Table 1 Prior distributions of the model parameters.
From: A state-space modelling approach to wildlife monitoring with application to flying-fox abundance
Parameter | Symbol | Prior | Rationale |
|---|---|---|---|
Rate of recruitment (mth−1) | ρ | U (0, 0.11) | Maximum 40% increase (over 3 months) |
Mortality rate (mth−1) | μ | U (0, 0.1) | Maximum 70% reduction (over 12 months) |
Cyclone effect on mortality | c μ | U (1, 10) | Initial 10-fold increase the maximum believed possible |
Cyclone effect on recruitment | c ρ | B (1.01, 1.01) | Largely uninformative |
Observation uncertainty std dev. (lognormal) | \({\sigma }_{obs}^{2}\) | U (0.05, 0.47) | Equates to belief CV of between 5% and 50%27 |
Process uncertainty std dev. (lognormal) | σ proc | U (0, 10) | Uninformative (on logit scale) |
Seasonality parameter 1 | α 1 | U (1, ∞) | Uninformative |
Seasonality parameter 2 | α 2 | U (α1 + 1, ∞) | Uninformative |
Cyclone in-camp roosting effect | c R | B (1.01, 1.01) | Uninformative, other than down-weighting 0 (total abandonment) and 1 (no effect) |
In-camp process uncertainty std dev. (normal on logit scale) | σ camp | U (0, 100) | Uninformative |
Exponential rate of increase (yr−1) | r | — | Calculated from the posterior distribution based on ρ, c ρ , μ, c μ and observed cyclone frequency |
