Sensitivity of primary model parameters on predicted prevalence of CWD infection and deer density after 25 years (2035).

aWe used Latin Hypercube Sampling for each parameter with S = 20 equal probability intervals and one random value from each interval. Values for each parameter were paired randomly with values from all other parameters.bMean SPC ± SE between input parameters and prevalence or deer density in 2035; stochastic analysis based on S = 20 replications. Probability of a t-statistic (with S-2 df; [43]) that evaluates SPC = 0 provided in parentheses.cBecause βmale, βfemale, and time since disease introduction (TDI) are correlated, we used a trivariate normal distribution, N({TDI,βmale,βfemale}, ∑), where ∑ = variance-covariance matrix, calculated for parameter combinations within the 95% confidence region [69]; TDI rounded to closest integer.dTransition (γ) from lymph-node positive (I) to obex-positive (O) represents differences in CWD progression among deer genotypes in terms of CWD susceptibility; standard error (SE) derived from 95% CI bounds = 8–16 months (e.g., representing transition to Obex infection for the two common genotypes).eGaussian distributions for harvest rates of antlered and antlerless deer are based on mean hunting rates and coefficients of variation (0.18 and 0.15, respectively) during 2002–2010 in the core area.We used Latin Hypercube Sampling and a semi-partial correlation coefficient (SPC) to measure the relative influence of model parameters; significant coefficients are bolded.