Fig. 2

Taxon assignment of 1369 Quercus individuals using linear discriminant analysis (LDA) with different thresholds of species assignment probability. This multivariate approach combines eight morphological and six synthetic genetic variables. Unambiguous species-specific hair types were used to predefine the groups of the training dataset. Shown are the position and classification of each tree along the two discriminant functions, using a highest probability (majority rule) and b 80 % as probability threshold for species assignment.