Measuring plant diversity in a two-stage sampling design by Bayesian updated relative abundances

The two-stage sampling design provides good local estimates of both the number of plant species and the relative abundances. However, it is a problem to calculate Hill diversity indices at the local scale, because some of the species found in the large plot are not present in the small plot and such species should then incorrectly be weighted with zero relative abundance. A new method for calculating local Hill diversity indices from species richness and relative abundances data is therefore needed. We suggest to replace the local relative abundances with Bayesian updated relative abundance estimates, where the prior probability distribution of the relative abundances are empirically estimated from all plots of the same habitat types. The method is applied on Danish Nardus grasslands.