Supporting Information - PNAS · Penn State Huck Institute of the Life Sciences or Beckman Coulter...
7
Supporting Information Estep et al. 10.1073/pnas.1404177111 SI Materials and Methods Taxon Sampling. We sampled 114 accessions representing two outgroup species (Paspalum malacophyllum and Plagiantha tenella), two species of Arundinella, and 100 species of Andropogoneae in 40 genera. Plant material came from our own collections, the US Department of Agriculture (USDA) Germplasm Resources In- formation Network (GRIN), the Kew Millenium Seed Bank, and material sent by colleagues. All plants acquired as seeds (e.g., those from USDA and from Kew) were grown to flowering in the greenhouse at the University of Missouri-St. Louis to verify iden- tification. Vouchers are listed in Table S1. Sequencing and Processing. Total genomic DNA was extracted using a modified cetyl triethylammonium bromide (CTAB) procedure (1) or Qiagen DNeasy kits, following the manu- facturer’s protocol (Qiagen) (2). Five regions of four loci were PCR amplified following Estep et al. (3). The loci are Aberrant panicle organization1 (apo1), Dwarf8(d8), two exons of Erect panicle2 (ep2), and Retarded palea1 (rep1). In sorghum (a dip- loid) apo1 and d8 are on chromosomes 1 and 10, respectively. ep2 and rep1 are on chromosome 2, 14.5 Mbp apart; based on estimated recombination frequency in euchromatic regions, this distance could be ca. 50 cM (4). Thus, the four markers are unlinked. Each marker has homologs on two chromosomes in maize, as expected in a tetraploid. Importantly for this study, we found no evidence that these loci are lost after polyploidization; in known polyploids, paralogues are found consistently as long as sequence depth is sufficient. PCR products were gel purified, cloned using pGEM-T easy kits, and transformed into JM109 high-efficiency competent cells, following the manufacturers’ protocols (Promega) or using a Topo-4 cloning vector and transformed into One Shot Top 10 cells (Invitrogen). At least eight positive clones for each PCR product were sequenced in both directions using universal pri- mers (M13, Sp6, or T7) on an ABI3730 DNA sequencer at the Penn State Huck Institute of the Life Sciences or Beckman Coulter Genomics. Chromatogram files were trimmed of vector and low quality sequences manually or using Geneious Pro-5.5.6 (BioMatters), and reverse and forward sequences for each clone were assembled. Only clones with 80% or more double-stranded sequence were used for downstream analyses. Internal primers were designed as necessary for loci over 1,000 bp long (D8 and Ep2 exon7). All good quality contigs for each sample were then aligned using Geneious, and primer sequences were removed. Recombinants were initially identified by eye and then con- firmed with networks in SplitsTree (5), and removed from the alignment. Use of computational methods alone missed many recombinant sequences that were readily identifiable by eye. Occasionally, it was difficult to determine which sequences were genomic and which were PCR recombinants; in these cases, we compared the problematic sequences with unambiguous se- quences from other species to determine the nonrecombined sequences. Singletons were identified with MEGA (6) and were interpreted as PCR errors. Sequences were translated and aligned using MUSCLE, as implemented in Geneious Pro-5.5.6. Data Matrix Assembly. The dataset for each locus consisted of numerous redundant clones. To reduce the number of sequences to one per paralogue per locus, preliminary phylogenetic analyses were conducted for each marker in RAxML (7) including all clones for all taxa. Clones that formed a clade in preliminary analyses and that differed by fewer than five nucleotides were inferred to represent a single locus and were combined into a majority-rule consensus sequence. Clones that did not meet these criteria were kept separate through another round of RAxML analyses. We identified clades with a bootstrap value ≥ 50 that comprised the same accession for each locus. Accessions in these clades were reduced to a single majority-rule consensus sequence using the perl script clone_reducer (github.com/ mrmckain). Gene trees were estimated using RAxML v.7.3.0 with the GTR + Γ model and 500 bootstrap replicates for each locus. We used individual gene-tree topologies as a guide to identify and concatenate paralogues from the same genome for each acces- sion. If a polyploid had two paralogues in each gene tree, and one paralogue was always sister to a particular diploid or other poly- ploid, then we inferred that those paralogues represented the same genome and used them to create a concatenated sequence. If a locus did not have a sequence congruent to the topology, then the locus was marked as missing data. Five datasets were assem- bled. One included only accessions with full sampling of all loci for all genomes, a second included accessions that had four out of five markers, a third, three out of five, and so forth. Preliminary trees for all datasets were congruent, but those trees in which some genomes of some taxa were represented by only one or two se- quences were less well supported. The results presented here are based on the dataset with a minimum of three out of five loci for each taxon, which maximized species inclusion while still providing enough information for robust results. All five datasets are depos- ited at Dryad (datadryad.org/resource/doi:10.5061/dryad.5kf38). Phylogenetic Analysis, Divergence Time, and Diversification Estimates. Concatenated trees were reconstructed using both maximum likelihood (ML) and Bayesian approaches and rooted at Paspalum. The ML tree was estimated with RAxML using the GTR + Γ model and 500 bootstrap replicates. The Bayesian tree was esti- mated using MrBayes v.3.2.1 (8) using a gamma model with six discrete categories. Two independent runs with 50 million gen- erations each were sampled every 1,000 generations. Convergence of the separate runs was verified using AWTY (9). Divergence times were estimated using BEAST 1.7.5 (10) on the CIPRES Science Gateway (11). The concatenated analysis was run for 100 million generations sampling every 1,000 under the GTR + Γ model with six gamma categories. The tree prior used the birth-death with incomplete sampling model (12), with the starting tree being estimated using unweighted pair group method with arithmetic mean (UPGMA). The site model fol- lowed an uncorrelated lognormal relaxed clock (13). The anal- ysis was rooted to Paspalum, with the age of the root estimated as a normal distribution describing an age of 25.5 ± 5 million y (14). Convergence statistics were estimated using Tracer v.1.5 (15) after a burn-in of 50,000 sampled generations. Chain con- vergence was estimated to have been met when the effective sample size was greater than 200 for all statistics. Ultimately, 10,000 trees were used in TreeAnnotator v.1.7.5 to produce the maximum clade credibility tree and to determine the 95% highest posterior density (HPD) for each node. The final tree was drawn using FigTree v.1.4.0 (tree.bio.ed.ac.uk/software/ figtree/). Tests for shifts in the underlying model of diversification were conducted using Bayesian Analysis of Macroevolutionary Mix- tures (BAMM) (16); priors were chosen as recommended using the function setBAMMpriors. Analyses were conducted on a tree constructed by pruning the BEAST tree to leave only a single Estep et al. www.pnas.org/cgi/content/short/1404177111 1 of 7 Supporting Information Corrected , Dec mber e 28 2015
Transcript of Supporting Information - PNAS · Penn State Huck Institute of the Life Sciences or Beckman Coulter...
Supporting InformationEstep et al. 10.1073/pnas.1404177111SI Materials and MethodsTaxon Sampling. We sampled 114 accessions representing twooutgroup species (Paspalum malacophyllum and Plagiantha tenella),two species of Arundinella, and 100 species of Andropogoneae in40 genera. Plant material came from our own collections, the USDepartment of Agriculture (USDA) Germplasm Resources In-formation Network (GRIN), the Kew Millenium Seed Bank, andmaterial sent by colleagues. All plants acquired as seeds (e.g., thosefrom USDA and from Kew) were grown to flowering in thegreenhouse at the University of Missouri-St. Louis to verify iden-tification. Vouchers are listed in Table S1.
Sequencing and Processing. Total genomic DNA was extractedusing a modified cetyl triethylammonium bromide (CTAB)procedure (1) or Qiagen DNeasy kits, following the manu-facturer’s protocol (Qiagen) (2). Five regions of four loci werePCR amplified following Estep et al. (3). The loci are Aberrantpanicle organization1 (apo1), Dwarf8 (d8), two exons of Erectpanicle2 (ep2), and Retarded palea1 (rep1). In sorghum (a dip-loid) apo1 and d8 are on chromosomes 1 and 10, respectively.ep2 and rep1 are on chromosome 2, 14.5 Mbp apart; based onestimated recombination frequency in euchromatic regions, thisdistance could be ca. 50 cM (4). Thus, the four markers areunlinked. Each marker has homologs on two chromosomes inmaize, as expected in a tetraploid. Importantly for this study, wefound no evidence that these loci are lost after polyploidization;in known polyploids, paralogues are found consistently as long assequence depth is sufficient.PCR products were gel purified, cloned using pGEM-T easy
kits, and transformed into JM109 high-efficiency competent cells,following the manufacturers’ protocols (Promega) or usinga Topo-4 cloning vector and transformed into One Shot Top 10cells (Invitrogen). At least eight positive clones for each PCRproduct were sequenced in both directions using universal pri-mers (M13, Sp6, or T7) on an ABI3730 DNA sequencer at thePenn State Huck Institute of the Life Sciences or BeckmanCoulter Genomics. Chromatogram files were trimmed of vectorand low quality sequences manually or using Geneious Pro-5.5.6(BioMatters), and reverse and forward sequences for each clonewere assembled. Only clones with 80% or more double-strandedsequence were used for downstream analyses. Internal primerswere designed as necessary for loci over 1,000 bp long (D8 andEp2 exon7). All good quality contigs for each sample were thenaligned using Geneious, and primer sequences were removed.Recombinants were initially identified by eye and then con-firmed with networks in SplitsTree (5), and removed from thealignment. Use of computational methods alone missed manyrecombinant sequences that were readily identifiable by eye.Occasionally, it was difficult to determine which sequences weregenomic and which were PCR recombinants; in these cases, wecompared the problematic sequences with unambiguous se-quences from other species to determine the nonrecombinedsequences. Singletons were identified with MEGA (6) and wereinterpreted as PCR errors. Sequences were translated andaligned using MUSCLE, as implemented in Geneious Pro-5.5.6.
Data Matrix Assembly. The dataset for each locus consisted ofnumerous redundant clones. To reduce the number of sequencesto one per paralogue per locus, preliminary phylogenetic analyseswere conducted for each marker in RAxML (7) including allclones for all taxa. Clones that formed a clade in preliminaryanalyses and that differed by fewer than five nucleotides were
inferred to represent a single locus and were combined intoa majority-rule consensus sequence. Clones that did not meetthese criteria were kept separate through another round ofRAxML analyses. We identified clades with a bootstrap value ≥50 that comprised the same accession for each locus. Accessionsin these clades were reduced to a single majority-rule consensussequence using the perl script clone_reducer (github.com/mrmckain).Gene trees were estimated using RAxML v.7.3.0 with the
GTR + Γ model and 500 bootstrap replicates for each locus.We used individual gene-tree topologies as a guide to identify andconcatenate paralogues from the same genome for each acces-sion. If a polyploid had two paralogues in each gene tree, and oneparalogue was always sister to a particular diploid or other poly-ploid, then we inferred that those paralogues represented thesame genome and used them to create a concatenated sequence.If a locus did not have a sequence congruent to the topology, thenthe locus was marked as missing data. Five datasets were assem-bled. One included only accessions with full sampling of all loci forall genomes, a second included accessions that had four out of fivemarkers, a third, three out of five, and so forth. Preliminary treesfor all datasets were congruent, but those trees in which somegenomes of some taxa were represented by only one or two se-quences were less well supported. The results presented here arebased on the dataset with a minimum of three out of five loci foreach taxon, which maximized species inclusion while still providingenough information for robust results. All five datasets are depos-ited at Dryad (datadryad.org/resource/doi:10.5061/dryad.5kf38).
Phylogenetic Analysis, Divergence Time, and Diversification Estimates.Concatenated trees were reconstructed using both maximumlikelihood (ML) and Bayesian approaches and rooted at Paspalum.The ML tree was estimated with RAxML using the GTR + Γmodel and 500 bootstrap replicates. The Bayesian tree was esti-mated using MrBayes v.3.2.1 (8) using a gamma model with sixdiscrete categories. Two independent runs with 50 million gen-erations each were sampled every 1,000 generations. Convergenceof the separate runs was verified using AWTY (9).Divergence times were estimated using BEAST 1.7.5 (10) on
the CIPRES Science Gateway (11). The concatenated analysiswas run for 100 million generations sampling every 1,000 underthe GTR + Γ model with six gamma categories. The tree priorused the birth-death with incomplete sampling model (12), withthe starting tree being estimated using unweighted pair groupmethod with arithmetic mean (UPGMA). The site model fol-lowed an uncorrelated lognormal relaxed clock (13). The anal-ysis was rooted to Paspalum, with the age of the root estimatedas a normal distribution describing an age of 25.5 ± 5 million y(14). Convergence statistics were estimated using Tracer v.1.5(15) after a burn-in of 50,000 sampled generations. Chain con-vergence was estimated to have been met when the effectivesample size was greater than 200 for all statistics. Ultimately,10,000 trees were used in TreeAnnotator v.1.7.5 to producethe maximum clade credibility tree and to determine the95% highest posterior density (HPD) for each node. The finaltree was drawn using FigTree v.1.4.0 (tree.bio.ed.ac.uk/software/figtree/).Tests for shifts in the underlying model of diversification were
conducted using Bayesian Analysis of Macroevolutionary Mix-tures (BAMM) (16); priors were chosen as recommended usingthe function setBAMMpriors. Analyses were conducted on a treeconstructed by pruning the BEAST tree to leave only a single
Estep et al. www.pnas.org/cgi/content/short/1404177111 1 of 7
Supporting Information Corrected , Dec mbere 28 2015
paralogue (genome) per species, as well as on the unpruned trees.Analyses were run for 10 million and 100 million generations;10 million provided an adequate effective sample size for bothrate shifts and log-likelihoods (>>500), and additional generationsdid not change the results. Incomplete taxon sampling was ac-commodated by assigning a backbone sampling fraction of 0.5,corresponding to our sample of 50% of the genera, and assigningspecific fractions for the sample for each clade. Preliminaryanalyses using a backbone sample of 0.08 or using a global esti-mate of sampling intensity produced largely similar results.Differences in speciation rate for polyploids versus diploids
were estimated using the Binary State Speciation and Extinction(BiSSE) model (17) as implemented in Mesquite (18) and di-versitree (19). Species were coded as allopolyploid or diploid,and the characters were mapped on the pruned BEAST tree.Adjustment for incomplete sampling was implemented in di-versitree, using a sample of 0.08 and also 0.5 to correspond to theanalyses in BAMM. Analyses were run for 100,000 generations.Values from an unconstrained analysis were compared withthose of an analysis in which speciation rates were constrained tobe equal.
Genome-Size Estimation. Genome size was measured by flowcytometry at the Flow Cytometry and Imaging Core laboratoryat the Benaroya Research Institute, Virginia Mason ResearchCenter. Genome sizes were measured a minimum of four times
using standard methods, and the values were averaged (20). Pi-cograms of DNA/2C cell were converted to megabasepairs(Mbp)/1C by multiplying the average experimental value by980 Mbp per 1 picogram of DNA and dividing by two.
Estimating the Number of Allopolyploidization Events. To obtaina minimum estimate of the number of allopolyploidy events, welooked for a clear phylogenetic signal of allopolyploidy, in theform of a multiple-labeled gene tree (Fig. S1). This methodprovides unambiguous evidence of allopolyploidy, but willoverlook genetic and taxonomic autopolyploidy, allopolyploidybetween two very similar species or populations of the samespecies, and allopolyploidy for which our sequencing was notsufficiently deep to retrieve paralogues. Thus, the number ofevents estimated here is a robust minimum estimate. We alsoreport genome size for many of the sequenced specimens (Figs. S2and S3). Although genome size gives a rough approximation ofploidy, it cannot distinguish between polyploids and plants whosegenomes have expanded via transposon amplification (21). Wedid not use published data on chromosome numbers because theliterature on these numbers is unreliable; many counts lackphotodocumentation or voucher specimens and thus cannot beverified as to number or species identity. In addition, manyspecies have several numbers reported; these may represent realvariation or errors.
1. Doyle JJ, Doyle JL (1987) A rapid DNA isolation procedure for small quantities of freshleaf tissue. Phytochem Bull 19:11–15.
2. Teerawatananon A, Jacobs SWL, Hodkinson TR (2011) Phylogenetics of Panicoideae(Poaceae) based on chloroplast and nuclear DNA sequences. Telopea (Syd) 13:115–142.
3. Estep MC, Vela Diaz DM, Zhong J, Kellogg EA (2012) Eleven diverse nuclear-encodedphylogenetic markers for the subfamily Panicoideae (Poaceae). Am J Bot 99(11):e443–e446.
4. Mace ES, Jordan DR (2011) Integrating sorghumwhole genome sequence informationwith a compendium of sorghum QTL studies reveals uneven distribution of QTL andof gene-rich regions with significant implications for crop improvement. Theor ApplGenet 123(1):169–191.
5. Huson DH, Bryant D (2006) Application of phylogenetic networks in evolutionarystudies. Mol Biol Evol 23(2):254–267.
6. Tamura K, et al. (2011) MEGA5: Molecular evolutionary genetics analysis using max-imum likelihood, evolutionary distance, and maximum parsimony methods. Mol BiolEvol 28(10):2731–2739.
7. Stamatakis A (2006) RAxML-VI-HPC: Maximum likelihood-based phylogenetic analy-ses with thousands of taxa and mixed models. Bioinformatics 22(21):2688–2690.
8. Ronquist F, Huelsenbeck JP (2003) MrBayes 3: Bayesian phylogenetic inference undermixed models. Bioinformatics 19(12):1572–1574.
9. Nylander JAA, Wilgenbusch JC, Warren DL, Swofford DL (2008) AWTY (are we thereyet?): A system for graphical exploration of MCMC convergence in Bayesian phylo-genetics. Bioinformatics 24(4):581–583.
10. Drummond AJ, Suchard MA, Xie D, Rambaut A (2012) Bayesian phylogenetics withBEAUti and the BEAST 1.7. Mol Biol Evol 29(8):1969–1973.
11. Miller MA, Pfeiffer W, Schwartz T (2010) Creating the CIPRES science gateway forinference of large phylogenetic trees. Proceedings of the 2010 Gateway ComputingEnvironments Workshop (GCE) (Institute of Electrical and Electronics Engineers, NewYork), pp 1–8.
12. Stadler T (2009) On incomplete sampling under birth-death models and connectionsto the sampling-based coalescent. J Theor Biol 261(1):58–66.
13. Drummond AJ, Ho SYW, Phillips MJ, Rambaut A (2006) Relaxed phylogenetics anddating with confidence. PLoS Biol 4(5):e88.
14. Vicentini A, Barber JC, Giussani LM, Aliscioni SS, Kellogg EA (2008) Multiple coincidentorigins of C4 photosynthesis in the Mid- to Late Miocene. Glob Change Biol 14:2963–2977.
15. Rambaut A, Suchard MA, Xie D, Drummond AJ (2013) Tracer, Version 1.5. Available attree.bio.ed.ac.uk/software/tracer/.
16. Rabosky DL (2014) Automatic detection of key innovations, rate shifts, and diversity-dependence on phylogenetic trees. PLoS ONE 9(2):e89543.
17. Maddison WP, Midford PE, Otto SP (2007) Estimating a binary character’s effect onspeciation and extinction. Syst Biol 56(5):701–710.
18. Maddison WP, Maddison DR (2011) Mesquite: A Modular System for EvolutionaryAnalysis, Version 2.75. Available at mesquiteproject.org.
19. FitzJohn RG (2012) Diversitree: Comparative phylogenetic analysis of diversification inR. Methods Ecol. Evol. 3:1084–1092.
20. Arumuganathan K, Earle ED (1991) Estimation of nuclear DNA amounts of plants byflow cytometry. Plant Mol Biol Rep 9:229–241.
21. Kellogg EA, Bennetzen JL (2004) The evolution of nuclear genome structure in seedplants. Am J Bot 91(10):1709–1725.
Estep et al. www.pnas.org/cgi/content/short/1404177111 2 of 7
ABCDE
diploids, divergent evolution
AE
tetraploid
Evolutionary history Gene tree, full sampling
ABCDE
AE
AE
Examples of gene trees, incomplete sampling
BE
AE
AE
AE1
AE4AE3AE2
ABCDE
AE1
AE4AE3AE2
AE1
AE4AE3AE2
BAE
AE
AE1
AE4AE1AE4
BCE
AE1
AE4AE3
AE1
AE4AE2
No
dive
rsifi
catio
n a
fter
pol
yplo
idy
Div
ersi
ficat
ion
afte
r pol
yplo
idy
ABCDE
hybridization
Fig. S1. Gene-tree patterns resulting from hybridization and polyploidy (genetic allopolyploidy). (Left) The actual evolutionary history of divergent evolutionat the diploid level, followed by hybridization between taxon A and taxon E (reticulation) and formation of a tetraploid. Subsequent speciation may (Lower) ormay not (Upper) occur. (Center) The double-labeled gene trees that will be produced by each pattern of evolution. Tetraploids appear twice in the tree; the Agenome (AE) will appear with its A genome progenitor and the E genome (AE) with its E genome progenitor. (Right) Some of the many possible gene treesthat will result from incomplete taxon sampling. In this study, any of these patterns were taken as indication of allopolyploidy.
Estep et al. www.pnas.org/cgi/content/short/1404177111 3 of 7
posterior probabtility
1
0.5163
0.03
Pseudosorghum_fasciculare_164
Ischaemum_afrum_020
Phacelurus_digitatus_025
Chrysopogon_serrulatus_056
Arundinella_nepalensis_115
Rottboellia_cochinchinensis_046
Dimeria_fuscescens_137
Sorghum_bicolor_1_037
Imperata_cylindrica_154
Polytrias_indica_073
Microstegium_vimineum_124
Dimeria_sinensis_139
Apocopis_courtllumensis_127
Saccharum_narenga_166
Dimeria_fuscescens_137
Ischaemum_santapaui_021
Coix_gasteenii_134
Tripsacum_dactyloides_041
Kerriochloa_siamensis_157
Pogonatherum_paniceum_072
Apluda_mutica_100
Miscanthus_ecklonii_089
Ischaemum_santapaui_021
Erianthus_arundinaceum_165
Arundinella_hirta-099
Apocopis_siamensis_129
Dimeria_ornithopoda_138
Sorghum_nervosum_039
Arthraxon_lanceolatus_131
Chrysopogon_zizanoides_104
Apocopis_intermedius_128
Polytrias_indica_073
Eulalia_aurea_098
Tripsacum_dactyloides_041
Sorghastrum_nutans_035
Miscanthus_ecklonii_089
Apocopis_intermedius_128
Sorghum_bicolor_182
Coix_lacryma-jobi_053
Eriochrysis_pallida_016
Chrysopogon_gryllus_006
Pogonatherum_crinitum_160
Germainia_capitata_145
Sorghastrum_elliotti_187
Chasmopodium_caudatum_120
Plagiantha_tenella_060
Paspalum_malacophyllum_058
Zea_maysA_997
Erianthus_ravennae_110
Spodiopogon_sibiricus_169
Miscanthus_sinensis_068
Thelepogon_elegans_170
Zea_maysB_998
Chasmopodium_caudatum_120
Ischaemum_muticum_156
Sorghum_bicolor_g_999
Ischaemum_rugosum_186
Chionachne_massiei_133
Sorghum_bicolor_2_038
Ischaemum_indicum_155
Andropterum_stolzii_050
Arthraxon_prionodes_103
Miscanthus_sinensis_068
Phacelurus_huillensis_024
Saccharum_narenga_166
Sehima_nervosum_034
Chrysopogon_gryllus_006
0.96 / 36
0.67 / 12
1 / 100
0.99 / 94
1/100
0.86 / 42
1 / 961 / 98
1/100
0.82 / 291 / 71
1 / 82
0.95 / 62
0.92 / 36
1/100
0.83 / 16
0.98 / 59
1/99
0.73 / 18
1 / 100
0.72 / 17
1 / 32
0.81 / 26
0.58 / 25
1.0 / 73
0.99 / 54
0.85 / 23
1/100
0.88 / 480.95 / 43
1/100
1/100
1
0.79 / 34
1 / 100
1 / 71
0.94 / 66
0.90 / 64
0.92 / 42
0.99 / 89
1/100
0.99 / 64
0.83 / 29
0.98 / 71
1/100
core Andropogoneae(Andropogoninae) (Figure S2)
1
2
3
4
5
6
7
955
3449
1878
677
1805
2597
960
717
2171
1760
2819
1751
2185
1058
1975
1675
1563
1161
8
1 / 100 1 / 100
1 / 100
1 / 100
1 / 100
1 / 1001 / 100
1 / 100
1/100
1 / 100
1 / 100
1 / 100
1 / 100
1 / 1001 / 100
1 / 100
1 / 100
Fig. S2. Bayesian phylogeny of Andropogoneae, species not included in the “core Andropogoneae.” Numbers on branches are posterior probability (pp)/ML.Branch color reflects pp values. Genome-size estimates as Mbp/1C nucleus are indicated after species names, where available. Numbered polyploidizationevents correspond to those in Fig. 1, Figs. S2 and S3, and Table 1. Accession numbers follow the species name for all accessions.
Estep et al. www.pnas.org/cgi/content/short/1404177111 4 of 7
posteriorprobability
1
0.5163
0.03
Andropogon_gerardii_002
Diheteropogon_hagerupii_183
Themeda_arundinacea_171
Bothriochloa_barbinodis_078
Bothriochloa_ischaemum_063
Bothriochloa_bladhii_093
Bothriochloa_bladhii_076
Hyparrhenia_hirta_189
Bothriochloa_decipiens_086
Dichanthium_annulatum_054
Bothriochloa_insculpta
Bothriochloa_bladhii_082
Hyparrhenia_diplandra_152
Bothriochloa_ischaemum_063
Bothriochloa_hybrida_092
Hyparrhenia_rufa_108
Hyparrhenia_hirta_189
Themeda_villosa_172
Capillipedium_spicigerum_090
Bothriochloa_laguroides_085
Bothriochloa_insculpta
Bothriochloa_woodrovii_083
Schizachyrium_sanguineum_var_hirtiflorum_007
Bothriochloa_bladhii_093
Bothriochloa_exaristata_094
Schizachyrium_sanguineum_032
Schizachyrium_exile_45
Cymbopogon_refractus_012
Capillipedium_parviflorum_005
Dichanthium_annulatum_013
Cymbopogon_parkeri_011
Eulalia_villosa_047Eulalia_quadrinervis_142
Bothriochloa_hybrida_092
Andropogon_perligulatus_003
Dichanthium annulatum_188
Cymbopogon_flexuosus_009
Themeda_triandra_116
Bothriochloa_woodrovii_083
Diheteropogon_amplectens_055
Bothriochloa_exaristata_094
Bothriochloa_insculpta
Bothriochloa_bladhii_082
Bothriochloa_ischaemum_102
Cymbopogon_citratus_135Cymbopogon_parkeri_011
Hyparrhenia_hirta_095
Dichanthium_annulatum_054
Hyparrhenia_rufa_108
Themeda_triandra_116
Hyparrhenia_hirta_095
Cymbopogon_distans_008
Bothriochloa_barbinodis_078
Eulalia_quadrinervis_142
Bothriochloa_hybrida_092
Bothriochloa_macra_079
Themeda_triandra_101
Iseilema_macratherum_061
Bothriochloa_bladhii_087
Schizachyrium_sanguineum_168
Bothriochloa_laguroides_085
Schizachyrium_sanguineum_032
Phacelurus_digitatus
Schizachyrium_sanguineum_var_hirtiflorum_007
Bothriochloa_exaristata_094
Hyparrhenia_hirta_018
Andropogon_glomeratus_192
Hyparrhenia_rufa_153
Hyparrhenia_hirta_018
Dichanthium_aristatum_014
Hyparrhenia_hirta_189
Hyparrhenia_diplandra_152
Bothriochloa_macra_079
Capillipedium_parviflorum_005
Hyparrhenia_hirta_018
Dichanthium_theinlwinii_136
Schizachyrium_brevifolium_167
Andropogon_virginicus _105
Andropogon_chinensis_125
Cymbopogon_obtectus_010
Themeda_triandra_101
Hyparrhenia_hirta_185
Bothriochloa_bladhii_076
Cymbopogon_citratus_135
Andropogon_gayanus_051
Andropogon_eucomus_001
Andropogon_virginicus_075
Hyparrhenia_hirta._185
Spodiopogon_sibiricus
Bothriochloa_alta_077
Schizachyrium_sanguineum_var_hirtiflorum_007
Andropogon_hallii_052
Cymbopogon_obtectus_091
Capillipedium_spicigerum_090
Hyparrhenia_diplandra_177
Andropogon_gerardii_002
Bothriochloa_bladhii_093
Eremopogon_foveolatum_181
Themeda_villosa_172
TK132_Capillipedium_assimile_132
Hyparrhenia_hirta_018
Outgroups
Cymbopogon_citratus_135
Hyparrhenia_rufa_153
Cymbopogon_jwarancusa_ssp_olivieri_049
Schizachyrium_sanguineum_168Schizachyrium_sanguineum_168
Cymbopogon_martinii_048
Bothriochloa_insculpta
Bothriochloa_woodrovii_083
Bothriochloa_decipiens_086
Heteropogon_triteceus_151
Bothriochloa_bladhii_087
1 / 81
0.59 / *
0.60 / 50
0.59 / 29
1 / 96
0.95 / 58
1 / 71
0.78 / 29
1 / 98
1 / 93
0.98 / 44
0.96 / 36
0.83 / 16
1 / 90
0.64 / 50
1 / 32
0.98 / 46
0.83 / 45
1 / 88
0.95 / 69
0.78 / 45
0.98 / 78
0.99 / 54
1 / 960.82 / 29
0.93 / 73
1 / 72
0.99 / 71
0.95 / 98
0.98 / 31
0.99 / 84
1 / 69
0.99 / 100
0.98 / 43
1 / 89
1 / 96
1 / 71
1 / 99
1 / 91
0.73 / 18
1 / 88
0.92 / 42
0.99 / 67
1 / 91
0.52 / 411 / 71
0.57 / 49
1 / 82
0.67 / 12
0.83 / 29
1 / 100
1 / 95
0.58 / 25
1 / 78
1 / 98
1 / 99
1 / 92
0.79 / 58
0.52 / 51
0.99 / 51
1 / 90
0.92 / 43
0.83 / 72
0.52 / 49
0.59 / 44
0.57 / 47
1 / 96
0.92 / 63
0.59 / 8
0.95 / 43
0.59 / *
0.87 / 68
0.99 / 71
1 / 99
0.96 / 34
0.98 / 72
0.99 / 54
13B
C
15
9
10
11
12
A
16
17
18
19
20
21
22
2324
25 26
27
D
E
28
F
ArundinellaArthraxon
Chrysopogon/ThelepogonPhacelurus/Tripsacum/Zea
Apluda/Coix
Sorghum
Saccharum/Ischaemum
GermainiinaeErianthus
542
785
872
1523
1034
1253
1819
1770
779
1105
993
821
1548
2005
765
4125
2273
2053
3880
921
3082
31301427
1236
1510
470
1881
1151
2175
2121
1762
921
1973
2106
1036
14
1 / 1001 / 100
1 / 1001 / 100
1 / 99
1 / 100
1 / 100
1 / 100
1 / 1001 / 100
1 / 100
1 / 100
1 / 100
1 / 100
1 / 100
1 / 100
1 / 1001 / 100
1 / 100
1 / 100
1 / 90
1 / 100
1 / 100
1 / 100
1 / 100
1 / 100
1 / 100
1 / 100
1 / 1001 / 98
1 / 100
1 / 100
1 / 100
1 / 100
1 / 100
1 / 100
1 / 100
1 / 1001 / 100
1 / 1001 / 100
1 / 100
1 / 100
1 / 100
1 / 100
1 / 100
“BCD clade”
“coreAndropogoneae”
A’
Fig. S3. Bayesian phylogeny of core Andropogoneae. Branch numbers, colors, and genome sizes are as in Fig. S2.
Estep et al. www.pnas.org/cgi/content/short/1404177111 5 of 7
0.00 0.310
35
69
Evolutionary Rate
Den
sity
Chrysopogon serrulatusChrysopogon zizanioidesChrysopogon gryllusThelepogon elegans
Arthraxon prionodesArthraxon lanceolatus
Zea mays genome ATripsacum dactyloidesChionachne massieiPhacelurus huillensisPhacelurus digitatusErianthus ravennaeErianthus arundinaceumSaccharum narengaMiscanthus eckloniiPseudosorghum fasciculareMiscanthus sinensisIschaemum muticumIschaemum indicumIschaemum rugosumIschaemum santapauiDimeria sinensisDimeria ornithopodaDimeria fuscescensAndropterum stolziiMicrostegium vimineumKerriochloa siamensisEulalia aureaPolytrias indicaSorghastrum elliottiSorghastrum nutansIschaemum afrumSehima nervosumSpodiopogon sibiricusSchizachyrium sanguineumSchizachyrium sanguineum var. hirtiflorumSchizachyrium exileSchizachyrium brevifoliumAndropogon halliiAndropogon gerardiiAndropogon eucomusAndropogon virginicusAndropogon glomeratusAndropogon perligulatusAndropogon chinensisDiheteropogon hagerupiiDiheteropogon amplectensHyparrhenia diplandraHyparrhenia rufaHyparrhenia hirtaEulalia quadrinervisEulalia villosaCymbopogon distansCymbopogon jwarancusa ssp. olivieriCymbopogon parkeriCymbopopogon martiniiiCymbopogon citratusCymbopogon flexuosusCymbopogon refractusCymbopogon obtectusHeteropogon triticeusThemeda villosaThemeda arundinaceaThemeda triandraDichanthium theinlwiniiDichanthium aristatumDichanthium annulatumCapillipedium parviflorumCapillipedium spicigerumCapillipedium assimileBothriochloa decipiensBothriochloa macraBothriochloa laguroidesBothriochloa insculptaBothriochloa ischaemumBothriochloa bladhiiBothriochloa altaBothriochloa exaristataBothriochloa hybridaBothriochloa woodroviiBothriochloa barbinodisEremopogon foveolatumIseilema macratherumSorghum bicolor genomeGermainia capitataApocopis intermediusApocopis siamensisApocopis cortallumensisPogonatherum paniceumPogonatherum crinitumImperata cylindricaEriochrysis pallidaCoix lacryma-jobiCoix gasteeniiChasmopodium caudatumRottboellia cochinchinensisApluda muticaArundinella nepalensisArundinella hirta
Fig. S4. Phylorate tree from BAMM (16), with species sample frequency adjusted for each clade, and backbone sampling of 0.5. Models of rate shifts with thehighest posterior probability indicate either no shifts or one. The marginal probabilities of a shift are similar across much of the tree; in other words, theposition of the shift cannot be assigned with certainty.
Estep et al. www.pnas.org/cgi/content/short/1404177111 6 of 7
0.0 0.2 0.4 0.6 0.8
0
2
4
6
8
10
12
Net Diversification Rate
Prob
abili
ty d
ensi
ty
DiploidPolyploid
Fig. S5. Net diversification rates (speciation minus extinction) for diploids (blue) and polyploids (yellow), calculated using BiSSE, as implemented in diversitree(19). ML estimates of the rates are significantly different (P < 0.02) compared with a model in which speciation rates are constrained to be equal. Analysisassumes a species sample of 8%.
Other Supporting Information Files
Table S1 (DOCX)
Estep et al. www.pnas.org/cgi/content/short/1404177111 7 of 7
