10,000 Years Institute - scientific research for natural resource management from Seattle, WA to Lake Baikal, Russia.
10,000 Years Institute - scientific research for natural resource management from Seattle, WA to Lake Baikal, Russia.
Summary: Comparative Risk Model to Identify
the Best Combination of Fish Passage Technologies

Strategies for restoration of anadromous fish runs which involve systems to pass fish over dams require that fish traverse numerous machines and man-made environments during all freshwater life stages. Because cumulative mortality within a fish passage system undermines the long-term success of the population, controlling mortality at key points in the passage system is the best means to a cost-effective and successful restoration program.

On the Lewis River in southwestern Washington, there are 50 to 60 miles of good spawning and rearing habitat upstream of a power plant and three high-head dams. Managers of hydropower projects are considering development of facilities for successful passage of salmonids to and from this area. 10,000 Years Institute applied ecological risk assessment concepts to model the cumulative population-level risks of different passage systems in the Lewis River.* The model was based on coho (Oncorhynchus kisutch) because of their relatively simple life history pattern, but can be adapted to accommodate other species. We performed a basic comparison between two very different passage systems, and used sensitivity analyses to identify which elements of each system would result in the best overall population performance.

Conceptual model of trap and haul
Figure 1. (click photo to enlarge) Conceptual model of trap and haul passage.
Conceptual Model
Within a risk assessment framework, each step in the fish passage system is a stressor. Exposure assessment consists of estimating what fraction of the fish population encounters each passage device (e.g., screen, bypass, truck, ladder). Risk is characterized as the cumulative mortality throughout the life cycle of one generation of fish. Potential success of a passage system is gauged by analyzing population sustainability over multiple generations.

Our passage risk model tracked a cohort of fish from their initial downstream migration as smolts through each step of a particular passage scenario to their eventual return to their natal streams as adults to spawn. Our model calculated incremental losses to the population across the passage system, and in natural habitats over the life cycle of the fish.

Conceptual model of trap and haul
Figure 2. (click photo to enlarge) Conceptual model of a volitional passage system.
We initially compared a "trap and haul' program of capturing fish (i.e., loading them into trucks, and driving them upstream or downstream; (Figure 1) with volitional passage (Figure 2). Both were considered because, while trap and haul is substantially less expensive, volitional passage results in a lesser degree of fish handling, which was preferred for cultural reasons by Native Americans involved in the planning process; anglers also prefer volitional passage. With marine survival as the only stochastic model parameter, we modeled coho populations in each passage system over 50 generations. In our sensitivity analyses, marine survival was held constant at four percent.

Effect of the D Factor
Figure 3. (click photo to enlarge) The effect of the D Factor, or return of trucked fish (TH) relative to returns of volitional (VOL) migrants. While much more expensive, volitional passage appears to have an inherent biological advantage because those fish that migrate in-river are more likely to return.
Results and Sensitivity Analysis
Initial estimates of smolt numbers were based on a deterministic Beverton-Holt equation which links smolt production to the number of spawning females in the previous generation, and is based on production per unit of habitat area. Other model assumptions were based on studies of passage systems elsewhere in the Columbia River Basin, or on site-specific studies.

Over 50 generations, a volitional system provided better overall population-level performance for coho than a trap and haul system. This result was largely dependent on the "D Factor" - the fraction of successfully returning adults that were transported by truck as juveniles to those which migrated in-river as juveniles. The D Factor is based on studies on the Columbia River that suggest that in-river migrants successfully return to their watersheds at a higher rate than trucked fish; this difference is expressed as a fraction (D). In our study, measured values of D ranged from 0.52 to 0.73, while our sensitivity analysis indicated that D would have to be about 0.8 before trucked fish would ultimately produce the same number of spawners as fish that migrated in-river (Figure 3).

Spawner abundance when FGE is 90 percent
Figure 4. (click photo to enlarge) Spawner abundance when FGE is 90 percent. Improvements in turbine survival, which can be achieved when new technologies are used, are key to successful restoration of coho.
The model sensitivity analysis indicated that a combination of improvements in turbine mortality and fish guidance efficiency (FGE) generated the best overall gains in numbers of returning adults. Increasing FGE meant fewer fish in either scenario would pass through turbines, thereby minimizing the presence of turbines on smolt success. If FGE is constant at 0.4, the highest spawner abundance is 4,000 fish. When both FGE and turbine survival are at 0.9 (90 percent), up to 5,000 spawners may return in any given year (Figure 4).

The model and sensitivity analysis guided management attention to those passage and migration elements for which incremental improvements could result in the greatest increase in population performance, and focused managers on passage elements that could be readily improved with investment in new technologies. This type of risk modeling approach allows risk managers to pinpoint where technological improvements can result in the greatest gains while achieving management objectives.

* The Institute thanks Steward and Associates for their invaluable contributions to model development.