Posterior estimates for the parameters in the Gompertz state-space model (eqns. 1–3), obtained for both prey species through Markov chain Monte Carlo integration.

*The range of possible models within the saturated one is ordered for each species according to increasing complexity. The modelled population process, denoted by the state variable Xi, can be affected by a range of parameters: β0 stands for the density-independent growth rate; β1 stands for first-order density-dependence; β2 denotes second-order density-dependence; γ stands for the weather effect on population size; and σ2, τ2 stand for the process and sampling variances, respectively. Shown are the mean ± Standard Error. Parameters in which the 95% credible interval does not overlap 0 are shown in bold type.
¶Bayesian Information Criterion; the model minimizing this quantity is selected as the best descriptor of the dataset within the pool of fitted models, and is shown in bold type.