Target capture methods offer insight into the evolution of rapidly diverged taxa and resolve allopolyploid homeologs in the fern genus Polypodium s.s.

Like many fern lineages comprising reticulate species complexes, Polypodium s.s. (Polypodiacaeae) has a history shaped by rapid diversification, hybridization, and polyploidy that poses substantial challenges for phylogenetic inference with plastid and single-locus nuclear loci. Using target capture probes for 408 nuclear loci developed by the GOFlag project and a custom bioinformatic pipeline, SORTER, we constructed multi-locus nuclear datasets for diploid temperate and Mesoamerican species of Polypodium and five allotetraploid species belonging to the well-studied Polypodium vulgare complex. SORTER employs a clustering approach to separate putatively paralogous copies of targeted loci into orthologous matrices and haplotype phasing to infer allopolyploid haplotypes across loci, resulting in datasets amenable to both concatenated maximum likelihood and multi-species coalescent phylogenetic analyses. By comparing phylogenies derived from maximum likelihood and multi- species coalescent analyses of unphased and phased datasets, as well as evaluating discordance among gene trees and species trees, we recover support for incomplete lineage sorting within Polypodium s.s., novel relationships among diploid taxa of the Polypodium vulgare complex and its Mesoamerican sister clade, and the placement of several Polypodium species within other genera. Additionally, we were able to infer well-supported phylogenies that identified the hypothesized progenitors of the allotetraploid species, indicating that SORTER is an effective and accurate tool for reconstructing homeolog haplotypes of allopolyploids in fern taxa and other non-model organisms from target capture data.Trees can be viewed with free FigTree software, and alignments can be observed with text editing software or Mesquite software.The resulting alignments and phylogenetic trees (Supplementals C and D) were built from target-capture data generated by the Geneology of Flagellate Plants Project (GoFlag). All raw data was sourced from various researchers that submitted samples to GoFlag with permission from respective parties. The resulting trees and alignments were processed by the SORTER pipeline (https://github.com/JonasMendez/SORTER). Supplementals C and D also include summary CSV tables for distribution of loci across samples for each dataset. Included are two supplemental texts describing the SORTER pipeline (Supplemental A) as well as effect of clustering on sequence copy number from Stage 1 of the SORTER pipeline (Supplemental B). Supplemental Table S1 summarizes read and sequence statistics for each sample used in the study