A proposed methodology for the determination of seed sources for tree native species based on environmental variables: the case of Quillaja saponaria Mol

There is an increasing interest in planting native tree species for commercialization and restoration purposes. However, the seed provenance zones of plant material to complete these activities successfully are not always available. This study aimed to develop a methodology that delineates seed zones based on provenances and using climatic and topographic variables. We created a database with spatial units throughout the natural distribution of Quillaja saponaria Mol. (30o to 38oS), containing climatic and topographic information and a relative abundance of the species. First, a principal component analysis (PCA) was performed to synthesize data, with the first three components explaining 89% of the total variance. Then, a non-hierarchical cluster analysis was performed to define homogenous groups, which was defined through linear discriminant analysis. Finally, an ordinal logistic regression (OLR) was performed using the abundance value as response and the environmental variables as predictors, obtaining a maximum of 40% of success in the prediction of abundance. Even though abundance was a complex response to be predicted in the case of Q. saponaria, the proposed method to delineate seed sources showed geographical coherence. The proposed methodology is easily replicable to other species using free databases and computing tools and allows a preliminary estimation of seed transfer zones for Q. saponaria.