Test of the relationship of genetic and particularly epigenetic variation with geographic isolation and environment is important to reveal potential environmental drivers for selection. Rhododendron oldhamii is widespread but inhabits fragmented subtropical forest landscapes and populations across its range may exhibit different levels of genetic and epigenetic structuring correlated to their environmental conditions. Here, we investigated the genetic and epigenetic variations and their ecological correlates in R. oldhamii. Genetic and epigenetic variations were surveyed using amplified fragment length polymorphism (AFLP) and methylation-sensitive amplification polymorphism (MSAP), respectively. Using variation partitioning by redundant analysis (RDA), we examined the pure and spatially-structured environmental effects and pure spatial factors on genetic and epigenetic variations among individuals collected from 18 localities across R. oldhamii distribution range in Taiwan. We found that environments compared to geographic isolation among sites explained more genetic and epigenetic variations. Patchy distribution of the contemporary R. oldhamii populations was revealed by correlograms with patch size of approximately around 20-30 km based on the total genetic and epigenetic data. Spatial variables derived from the method of principal coordinates of neighbor matrices (PCNM), including PCNM3, PCNM5, PCNM7 and PCNM8 representing biotic processes, such as individual dispersal, were found to be important influencing potentially adaptive genetic and epigenetic variations. Annual mean temperature, annual precipitation, precipitation of the warmest quarter, aspect, slope and soil moisture were the most important environmental variables influencing potentially adaptive genetic and epigenetic variations, and could be particularly important for the evolution of local adaptation in R. oldhamii.

R.oldhamii_AFLP and MSAPHere, we investigated the genetic and epigenetic variations and their ecological correlates in R. oldhamii. Genetic and epigenetic variations were surveyed using amplified fragment length polymorphism (AFLP) and methylation-sensitive amplification polymorphism (MSAP), respectively. For MSAP, we generated two binary matrices for analyses representing two informative patterns of methylation susceptible loci (i.e. MSAP-nonMeth for non-methylation and MSAP-Meth for methylation). In the MSAP-nonMeth dataset, the loci of methylation state (i) were coded with "1" and loci of methylation states (ii) and (iii) were coded with "0". Alternatively, the loci of methylation state (ii) were coded with "1" and loci of methylation states (i) and (iii) were coded with "0" in the MSAP-Meth dataset.