomes available in the long term.
Uncertain Outputs: Just as conclusions from a valid argument will be unsound if its premises are wrong, predictions of future spawner abundances and VSP scores from the existing life-cycle model should be considered highly uncertain given limitations on the quality of data currently available. The ISAB suggests that this uncertainty has been underestimated and that the confidence intervals on outcomes are probably too narrow. In population viability analyses, it is typically more feasible to rank the relative performance of different management scenarios than to estimate absolute values of survival or extinction risk under those scenarios. Even so, given the wide confidence intervals on the final VSP scores, it is not clear how significantly the results differ among scenarios. It might also be true that self-sustaining, natural-origin salmon runs cannot be achieved under any of the scenarios considered.
Additional Considerations: Because quantitative outputs from the existing models may be unreliable to differentiate among the options, additional steps should be considered to support decision making in the near term. Weighting factors, including expected survival benefits and other considerations, might be useful for ranking the alternatives under review. Incremental implementation of some alternatives (i.e., where feasible) might be advantageous provided the decision-making process remains flexible and monitoring is adequate to evaluate early results. An adaptive approach would allow decisions to be revisited if benefits expected from an operational regime were not being realized or if new information were to suggest that greater benefits could be achieved by different actions.
The current FBW and life-cycle models provide a systematic foundation for exploring the potential value of collecting additional data. Further sensitivity analysis could help to prioritize research by identifying the kinds of additional data that would improve quantitative assessment of management options in the long term. This approach seems warranted given that the performance of these models is currently limited by the availability of data. With more complete data, these models could be used to predict and evaluate the long-term success of actions to improve dam passage, as well as to assess the probable impacts of other factors affecting viability such as hatchery fish, habitat conditions, prespawning mortality, predators in the reservoirs, and climate change.
Uncertainty and Data Gaps:
- Improve documentation of assumptions, data sources, parameter values, and reasons for choosing parameter values in the FBW and VSP life-cycle models. More explicit documentation would improve transparency and aid in communicating the appropriate use of the results, and in refining the models as more information becomes available. In addition, more ResSim documentation is needed because the FBW is dependent on ResSim outputs.
- Clarify how stochastic variability is represented in the life-cycle model. In particular, it is not clear which components of the model include year-to-year variability and which do not.
- Conduct simulations with ResSim that replicate past operations at each reservoir in order to verify that the model can adequately predict flows in the FBW (or document the magnitude of errors).
- Conduct further sensitivity analyses to identify and prioritize data gaps that must be filled to enable a more reliable assessment of options for structural, operational, and other changes to dam management and other parts of the system.
Research to Address Known Issues:
- Conduct field research and additional modeling to improve capabilities to predict and reduce pre-spawning mortality. The sensitivity analysis indicates that observed levels of pre-spawning mortality could compromise efforts to develop self-sustaining populations of Chinook salmon above the dams in the Willamette basin.
- Conduct field research to determine how reservoirs in the Willamette basin affect growth, migration, and survival of juvenile salmon and steelhead. This information appears essential for evaluating how various options for structural, operational, and other changes to dam management and other parts of the system affect population viability. Model predictions might also be improved by acquiring better data on the quantity and quality of habitat both above and below the projects, together with more explicit modeling of the influence of this habitat on VSP parameters.