Spatial PCA and Random Forest Analyses of Spatial Patterns in Genetic Diversity of Calotropis procera

Abstract

Landscape genetic study can provide data on spatial effects on the genetic structure of plants. Calotropis procera is an economically and medicinally important plant species that grows in limited regions of southern parts of Iran. Identifying genetic variability and corridor of gene flow may be used in the conservation of this valuable plant. For this reason, we used a multivariate spatial PCA (sPCA) and a machine learning approach of random forest (RF), to reveal the genetic divergence of twelve populations in Calotropis procera with respect to spatial and geographical features. sPCA analyses revealed significant isolation by distance and spatial genetic structuring at both global and local levels. RF analysis identified temperature and altitude as major spatial features affecting the genetic differentiation of these populations based on the Fst value. We also noticed the genetic differentiation of populations within a single area due to local spatial effects. However, these populations did not force a separate infraspecific taxonomic entity. The genetic connectivity graph is constructed and a gene flow corridor is suggested for these populations. Similarly, resistance surface including temperature, latitude, and soil type has been proposed for these plant species.