Results of model fitting and model choice of the demographic models tested in <i>C</i>. <i>impatiens</i> and <i>C</i>. <i>resedifolia</i>.

<p>^a Demographic model: CON = constant populations size; BOT = population size bottleneck; EXP = population size expansion; RED = population size reduction.</p><p>^b Marginal density.</p><p>^c Fraction of the retained simulations (<i>n</i> = 5,000, corresponding to 0.5% of the total simulations) with a likelihood smaller or equal to that obtained for the observed data (Model fit).</p><p>^d Posterior probability, estimated as ratio between the marginal density (MD) of a model and sum of the marginal densities across models.</p><p>^e Fraction of pseudo-observed datasets generated under the model (PODs) for which such model was correctly identified as the best based on posterior probability (p<i>P</i>). Random expectation is 1/(no. of models) = 0.25.</p><p>^f Posterior probability corrected using multivariate kernel density estimation with bandwidth = 0.1.</p><p>Results of model fitting and model choice of the demographic models tested in <i>C</i>. <i>impatiens</i> and <i>C</i>. <i>resedifolia</i>.