Population viability analysis of Snake River chinook: What do we learn by including climate variability? Rich Zabel NOAA Fisheries Seattle, WA
Population Viability Analyses Count-based PVA (Dennis et al. 1991) Based on time series of abundance ln(N t ) t
Population Viability Analyses Count-based PVA (Dennis et al. 1991) Based on time series of abundance ln(N t ) t
Population Viability Analyses Count-based PVA → (mean annual growth rate) → Prob [falling below a threshold] → Salmon Example: McClure et al. 2003
Population Viability Analyses Demographic PVA (Leslie 1945) Typically based on short-term demographic data. Demographic Rates are fixed s 1 b 3 m 3 /2 s 1 b 4 m 4 /2 s 1 b 5 m 5 /2 2s2s2 3s3s3 4(1-b 3 )s 4 5(1-b 4 )s 5
Population Viability Analyses Demographic PVA → (Mean annual growth rate) → Sensitivity analysis: How does change in response to changes in demographic rates? → Kareiva et al. 2000, Wilson 2003
Population Viability Analyses But climate effects notably absent from most PVAs PVAs are data-driven, and considerable data are required to characterize climate effects
Population Viability Analyses Spawners Parr Smolts Estuary Early ocean Ocean
Population Viability Analyses Spawners Parr Smolts Estuary Early ocean Ocean Climate
Population Viability Analyses “Mechanistic” PVA Relate variability in specific demographic rates to intrinsic (population density) or extrinsic (environmental) factors
Population Viability Analyses “Mechanistic” PVA → More realism by capturing important drivers → Combination of count-based and demographic PVA, thus can produce viability measures of both
Population Viability Analyses “Mechanistic” PVA → Snake River spring summer chinook → Long-term data at several life stages → Important drivers: 1) Ocean conditions upon entry 2) Density dependence in freshwater productivity
General Question How does adding complexity to the models enhance our understanding of population dynamics, and hence our ability to manage populations?
Snake River spring/summer Chinook Listed as a threatened ESU Meta-population with 31 identified sub-populations
Migratory Route in the Snake and Columbia Rivers
Migration of Adult Snake River Spring Chinook In the Pacific Ocean
12345 s2s2 s 3 (t) soso s o ·(1-b 4 ) b 4 ·F 4 (n) F 5 (n) Age-structured Life Cycle Model for Snake River spring/summer chinook
12345 s2s2 s 3 (t) soso s o ·(1-b 4 ) b 4 ·F 4 (n) F 5 (n) Age-structured Life Cycle Model for Snake River spring/summer chinook Survival
12345 s2s2 s 3 (t) soso s o ·(1-b 4 ) b 4 ·F 4 (n) F 5 (n) Age-structured Life Cycle Model for Snake River spring/summer chinook Propensity to breed
12345 s2s2 s 3 (t) soso s o ·(1-b 4 ) b 4 ·F 4 (n) F 5 (n) Age-structured Life Cycle Model for Snake River spring/summer chinook Fertility
12345 s2s2 s 3 (t) soso s o ·(1-b 4 ) b 4 ·F 4 (n) F 5 (n) Age-structured Life Cycle Model for Snake River spring/summer chinook Related to Ocean Conditions
3 year olds 4 year olds 5 year olds
Smolts per spawner Freshwater productivity Smolt-to-Adult Ocean Survival
Third-year survival and Climate Effects 1) Back-calculate from Smolt-to-Adult data (and estimates of riverine survival, ocean survival, harvest, age composition) 2) Relate to Monthly Pacific Decadal Oscillation Index (PDO)
Third-year survival and Climate Effects APRMAR FEB JAN DEC NOV OCT SEP AUG JUL JUN MAY Monthly PDO Indices Estuary Entry
Third-year survival and Climate Effects APR MAR FEB JAN DEC NOV OCT SEP AUG JUL JUN MAY Monthly PDO Indices Estuary Entry
Third-year survival and Climate Effects
Third-year Survival Year R 2 = Fit of Third-Year survival to Climate Data
Third-year Survival Year R 2 = Fit of Third-Year survival to Climate Data Data were autocorrelated, Residuals were not
Predicted Third-Year Survival Predicted Third-Year survival (and 95% CI) over the 100 year PDO record
Freshwater Density-Dependent Recruitment
Beverton-Holt fit to freshwater productivity a = density-independent slope a/b = carrying capacity
R 2 = 0.776
Putting it all together: Sample Model Output
Effects of Ocean Conditions
Four climate scenarios: “Historic” “Recent” “Bad” “None” mean and variance from
Effects of Ocean Conditions “Historic” Ocean “Bad” Ocean
Climate Produces Autocorrelation…
Effects of Ocean Conditions
Quasi extinction defined as < 3100 spawners
Effects of Ocean Conditions
Interactions between ocean conditions and freshwater productivity?
Sensitivity Analysis
Sensitivity of to 20% increase in DD-independent Survival Sensitivity of to 20% increase in Carrying Capacity (t) Year
Sensitivity Analysis Sensitivity of to 20% increase in DD-independent Survival Sensitivity of to 20% increase in Carrying Capacity (t) Year = 0.95 = -0.70
In other words… In time of favorable ocean conditions → More important to increase freshwater Carrying Capacity In times of unfavorable ocean conditions → More important to increase DD-independent Survival
Future Directions Meta-population structure Other drivers: Freshwater climate effects, seawater Density dependent effects Next step: How can we incorporate fish condition into viability models?
Conclusions Very useful to relate important drivers to the specific life stages upon which they act Climate clearly important factor for viability, both good versus bad and autocorrelation.
Conclusions Need to consider “worst case” climate scenarios in management Climate produces complexity in population dynamics: May be an interaction between ocean climate effects and freshwater productivity