Phylogenomics in the Hard Pines (Pinus subsection Ponderosae; Pinaceae) Confirms Paraphyly in Pinus ponderosa, and Places Pinus jeffreyi with the California Big Cone Pines

Abstract— We sampled 130 individuals (2 to 25 per taxon) of Pinus subsections Ponderosae and Sabinianae . Nucleotide sequences were obtained by targeting 703 low copy nuclear genes. From the unenriched portion of the short reads, we assembled nearly complete plastome nucleotide sequences. We used 600 nuclear genes and the plastome sequences to create phylogenies and species trees that we compared to evaluate cytonuclear concordance and reticulation. We found that Pinus jeffreyi belongs with Pinus subsect. Sabinianae based on morphological synapomorphies as well as strong molecular phylogenetic support. Pinus ponderosa sensu lato is paraphyletic, and we suggest treatment as three species: P. ponderosa sensu stricto (with P. ponderosa var. ponderosa , P. ponderosa var. benthamiana , and P. ponderosa var. washoensis ), P. scopulorum , and P. brachyptera . The persistence of lineages with the footprints of ancient nuclear introgression (labeled bpw in clade N4) and chloroplast capture (labeled bpw in clade P1) should caution species identification in Pinus subsection Ponderosae based on limited molecular data. The hybrid frequency was low based on cytonuclear discordance, and the persistence of an ancient P1 plastid clade is a better explanation than hybridization between P. ponderosa and P. jeffreyi for unexpected plastid associations in the western Sierra Nevada, USA. We identified a new potential zone of ancient admixture between P. ponderosa and P. scopulorum in Idaho, USA. Some populations of P. arizonica , P. brachyptera , P. engelmannii , and P. scopulorum in the USA are more closely related to taxa with distributions limited to Mexico than they are to each other. To integrate phylogeny and taxonomy, future work should sample widely in Mexico and the USA, score morphological characters (including seedling characters from the known seed parent), on the same individual as used for molecular data, and use methods that are based on individuals rather than population frequencies.