Read me first, translation key for taxon names in filesDryad_readme_first.rtfStarting alignment of 13 plastomesStarting alignment of 13 plastomes (including partial Anodendron plastome), 145,064 base pairs long, Phylip format.infileStarting alignment of 12 plastomesStarting alignment of 12 plastomes (excluding partial Anodendron plastome) 144,864 base pairs long, created from the 145,064 alignment by 1) deletion of Anodendron plastome and 2) deletion of all gap and all “N” characters, Phylip format.infile_144864.reducedRaxML partition file for 13 plastome, 145,064 base pair, alignment from Partition Finder analysis.partRaxML partition file for 12 plastome, 144,864 base pair alignment from Partition Finder analysis.part_144864.reducedRaxML exclude file for 13 plastome, 145,064 base pair alignment; 1075 segments excluded based on GBLOCKS analysis.exclRaxML exclude file for 12 plastome, 144,864 base pair alignment; 1065 segments excluded based on GBLOCKS analysis.Gblocks_exclude_20131001.txtRaxML input matrix “Allsites_alltaxa”Anodendron included; 125,369 aligned base pairs, undetermined and ambiguously aligned sites identified by GBLOCKS removed; Phylip format.infileRaxML partition file for matrix “Allsites_alltaxa”Anodendron included, 125,369 aligned base pairs, undetermined and ambiguously aligned sites identified by GBLOCKS removed.partMaximum likelihood bootstrap tree for matrix “Allsites_alltaxa” analyzed with linked branch lengths.RAxML_bipartitions.resultMaximum likelihood bootstrap tree for matrix “Allsites_alltaxa” analyzed with unlinked branch lengths.RAxML_bipartitions.resultMost parsimonious tree for matrix “Allsites_alltaxa”, Winclada tree file.125369_names.treMaximum likelihood bootstrap tree for matrix “Allsites_alltaxa” with Rhabdadenia excluded, linked branch lengths.RAxML_bipartitions.resultMaximum likelihood bootstrap tree for matrix “Allsites_alltaxa” with Rhabdadenia excluded, unlinked branch lengths.RAxML_bipartitions.resultMost parsimonious trees for matrix “Allsites_alltaxa” with Rhabdadenia excluded, Winclada tree file.125369_rhab_names.treMaximum likelihood bootstrap tree for matrix “Allsites_alltaxa” with Periploca excluded, linked branch lengths.RAxML_bipartitions.resultMaximum likelihood bootstrap tree for matrix “Allsites_alltaxa” with Periploca excluded, unlinked branch lengths.RAxML_bipartitions.resultMost parsimonious trees for matrix “Allsites_alltaxa” with Periploca excluded, Winclada format.125369_peri_names.treRaxML input matrix “Allsites_noAno”RaxML input matrix “Allsites_noAno”, Anodendron excluded, 126,430 aligned base pairs, undetermined and ambiguously aligned sites identified by GBLOCKS removed, Phylip format.infileRaxML partition file for matrix “Allsites_noAno”, Anodendron excluded, 126,430 aligned base pairs, undetermined and ambiguously aligned sites identified by GBLOCKS removed.partMaximum likelihood bootstrap tree for matrix “Allsites_noAno” analyzed with linked branch lengths.RAxML_bipartitions.resultMaximum likelihood bootstrap tree for matrix “Allsites_noAno” analyzed with unlinked branch lengths.RAxML_bipartitions.resultMost parsimonious trees for matrix “Allsites_noAno”, Winclada format.126430_names.treMaximum likelihood bootstrap tree for matrix “Allsites_noAno” with Rhabdadenia excluded, branch lengths linked.RAxML_bipartitions.resultMaximum likelihood bootstrap tree for matrix “Allsites_noAno” with Rhabdadenia excluded, branch lengths unlinked.RAxML_bipartitions.resultMost parsimonious trees for matrix “Allsites_noAno” with Rhabdadenia excluded, Winclada format.126430_rhab_names.treMaximum likelihood bootstrap tree for matrix “Allsites_noAno” with Periploca excluded, branch lengths linked.RAxML_bipartitions.resultMaximum likelihood bootstrap tree for matrix “Allsites_noAno” with Periploca excluded, branch lengths unlinked.RAxML_bipartitions.resultMost parsimonious trees for matrix “Allsites_noAno” with Periploca excluded, Winclada format.126430_peri_names.treRaxML input matrix “>3.4_excluded”RaxML input matrix “>3.4_excluded”, Anodendron excluded, created from matrix “Allsites_noAno” by exclusion of sites with AIR rates > 3.4, 125,871 base pairs retained, 559 base pairs excluded, Phylip format.infileRaxML exclusion file, sites excluded from “Allsites_noAno” to create matrix “>3.4_excluded”.AIR_exclude_559_7tre.txtMaximum likelihood bootstrap tree for matrix “>3.4_excluded” analyzed with linked branch lengths.RAxML_bipartitions.resultMaximum likelihood bootstrap tree for matrix “>3.4_excluded” analyzed with unlinked branch lengths.RAxML_bipartitions.resultMost parsimonious trees for matrix “>3.4_excluded”125871_names.treRaxML input matrix “>3.3_excluded”RaxML input matrix “>3.3_excluded”, Anodendron excluded, created from matrix “Allsites_noAno” by exclusion of sites with AIR rates > 3.3, 125,487 base pairs retained, 943 base pairs excluded, Phylip format.infileRaxML exclusion file, sites excluded from “Allsites_noAno” to create matrix “>3.3_excluded”.AIR_exclude_943_7treA.txtMaximum likelihood bootstrap tree for matrix “>3.3_excluded” analyzed with linked branch lengths.RAxML_bipartitions.resultMaximum likelihood bootstrap tree for matrix “>3.3_excluded” analyzed with unlinked branch lengths.RAxML_bipartitions.resultMost parsimonious trees for matrix “>3.3_excluded”125487_names.treRaxML input matrix “>3.0_excluded”RaxML input matrix “>3.0_excluded”, Anodendron excluded, created from matrix Allsites_noAno (4b) by exclusion of sites with AIR rates > 3.0, 123,995 base pairs retained, 2,435 base pairs excluded.infileRaxML exclusion file, sites excluded from “Allsites_noAno” to create matrix “>3.0_excluded”.AIR_exclude_2435_7tre.txtMaximum likelihood bootstrap tree for matrix “>3.0_excluded” analyzed with linked branch lengths.RAxML_bipartitions.resultMaximum likelihood bootstrap tree for matrix “>3.0_excluded” analyzed with unlinked branch lengthsRAxML_bipartitions.resultMost parsimonious trees for matrix “>3.0_excluded”, Winclada tree file format123995_names.treRaxML partition file for matrix “>3.4_excluded”partRaxML partition file for matrix “>3.3_excluded”partRaxML partition file for matrix “>3.0_excluded”partRaxML input matrix, inverse exclusion matrix for matrix “>3.4_excluded”, includes only 559 sites with AIR rate >3.4, analyzed without partitionsinfileMaximum likelihood bootstrap tree for inverse exclusion matrix for matrix “>3.4_excluded”RAxML_bipartitions.resultRaxML input matrix, inverse exclusion matrix for matrix “>3.3_excluded”, includes only 943 sites with AIR rate >3.3infileMaximum likelihood bootstrap tree for inverse exclusion matrix for matrix “>3.3_excluded”RAxML_bipartitions.resultRaxML input matrix, inverse exclusion matrix for matrix “>3.0_excluded”, includes only 2,435 sites with AIR rate >3.0infileMaximum likelihood bootstrap tree for inverse exclusion matrix for matrix “>3.0_excluded”RAxML_bipartitions.resultRaxML input tree file of 105 topologies for SH test of matrix “Allsites_alltaxa”.topologies_file.treResult of SH test on matrix “Allsites_alltaxa” with linked branch lengths.tab delimited text file125369_SH_linked.txtResult of SH test on matrix “Allsites_alltaxa” with unlinked branch lengths.tab delimited text file.125369_SH_unlinked.txtAlternative topologies (105) for SH test of matrices “Allsites_noAno”, “>3.4_excluded”, “>3.3_excluded”, “>3.0_excluded”.tab delimited text file.105_topologies.txtResult of SH test on matrix “Allsites_noAno” with linked branch lengths.SH_linked_2013_10_01.txtResult of SH test on matrix “Allsites_noAno” with unlinked branch lengths.SH_unlinked_2013_10_01.txtResult of SH test on matrix “>3.4_excluded” with linked branch lengths.SH_linked_AIR559excl_3.4_7tr.txtResult of SH test on matrix “>3.4_excluded” with unlinked branch lengths.SH_unlinked_AIR559excl_3.4_7tr.txtResult of SH test on matrix “>3.3_excluded” with linked branch lengths.Tab delimited text file.SH_linked_AIR943excl_3.3_7tr.txtResult of SH test on matrix “>3.3_excluded” with unlinked branch lengths.SH_unlinked_AIR943excl_3.3_7tr.txtResult of SH test on matrix “>3.0_excluded” with linked branch lengths.tab delimited text file.SH_linked_AIR2435excl_3_7tr.txtResult of SH test on matrix “>3.0_excluded” with unlinked branch lengths.SH_unlinked_AIR2435excl_3_7tr.txtAIR analysis output, based on 7 treesAIR analysis output, average rates for each site in matrix “Allsites_noAno” for 7 trees with p>0.05 from SH test of “Allsites_noAno” with linked branch lengths. Tab delimited text file.AIR7trees.txtAIR analysis output, based on 12 treesAIR analysis output, average rates for each site in matrix “Allsites_noAno” for 12 trees with p>0.05 from SH test of “Allsites_noAno” with unlinked branch lengths. Tab delimited text file.AIR12trees.txtSites removed by GBLOCKS and AIR sorted by genomic featureGenomic features associated with sites excluded from the alignment of 12 full plastomes by GBLOCKS to produce matrix Allsites_noAno and by AIR to produce matrix >3.0_excluded, data for Figure 4. Tab delimited text file.summary_sorted_for_graph.txtMorphological character matrix: pollen aggregation.character_matrix.outAnodendron oblongifolium partial plastome assemblyfasta fileAnodendron_oblongifolium_plastome_assembly.fasta

Crown clade Apocynaceae comprise seven primary lineages of lianas, shrubs, and herbs with a diversity of pollen aggregation morphologies including monads, tetrads, and pollinia, making them an ideal group for investigating the evolution and function of pollen packaging. Traditional molecular systematic approaches utilizing small amounts of sequence data have failed to resolve relationships along the spine of the crown clade, a likely ancient rapid radiation. The previous best estimate of the phylogeny was a five-way polytomy, leaving ambiguous the homology of aggregated pollen in two major lineages, the Periplocoideae, which possess pollen tetrads, and the milkweeds (Secamonoideae plus Asclepiadoideae), which possess pollinia. To assess whether greatly increased character sampling would resolve these relationships, a plastome sequence data matrix was assembled for 13 taxa of Apocynaceae, including nine newly generated complete plastomes, one partial new plastome, and three previously reported plastomes, collectively representing all primary crown clade lineages and outgroups. The effects of phylogenetic noise, long-branch attraction, and model selection (linked versus unlinked branch lengths among data partitions) were evaluated in a hypothesis-testing framework based on Shimodaira–Hasegawa tests. Discrimination among alternative crown clade resolutions was affected by all three factors. Exclusion of the noisiest alignment positions and topologies influenced by long-branch attraction resulted in a trichotomy along the spine of the crown clade consisting of Rhabdadenia + the Asian clade, Baisseeae + milkweeds, and Periplocoideae + the New World clade. Parsimony reconstruction on all optimal topologies after noise exclusion unambiguously supports parallel evolution of aggregated pollen in Periplocoideae (tetrads) and milkweeds (pollinia). Our phylogenomic approach has greatly advanced the resolution of one of the most perplexing radiations in Apocynaceae, providing the basis for study of convergent floral morphologies and their adaptive value.