Optimal pooling of data for the reliable estimation of trait probability distributions

Aim Functional diversity indices that summarize trait variation between organisms within communities are widely used tools in community ecology. Although intraspecific is often an important component of the total variation, traditional functional neglect this component. Recently, Carmona et al. proposed a framework to overcome limitation based on fitting probability distributions (TPDs). Unfortunately, article does not contain any guidance pooling data before TPDs. Innovation Reliable TPDs needs larger sample size than usually used. This can be reached comes from different localities. If TPD same these localities, results more reliable estimation. Moreover, if pooled represent TPDs, fitted distribution artefact. In paper, we suggest algorithm for automated selection optimal pooling. It Gaussian mixture models and model using Bayes information criterion (BIC). Main conclusions The new was able select data, which illustrated artificial data. Analysis field examples showed obvious: both merging all analysing measurements each locality separately might result unreliable estimates.