WLF 448: Fish & Wildlife Population Ecology

Fall 2010

Lab 7: Model Selection and Time Series Analyses

I. Viable Population Monitoring

A. Stochasticity results in some modeling scenarios where population would go extinct at some point in future

B. Probability of reaching quasi-extinction threshold

II. Model Selection

          A. Exponential Growth (EGPN)

        B.  Ricker

        C. Gompertz

        D.  Use Akaike Information Criteria (AIC) to find which model fits best

III.  Use of Residuals

A.  positive = "good year"

B.  negative = "bad year"

C.  Residuals plotted against covariate ==> high R² = good fit, low R² = bad fit

IV.  Lag Effects

A.  Some covariates may influence subsequent year(s) population abundance

B.  Plot residuals versus covariate with time lag and see if R² value improves

C.  Must make biological "sense"

Revised: 05 October 2010