Understanding the consequences of exotic diseases on native forests is important to evolutionary ecology and conservation biology because exotic pathogens have drastically altered US eastern deciduous forests. Cornus florida L. (flowering dogwood tree) is one such species facing heavy mortality. Characterizing the genetic structure of C. florida populations and identifying the genetic signature of adaptation to dogwood anthracnose (an exotic pathogen responsible for high mortality) remains vital for conservation efforts. By integrating genetic data from genotype-by-sequencing (GBS) of 289 trees across the host species range and distribution of disease, we evaluated the spatial patterns of genetic variation and population genetic structure of C. florida and compared the pattern to the distribution of dogwood anthracnose. Using GWAS and gradient forest analysis, we identified genetic loci under selection and associated with ecological and diseased regions. The results revealed signals of weak genetic differentiation of three or more subgroups nested within two clusters—explaining up to 2-6% of genetic variation. The groups largely corresponded to the regions within and outside the eastern Hot-Continental ecoregion, which also overlapped with areas within and outside the main distribution of dogwood anthracnose. The fungal sequences contained in the GBS data of sampled trees bolstered visual records of disease at sampled locations and were congruent with the reported range of D. destructiva, suggesting fungal sequences within host genomic data were informative for detecting or predicting disease. The genetic diversity between populations at diseased vs. disease-free sites across the range of C. florida showed no significant difference. We identified 72 SNPs from 68 loci putatively under selection, some of which exhibited abrupt turnover in allele frequencies along the borders of the Hot-Continental ecoregion and the range of dogwood anthracnose. One such candidate SNP was independently identified in two prior studies as a possible L-type lectin-domain containing receptor kinase. While diseased and disease-free areas do not significantly differ in genetic diversity, overall there are slight trends to indicate marginally smaller amounts of genetic diversity in disease-affected areas. Our results were congruent with previous studies that were based on a limited number of genetic markers in revealing high genetic variation and weak population structure in C. florida.

Population genetic datasets and sequences of GBS-tags per sample were exported as Fasta, Genepop, and PLINK files after processing GBS data via STACKS (version 1.35).

Various population genetic programs can process, analyze, or convert Genepop (.gen) or PLINK (.ped and .map) file formats, and sequences in Fasta (.fa) file may be searched against existing databases to find similar sequences and functional annotations. For aligning GBS-tags back to draft genomes of Cornus florida, Erysiphe pulchra, and Discula destructiva please contact original owners for latest version of draft genomes (see acknowledgement section of Pais et al. 2020).