the film
Economic and dam related articles

Recovery and Management Options for
Spring/Summer Chinook Salmon
in the Columbia River Basin

by Peter Kareiva & Michelle Marvier
Nature - November 3, 2000


Construction of four clams on the lower Snake River (in northwestern United States) between 1961 and 1975 altered salmon spawning habitat, elevated smolt and adult migration mortality, and contributed to severe declines of Snake River salmon populations. By applying a matrix model to long-term population data, we found that (i) dam passage improvements have dramatically mitigated direct mortality associated with dams; (ii) even if main stern survival were elevated to 100%, Snake River spring/summer chinook salmon (Oncorhynchus tshawytscha) would probably continue to decline toward extinction; and (iii) modest reductions in first-year mortality or estuarine mortality would reverse current population declines.
Graphic: If this indirect mortality were 9% or higher, then dam breaching could reverse the declining trend of Snake River Chinook Salmon. Dams in the Columbia River Basin of North America almost certainly contributed to severe declines in wild salmon runs. Some dams in this basin, such as the Hell's Canyon Dam, completely blocked salmon passage, eliminating much spawning habitat . Other dams allow fish passage, but turbines, predation in reservoirs, and other alterations in the migration corridor presumably increase salmon mortality . Ecological problems associated with dams are widespread and are leading to societal questions weighing the benefits of dams against their costs to depleted fish populations. The most dramatic decision yet faced involves four hydroelectric dams on the lower Snake River.

Salmonid evolutionarily significant units (ESUs) represent genetically distinct collections of populations . In western North America, 24 salmonid ESUs are listed under the Endangered Species Act; 12 of these are in the Columbia River Basin and four must pass the four lower Snake River dams. The U.S. Army Corps of Engineers is currently considering removing these dams to recover Snake River salmon. Although most scientists agree that dam removal will help salmon, it is not known how much benefit would be derived from this action or whether alternative modifications of fish passage could lead to population recovery.

We used an age-structured matrix model for Snake River spring/summer (SRSS) chinook salmon to describe the current situation and explore the demographic effects of reducing mortality at different life stages. Seven index stocks of SRSS chinook salmon have been intensively monitored since the late 1950s; all are declining (Fig. 1 and Web fig. 1), with current spawning populations averaging less than 10% of their 1950 levels. Using age-specific spawner data, we estimated demographic projection matrices for these index stocks (Table 1). The matrices isolate survival during upriver and downriver migration from survival in other life stages, allowing direct examination of the effect of mortality during in-river migration on population growth. These simple matrix models are density-independent; we found little evidence supporting a density-dependent model.

We used data for 1990-1994 brood years to estimate parameters for matrices for all index stocks (Table 2), restricting analyses to recent years because these stocks have suffered progressively declining productivity. We thus examined a worst case scenario, taking a precautionary approach to the evaluation of endangered species. The dominant eigenvalues of these matrices indicate the long-term annual rates of population change (assuming that demographic rates remain constant) and all are substantially less than one.

We used these matrices to determine the effect of eliminating all migration mortality except for a small tribal harvest. Although perfect survival during in-river migration is unobtainable, it is a useful numerical experiment because one goal of both dam breaching and modification of intact dams is to reduce in-river migration mortality. Remarkably, even if every juvenile fish that migrated downstream survived to the mouth of the Columbia, and every returning unharvested adult fish survived to reach the spawning grounds, the index stocks would continue to decline (Fig. 2). Thus, management aimed solely at improving in-river migration survival cannot reverse the SRSS chinook decline.

We also tested the effectiveness of three past management actions: (i) reductions of harvest rates, from approximately 50% in the 1960s to less than 10% in the 1990s; (ii) engineering improvements that increased juvenile downstream migration survival rates from approximately 10% just after the last turbines were installed to 40 to 60% in most recent years; and (iii) the transportation of approximately 70% of juvenile fish from the uppermost dams to below Bonneville Dam, the lowest dam on the Columbia River. If such improvements had not been made, the rates of decline would likely have been 50 to 60% annually (Fig. 3), and spring/summer chinook salmon might well have already disappeared from the Snake River. Hence, past management actions have reduced in-river mortality but have not reversed population declines.

Finally, we tested whether improved survival at other life stages could reverse the population declines. Choosing the matrix with the median dominant eigenvalue (Poverty Flat) as a benchmark, we calculated combinations of first-year survival (s1) and early ocean/estuarine survival (se) values that give a dominant eigenvalue of 1.0 [a steady-state population in a deterministic world (Fig. 4)]. We neglected adult mortality because ocean harvest is negligible on these stocks, and management opportunities for enhancing open ocean survival are limited. For Poverty Flat, management actions that reduce mortality during the first year by 6%, or reduce early ocean/estuarine mortality by 5%, would be sufficient. If reductions in mortality are simultaneously accomplished in both the first year of life and the early ocean/ estuarine stage, then the combinations of mortality reductions required to produce an eigenvalue >/= 1.0 are as modest as a 3% reduction in first-year mortality and a 1% reduction in estuarine mortality. These required improvements are surely underestimates because the analyses are deterministic. Although we lack data to parameterize a stochastic matrix model, environmental variability reduces long-term population growth as compared to deterministic analogs. To accommodate this effect, we repeated the calculation with a target of 10% annual growth. When we made this precautionary adjustment for stochasticity, we found that first-year mortality must be reduced by 11% or early ocean/estuarine mortality must be reduced by 9%. In addition, our conclusions are qualitatively robust to a wide range of parameter values for chinook salmon.

The challenge of increasing first-year and estuarine survival shifts scientific inquiry from demographic modeling to identifying management actions that might produce the desired improvements. Because SRSS chinook salmon spawn in the upper reaches of Snake River tributaries, dam breaching is unlikely to affect available spawning habitat or first-year survival but could improve estuarine survival considerably. Although survival of juvenile fish during barging is quite high, barging might reduce the subsequent survival of barged fish relative to those that swim downstream. Breaching the lower Snake River dams would mean the end of fish transportation operations and would therefore eliminate any delayed mortality from transportation. Additionally, the removal of four of the eight dams encountered by Snake River salmon might increase the physiological vigor of salmon that swim downriver, thus improving survival during the critical estuarine phase. If this indirect mortality were 9% or higher, then dam breaching could reverse the declining trend of SRSS chinook salmon (Figure 5 above). Unfortunately, estimating the magnitude of any indirect mortality from passage through the Snake River dams is difficult; identifying fish appropriate as a "control" for the potential effects of these dams is problematic. Also, even if the Snake River dams were removed, the fish would still have to negotiate four Columbia River dams, and baseline mortality would still include any indirect mortality attributable to passage through those dams.

For the Snake River, deliberation regarding dam removal will require us to examine the effects of dams that may be manifested outside the migration corridor. Given the current uncertainty, policy-makers may have to view the decisions they make as large experiments, the outcomes of which cannot be predicted but from which we can learn a great deal pertaining to endangered salmonids worldwide.

Peter Kareiva & Michelle Marvier
Recovery and Management Options for Spring/Summer Chinook Salmon in the Columbia River Basin
Nature, Vol. 290 Issue 5493, p977, November 3, 2000

See what you can learn

learn more on topics covered in the film
see the video
read the script
learn the songs
discussion forum
salmon animation