Appendix D. Sensitivity analysis of the bias when varying the expectation and the variance of the natural survival without any competitive mortality (Φ0) in the wild boar population (this study, Table D1) and in the white stork population (Schaub and Lebreton 2004, Table D2).
The sensitivity analysis of the bias may differ according to the studied species or populations.
When looking at the sensitivity of the bias in the case of the wild boar population, we found that the bias was little sensitive to variations in the expectation and variance of the natural survival without any competitive mortality (Table D1). On the contrary, when looking at the sensitivity of the bias in the original Schaub and Lebreton (2004) study of white storks, we found that the bias was highly sensitive to variations in the expectation and variance of the natural survival without any competitive mortality (Table D2). This led to large variations in the bias-reduced correlation between the two mortalities. In such a case, it is therefore crucial to get a reliable estimate of the survival without any competitive mortality.
TABLE D1. Sensitivity analysis of the bias estimate in the wild boar population (this study) to the expectation and the variance of the natural survival in absence of hunting (f0). For each case, we indicated the estimate of the bias (B and 95% Bayesian confidence interval) and the resulting estimate of the bias-reduced correlation between the two causes-specific mortalities (BRC and 95% Bayesian confidence interval). |
TABLE D2. Sensitivity analysis of the bias estimate in the white stork population (Schaub and Lebreton 2004) to the expectation and the variance of the natural survival without any competitive mortality (f0). For each case, we indicated the estimate of the bias (B) and the resulting estimate of the bias-reduced correlation between the two causes-specific mortalities (BRC). The values considered by Schaub and Lebreton (2004) are in bold. As a single estimate is used here in the bias estimation, no level of uncertainty can be associated with this value and thus 95% Bayesian confidence interval can not be indicated. |
LITERATURE CITED
Schaub, M., and J.-D. Lebreton. 2004. Testing the additive versus the compensatory hypothesis of mortality from ring recovery data using a random effects model. Animal Biodiversity and Conservation 27:73–85.