Evolutionary insights from de novo transcriptome assembly and SNP discovery in California white oaks

Background: Reference transcriptomes provide valuable resources for understanding evolution within and among species. We de novo assembled and annotated a reference transcriptome for Quercus lobata and Q. garryana and identified single-nucleotide polymorphisms (SNPs) to provide resources for forest genomicists studying this ecologically and economically important genus. We further performed preliminary analyses of genes important in interspecific divergent (positive) selection that might explain ecological differences among species, estimating rates of nonsynonymous to synonymous substitutions (d(N)/d(S)) and Fay and Wu's H. Functional classes of genes were tested for unusually high dN/dS or low H consistent with divergent positive selection. Results: Our draft transcriptome is among the most complete for oaks, including 83,644 contigs (23,329 >= 1 kbp), 14,898 complete and 13,778 partial gene models, and functional annotations for 9,431 Arabidopsis orthologs and 19,365 contigs with Pfam hits. We identified 1.7 million possible sequence variants including 1.1 million high-quality diallelic SNPs - among the largest sets identified in any tree. 11 of 18 functional categories with significantly elevated d(N)/d(S) are involved in disease response, including 50+ genes with d(N)/d(S) > 1. Other high-d(N)/d(S) genes are involved in biotic response, flowering and growth, or regulatory processes. In contrast, median d(N)/d(S) was low (0.22), suggesting that purifying selection influences most genes. No functional categories have unusually low H. Conclusions: These results offer preliminary support for the hypothesis that divergent selection at pathogen resistance are important factors in species divergence in these hybridizing California oaks. Our transcriptome provides a solid foundation for future studies of gene expression, natural selection, and speciation in Quercus.