Pyper Covariation in Mid Col Steelhead Productivit

50 %
50 %
Information about Pyper Covariation in Mid Col Steelhead Productivit
News-Reports

Published on September 6, 2007

Author: Crystal

Source: authorstream.com

Covariation in Productivity ofMid-Columbia Steelhead Populations:  Covariation in Productivity of Mid-Columbia Steelhead Populations S.P. Cramer andamp; Associates, Inc. 600 N.W. Fariss Road Gresham, OR 97030 www.spcramer.com Brian Pyper andamp; Steve Cramer Slide2:  Mid – Columbia Study Area Background:  Background Population abundance Population growth rate (productivity) Spatial structure Diversity Mid-Columbia steelhead ESU listed as threatened NMFS uses four measures to evaluate viable salmonid populations (McElhany et al. 2000): Background:  Background 'Lambda' analysis a key tool used by NMFS to assess productivity (Homes 2001; McClure et al. 2003) 'Lambda' measures population growth rate and extinction risk using time series of escapement data (increasing or decreasing trend?) Model is not mechanistic Assumes no density dependence in spawner-recruit dynamics Slide5:  Slide6:  Slide7:  Spawner-recruit analysis:  Spawner-recruit analysis Examined spawner-recruit data for 8 populations (Cramer et al. 2005) Estimated intrinsic growth rates and capacity Compared 4 spawner-recruit models: Density independent model Ricker model Beverton-Holt model Hockey-stick model Used simulations to examine potential bias Data:  Data Dam counts of natural-origin spawners : Deschutes Yakima Umatilla Redd counts (index) for 5 John Day subpopulations: Upper and Lower Mainstem South, Middle, and North Forks Recruitment indices based on available harvest and age-structure data Slide10:  Population abundance of natural-origin steelhead in the Middle Columbia ESU, 1978-2004 Slide11:  Population abundance of natural-origin steelhead in the Middle Columbia ESU, 1978-2004 Slide12:  Population abundance of natural-origin steelhead in the Middle Columbia ESU, 1978-2004 Covariation in recruitment:  Covariation in recruitment Escapement indices correlated (Avg. r = 0.63) Suggests shared influence of freshwater or marine conditions on survival Suggests limited measurement error Next step: Fit spawner-recruit models … Slide14:  0 2 4 6 8 0 2 4 6 8 10 85 86 87 88 DI RK 1:1 BH HS Spawner Index Recruit Index Fits of the spawner-recruit models to the North Fork data set of the John Day population (DI = density-independent model, RK = Ricker model, HS = logistic hockey-stick model, and BH = Beverton-Holt model). Model comparisons:  Model comparisons Used the AIC model-selection criterion Beverton-Holt and Hockey-stick models 'best' across data sets But many unstable fits and unreasonably high estimates of intrinsic growth rate (alpha) Range in Alpha (Recruits per spawner) Beverton-Holt: 5.5 to 72.9 Hockey-stick: 2.4 to 20.8 Ricker: 2.6 to 5.2 Model comparisons:  Model comparisons Ricker model stable with biologically reasonable estimates of growth rate (alpha) Ricker fits much better than Density- Independent model for all 8 data sets Note: Estimates of capacity similar across forms (Ricker, Beverton-Holt, Hockey-stick) Density Independent model assumes no limit to capacity Slide17:  Fits of the Ricker and Density-independent models 0 2 4 6 8 0 2 4 6 8 10 JD North Fork 0 2000 4000 6000 8000 10000 0 2000 6000 10000 Deschutes 0 1000 2000 3000 0 1000 2000 3000 Umatillla 0 500 1000 1500 2000 2500 0 1000 2000 3000 Yakima Spawner Index Recruit Index 85 86 87 88 85 86 87 88 85 86 87 88 85 86 87 88 Slide18:  0 5 10 15 0 5 10 15 JD Upper Mainstem 0 2 4 6 8 10 12 14 0 5 10 15 JD Lower Mainstem 0 5 10 15 20 0 5 10 15 20 JD South Fork 0 5 10 15 0 5 10 15 JD Middle Fork Spawner Index Recruit Index 85 86 87 88 85 86 87 88 85 86 87 88 85 86 87 88 Fits of the Ricker and Density-independent models Slide19:  Ricker estimates of intrinsic growth rate (alpha) Average = 3.4 recruits per spawner Slide20:  Ricker estimates of intrinsic growth rate (alpha) Average = 3.4 recruits per spawner 0 2 4 6 8 10 12 14 Upper Mainstem Lower Mainstem South Fork Middle Fork North Fork Deschutes Umatillla Yakima Ricker Alpha (Recruits/Spawner) ` Ricker Average for Density- Independent models = 1.4 Recruits/Spawner Slide21:  Ricker estimates of capacity: unfished equilibrium spawner abundance (S*) 0 2,000 4,000 6,000 8,000 10,000 Deschutes Umatillla Yakima Ricker S* ` Spawner Abundance Slide22:  Ricker estimates of capacity: unfished equilibrium spawner abundance (S*) 0 2,000 4,000 6,000 8,000 10,000 Deschutes Umatillla Yakima Recent 5-yr geometric mean Ricker S* ` Spawner Abundance Slide23:  Ricker estimates of capacity: John Day 0 5 10 15 20 Upper Mainstem Lower Mainstem South Fork Middle Fork North Fork Recent 5-yr geometric mean Ricker S* Redds per Mile Slide24:  0 5 10 15 0 5 10 15 JD Upper Mainstem 0 2 4 6 8 10 12 14 0 5 10 15 JD Lower Mainstem 0 5 10 15 20 0 5 10 15 20 JD South Fork 0 5 10 15 0 5 10 15 JD Middle Fork Spawner Index Recruit Index 85 86 87 88 85 86 87 88 85 86 87 88 85 86 87 88 Influence of 1985 – 1988 brood years: Density dependence or poor ocean survival? Slide25:  Removed years and re-fit Ricker models Similar results – still get strong evidence of density dependence (P andlt; 0.01) for 8 data sets Consistent estimates of growth rate (alpha) Influence of 1985 – 1988 brood years Slide26:  Combined data (spawner index standardized so median = 1 for each data set) Slide27:  Combined data (spawner index standardized so median = 1 for each data set) Slide28:  Possible bias in Ricker parameters related to: Short data sets Measurement errors Autocorrelation Harvest rates Estimates of parameters uncertain Strong concern for NMFS (McElhany et al. 2000) Can use simulations to assess potential bias Potential problems with spawner-recruit analyses Slide29:  Simulated spawner-recruit data with same characteristics as Mid-Columbia data True alpha = 3 High autocorrelation Low harvest rates Assumed measurement error in age structure and escapement estimates (CV = 30%) Estimated Ricker parameters for each simulated data set to assess potential bias Simulations Slide30:  Results (500 simulations) 0 10 20 30 40 50 60 1.0 2.0 3.0 4.0 5.0 6.0 Estimate of Ricker alpha Number of Simulations True value = 3.0 Median estimate = 3.2 Slide31:  Bias in Ricker parameters was minimal (10 to 20%) for range of conditions typical of Mid-Columbia steelhead data sets Primary reason was low harvest rates (20% across most years) Significant bias expected for harvest rates = 40% or greater across years Simulations results Slide32:  Widespread evidence of density dependence in Mid-Columbia steelhead data sets Consistent estimates of intrinsic growth rates (avg. = 3.4 recruits per spawner) No evidence that one or more populations experienced relatively poor productivity 'Lambda' only useful as a red-flag indicator Intrinsic growth rates suggest resilience to short-term increases in mortality Summary

Add a comment

Related presentations