Emerging plant pathogens have been increasing exponentially over the last century. To address this issue, it is critical to determine whether these pathogens are native to ecosystems or have been recently introduced. Understanding the ecological and evolutionary processes fostering emergence can help to manage their spread and predict epidemics/epiphytotics. Using restriction site-associated DNA sequencing data, we studied genetic relationships, pathways of spread, and evolutionary history of Phellinus noxius, an emerging root-rotting fungus of unknown origin, in eastern Asia, Australia, and the Pacific Islands. We analyzed patterns of genetic variation using Bayesian inference, maximum likelihood phylogeny, populations splits and mixtures measuring correlations in allele frequencies and genetic drift, and finally applied coalescent based theory using Approximate Bayesian computation (ABC) with supervised machine learning. Population structure analyses revealed five genetic groups with signatures of complex recent and ancient migration histories. The most probable scenario of ancient pathogen spread is movement from an unsampled population to Malaysia and the Pacific Islands, with subsequent spread to Taiwan and Australia. Furthermore, ABC analyses indicate P. noxius spread occurred thousands of generations ago, contradicting previous assumptions that this pathogen was recently introduced to multiple geographic regions. Our results suggest that recent emergence of P. noxiusin eastern Asia, Australia, and the Pacific Islands is likely driven by anthropogenic and natural disturbances, such as deforestation, land-use change, severe weather events, and/or introduction of exotic plants. This study provides a novel example of applying genome-wide allele frequency data to unravel dynamics of pathogen emergence under changing ecosystem conditions.

The dataset contains three input files obtained from raw RADseq reads with Stacks v.2 to generate results in our paper. It also contains configuration files of five DIYABC analyses performed with DIYABC Random Forest v.1.

Users will need sequence data () and some knowledge of linux.