SummarySeveral population groups of Pacific salmon (Oncorhynchus spp.) have been listed as Threatened or Endangered under the U.S. Endangered Species Act. A number of approaches have been used to assess risk to salmon populations, including simple rule-based class-ifications and model-based analyses. Here, a statistical population dynamics model was developed and applied to evaluate demographic risks facing populations of coho salmon (Oncorhynchus kisutch) from the Oregon Coast. The model applies Bayesian risk assessment methods to combine density-dependent freshwater production with density-independent, environ-mentally driven marine survival to predict future population abundance and estimate risk of extinction. Density-dependence includes both depensation at low abundance and compensation at high abundance. Risk estimates combine effects of both environmental stochasticity and parameter uncertainty via a doubly-stochastic simulation, with random environmental replicates nested within a

nft=pN ft

1N ft

k

1−exp [N ft ln0.5

n50]

Beverton-Holt Curve

Depensation

Smolt Production Model

Thomas C. WainwrightNorthwest Fisheries Science Center, National Oceanic and Atmospheric Administration

2032 SE OSU Drive, Newport, OR, 97365, U.S.A. E-mail: thomas.wainwright@noaa.gov

Demographic Extinction Risk for Coho Salmon Populations

However, the results provide a basis for evaluating risks resulting from the major demographic factors of population abundance, productivity, and variability.

AcknowledgementsThe model was developed jointly with Paul Spencer and Steven Lindley. This work was initiated as part of a working group on “Predicting Extinction: The Dynamics of Populations at Low Densities” supported by the National Center for Ecological Analysis and Synthesis, a Center funded by NSF (Grant #DEB-94-21535), the University of California at Santa Barbara, and the State of California. The model formulation was contributed to by other members of the working group: L. Botsford, M. Bradford, D. Goodman, R. Hilborn, J. Hutchings, C. Lamon, M. Liermann, R. Myers, and T. Sinclair.

This poster was prepared entirely with open source software -- special thanks to the developers of Linux, OpenOffice.org, GMT, and R.

Markov-chain Monte Carlo (MCMC) sample of the parameter space. Population parameter estimates combine prior information from meta-analysis of production in other coho salmon populations with information derived from recent abundance time series.

The model was applied to 20 coastal Oregon (U.S.A.) populations of coho salmon. Risks of absolute extinction (zero adults) and quasi-extinction (adult population abundance below 50) over 100 years were estimated. Posterior est-imates of key parameters (depensation threshold, intrinsic population growth rate, and habitat capacity) varied widely among populations, but integrated risk of extinction for the individual populations was uniformly low (typically between zero and 5%). However, estimation error was high for some populations.

The model does not provide a full assessment of extinction risk, as it excludes important issues such as population genetics, changes in habitat condition, the influence of hatchery production, and the influence of migration among populations.

Marine Survival Model

N ft = nf , t−3mt 1−ht mt = m⋅exp t t = AR 1

Symbols

nft ≡ Female smolts, generation tN ft ≡ Spawning females, generation tp ≡ Beverton-Holt slope parameterk ≡ Beverton-Holt capacity parametern50 ≡ Spawner abundance at 50% depensationmt ≡ Marine survival fraction, generation tht ≡ Harvest fraction, generation tt ≡ Autoregressive AR(1) process noise

DATA: Adult Abundance History (Nt)

DATA: Marine Survival History (mt) Population Locations

RESULTS: Risk Estimates

RESULTS: Posterior Parameter Estimates

Biological Process Model

Median

Quartile 1

Quartile 3

+Mean

Outlier