Phylogeny and Reticulation in Poinae 589 d i v e r s i t y , p h y l o g e n y , a n d e v o l u t i o n i n t h e m o n o c o t y l e d o n s e d i t e d b y s e b e r g , p e t e r s e n , b a r f o d & d a v i s a a r h u s u n i v e r s i t y p r e s s , d e n m a r k , 2 0 1 0 Phylogeny and Reticulation in Subtribe Poinae and Related Subtribes (Poaceae) Based on nrITS, ETS, and trnTLF data Lynn J. Gillespie,1 Robert J. Soreng,2 Michel Paradis,1 and Roger D. Bull1 1Canadian Museum of Nature, Research Division, P.O. Box 3443, Station D. Ottawa, ON, K1P 6P4, Canada 2Smithsonian Institution, Department of Botany MRC-166, National Museum of Natural History, Washington, District of Columbia 20013-7012 U.S.A. Author for correspondence (lgillespie@mus-nature.ca) Abstract?The worldwide temperate to arctic distributed subtribe Poinae com- prises the largest grass genus Poa (500+ species) and 14 smaller genera. Phyloge- netic analyses of combined nuclear ribosomal ITS and ETS and chloroplast trnT- trnL-trnF sequence data for Poinae and the closely related subtribes Alopecurinae, Cinninae, Miliinae, and Puccinelliinae resolve three strongly supported major clades: 1) Poa; 2) all Poinae taxa except Poa, and including Alopecurus, Beckmannia (both Alopecurinae), and Cinna; and 3) subtribe Puccinelliinae. Phleum (Alo- pecurinae) and Milium resolve as two separate lineages, but with low support. Plastid, ITS, and ETS data partitions were not significantly incongruent, with some exceptions in Poa, and combining these datasets resulted in stronger sup- port for many clades. Our results indicate that subtribe Poinae, as currently circumscribed, is paraphyletic, and suggest that subtribe Alopecurinae may be polyphyletic since none of its three sampled genera were resolved as sister taxa. Analyses with all three Alopecurinae genera constrained as monophyletic resulted in trees considerably longer than those from unconstrained analyses. Placements of Arctopoa and Nicoraepoa pugionifolia were incongruent between plastid and nrDNA trees suggesting hybrid origins involving members of the Poa and Poinae (minus Poa) clades. The presence of two distinct ITS sequences in Aniselytron also suggest evolution involving hybridization between these same two major clades. Keywords?Alopecurinae, Aniselytron, Arctopoa, hybridization, phylogeny, re- ticulate evolution running head: Gillespie et al.: phylogeny and reticulation in poi- nae 1120_kap_18_32_6k.indd 589 07/06/10 21.23 590 Gillespie et al. The grass subtribe Poinae (Poaceae, Pooideae, Poeae) currently includes 15 genera and an estimated 550 species distributed worldwide, mostly in cool tem- perate regions (Soreng et al. 2007; Gillespie et al. 2008). Poa is by far the largest genus with about 500 species. The number and circumscription of Poinae genera have changed considerably since the subtribe was established by Dumortier (1829), particularly over the past two decades as a result of new molecular studies combined with morphological data (Soreng et al. 1990; Soreng and Davis 2000; reviewed in Gillespie et al. 2008). For the current circumscription of subtribe Poinae and the arrangement of genera and subtribes in tribe Poeae see Soreng et al. (2003, classification version 27 Oct 2008). In the most recent modification Gillespie et al. (2008) tentatively placed the three genera of subtribe Cinninae (Cinna, Cyathopus, Limnodea) within Poinae, which, if accepted, would bring the total number of Poinae genera to 18. Plastid and nuclear ribosomal DNA (nrDNA) ITS data have inferred nine Poinae genera to be nested within the Poa clade (Hunter et al. 2004; Gillespie and Soreng 2005; Gillespie et al. 2007, 2008; Refulio-Rodriguez 2007). Conse- quently, Anthochloa Nees & Meyen, Austrofestuca (Tzvelev) E. B. Alexeev, Dasypoa Pilg., Eremopoa Roshev., Neuropoa W. D. Clayton, and Parodiochloa C. E. Hubb. have been subsumed within Poa (Gillespie and Soreng 2005; Gillespie et al. 2007, 2008), and combinations within Poa are pending for Aphanelytrum, Dissanthelium, and Tovarochloa (Refulio-Rodriguez 2007). Poa subg. Andinae Nicora was deter- mined not to be part of the Poa clade, and the new Poinae genus Nicoraepoa was described for this group of Patagonian species (Gillespie et al. 2007, 2008; Soreng and Gillespie 2007). With regard to other Poinae genera, Festucella E. B. Alexeev was recently synonymized under Hookerochloa, both endemic to Australia (Gillespie et al. 2008; Jacobs et al. 2008). Ventenata and Gaudiniopsis, sometimes separated as subtribe Ventenatinae (nom. inval., and traditionally allied to Avena L. and Trisetum Pers. [Aveninae]), have been added into the Poinae mix of genera based on recent DNA studies (D?ring et al. 2007; Quintanar et al. 2007; Gillespie et al. 2008). Phylogenetic relationships among Poinae and related subtribes were investi- gated most recently using nrDNA ITS and plastid trnT-trnL-trnF (TLF) sequence data (Gillespie et al. 2008). Separate ITS and TLF analyses provided evidence for Poa as a lineage distinct from all other Poinae genera. Both analyses resolved a clade of subtribes Alopecurinae, Miliinae, Poinae, and Puccinelliinae (PPAM clade, Gillespie et al. 2008), but with very low bootstrap support. Five lineages were resolved in the PPAM clade: Puccinelliinae, Poa, Poinae (minus Poa) (this clade also includes two Alopecurinae genera), Phleum, and Milium (with evi- dence in ITS of a possible broader Miliinae lineage). The plastid trees provided strong support for a sister-group relationship between Puccinelliinae and an Alopecurinae-Miliinae-Poinae clade (PAM, Gillespie et al. 2007), but apart from this, relationships among PPAM lineages were poorly supported in plastid and nrDNA analyses. Resolution and support within PPAM lineages varied consider- 1120_kap_18_32_6k.indd 590 07/06/10 21.23 Phylogeny and Reticulation in Poinae 591 ably between ITS and TLF trees. For example, relationships within the Poinae (minus Poa) and Puccinelliinae clades were more resolved in ITS than TLF, but the reverse was true for the Poa clade. Multiple plastid and ITS analyses suggest that subtribe Alopecurinae (cur- rently defined as including: Alopecurus, Beckmannia, Cornucopiae, Limnas, Phleum, Pseudophleum, and Rhizocephalus) may be polyphyletic (Soreng and Davis 2000; Davis and Soreng 2007; D?ring et al. 2007; Gillespie et al. 2007; Quintanar et al. 2007; Soreng et al. 2007; Gillespie et al. 2008). Three of seven genera have been examined so far: Alopecurus (ca. 37 spp., widespread), Beckmannia (2 spp., boreal), and Phleum (ca. 13 spp., widespread). The other four genera have one or two species each, and are of more limited distribution (Mediterranean, Turkey, NE Asia [Limnas]). In the most recent analysis Alopecurus and Beckmannia are part of the Poinae (minus Poa) clade, but do not resolve as sister taxa, and Phleum resolves as a separate lineage (Gillespie et al. 2008). These molecular results ap- pear to contradict the close morphological similarity of Alopecurinae members. These seven genera share a unique combination of morphological characters in the PPAM clade, including dense spicate panicles and one-flowered spikelets that disarticulate below the glumes (Soreng and Davis 2000; Soreng et al. 2007; Gillespie et al. 2008). Conflicts between plastid and ITS trees suggest that three Poinae taxa may have originated from hybridization between Poa and other Poinae (including Cinninae) genera (Gillespie et al. 2008). All collections (eight) and species (three) examined of Arctopoa have ITS DNA placing it in the Poinae (minus Poa) clade, while plastid data places the genus in the Poa subg. Sylvestres clade, with which it bears little morphological resemblance (Gillespie et al. 2008). Though gener- ally treated as a subgenus within Poa (subg. Arctopoa (Griseb.) Prob.), Arctopoa was resurrected based on these molecular data and morphology (Gillespie et al. 2008). Similarly, but with reversed placements, the south-east Asian genus Aniselytron was positioned in the P. subg. Sylvestres (V.L. Marsh ex Soreng) Soreng & L.J. Gillespie clade in ITS analyses, but in the Poinae (minus Poa) clade in plastid analyses. However, this finding needs to be verified since it was based on different samples for ITS and TLF. Nicoraepoa pugionifolia also resolved with conflicting placements. Plastid and morphological data placed the species in the Poinae (minus Poa) clade consistent with all other species examined of the newly described Poinae genus Nicoraepoa (Soreng and Gillespie 2007; Gillespie et al. 2008), whereas ITS analyses placed it within Poa in the primarily sub-Antarctic island section Parodiochloa (C. E. Hubb.) Soreng & L. J. Gillespie clade. Since the number of parsimony informative characters was limited and character consistency was lower in ITS (compared to TLF, which had 10% more characters and a substantially higher overall consistency index, 0.633 versus 0.322, Gillespie et al. 2008), we decided to increase the size of the nuclear ribosomal DNA (nrDNA) dataset by sequencing part of the nrDNA external transcribed spacer (ETS) (Baldwin and Markos 1998). ETS has been successfully combined 1120_kap_18_32_6k.indd 591 07/06/10 21.23 592 Gillespie et al. with ITS to enhance resolution and support of nrDNA phylogenies (e.g., Bald- win and Markos 1998; Markos and Baldwin 2001; Li 2002; Urbatsch et al. 2003; Acevedo-Rosaset al. 2004; Razafimandimbisona et al. 2005). In this paper we use TLF, ITS, and ETS sequence data to infer phylogenetic relationships in the PPAM clade with a focus on genera of subtribes Poinae and Alopecurinae. Our goals are to: 1) examine congruence among the three datasets; 2) generate more robust phylogenetic hypotheses based on combined nrDNA ITS and ETS data, and on combined plastid TLF and nrDNA ITS/ETS data; 3) ex- plore relationships among Alopecurinae genera and investigate the monophyly of the subtribe; and 4) further investigate the origin and evolution of three pu- tative intergeneric hybrid taxa, Aniselytron, Arctopoa, and Nicoraepoa pugionifolia. Materials and Methods Taxon Sampling?Seventy-two accessions representing 38 Poinae species, 14 other PPAM species, and three outgroup species were included in this study (Ap- pendix 1). New ETS sequence data were collected for 64 accessions including 37 Poinae species and 10 other PPAM species. TLF and ITS sequences were previous- ly published for many of these accessions (Gillespie et al. 2007, 2008); new TLF and ITS sequences were generated for 12 and 16 collections, respectively. Com- plete TLF sequences were obtained for Aniselytron where we previously had only partial (trnL intron and trnLF) sequences. For each operational taxonomic unit (OTU) in the combined data matrix TLF, ITS, and ETS sequences were from the same collection, except for Cinna latifolia and Milium effusum (Appendix 1). In this paper we included only those species/accessions for which we had 1) both ITS and ETS sequences (50 species, 64 accessions), and 2) sequences of all three DNA regions TLF, ITS, and ETS (49 species, 61 accessions) (Appendix 1). Among PPAM clade members this included eight Poinae and three Alopecurinae genera, plus Cinna (Cinninae), Milium (Miliinae), and Catabrosa and Puccinellia (both Puccinelliinae). We included a reduced set of Poa species similar to Analy- sis I of Soreng et al. (2010): diploids (2n = 14) plus P. subg. Sylvestres (species sampled are tetraploid), with one accession of each species. Two additional species of Poa, P. flabellata (tetraploid; also included in Analysis II of Soreng et al. 2010) and P. kerguelensis (chromosome number unknown), were added in an attempt to narrow down the affinities of one of the putative hybrid taxa. Three outgroup genera were included as representatives of the other two major Poeae s.l. clades: Deschampsia P. Beauv. (subtribe Airinae), Festuca L.(Loliinae), and Helictotrichon Besser ex Schult. & Schult. f. (Aveninae) (Quintanar et al. 2007, Gillespie et al. 2008). This selection was in part based on outgroup taxa for which we were able to generate ETS sequence data. DNA Sequences?Methods of DNA extraction, amplification, and direct sequencing of TLF and ITS were outlined in Gillespie et al. (2008). A region of approximately 500 base pairs (bp) at the 3? end of the nrDNA external tran- 1120_kap_18_32_6k.indd 592 07/06/10 21.23 Phylogeny and Reticulation in Poinae 593 scribed spacer (ETS) was amplified and sequenced using the same reaction conditions as ITS and primers 18S-R (Starr et al. 2003) and RETS4-F, a primer newly designed for this study (5?-TGGCTACGCGAGCGCATGAG-3?). Although designed specifically for use in Poa, the RETS4-F primer was also found to suc- cessfully amplify other Poinae genera and more widely in Poeae. Sequence assembly, alignment, and editing were performed using Sequencher vers. 4.7 (Gene Codes Corp., Ann Arbor, Michigan), ClustalX vers. 1.83 (Jean- mougin et al. 1998), and BioEdit vers. 5.0.9 (Hall 1999), respectively, as outlined in Gillespie et al. (2008). Obvious and unambiguous nucleotide variants within a sequence (i.e., double peaks on electropherogram trace of approximately equal strength, or at least unambiguous double peaks in an otherwise clean sequence) were coded using standard IUB ambiguity codes. Within sample insertion- deletion (indel) variants were inferred via direct sequencing. Minor indel variants that were not parsimony informative were coded for the dominant variant, or where variants were of approximate equal intensity the longest. Where possible, more substantial, parsimony informative indel variation within a sample was coded as separate variant sequences. Analyses?TLF, ITS, and ETS alignments were merged into a single nexus format data matrix. Indel characters were not included in this matrix. To de- termine if it was appropriate to analyze combined plastid and nuclear ribos- omal data partitions, we used the Incongruence Length Difference (ILD) test of Farris et al. (1995) as implemented with the Partition-Homogeneity Test in PAUP* 4.0b10 (Swofford 2002) with 100??1000 replications (depending on run times). We determined the extent of conflict between 1) the two nrDNA data partitions, ITS and ETS, and 2) the plastid and nrDNA partitions, TLF, ITS, and ETS. Collections resulting in substantial incompatibility between the TLF and ITS/ETS partitions were subsequently each treated as two separate OTUs: plastid and nrDNA. Following exploratory parsimony analyses of TLF, ITS, ETS, ITS/ETS and TLF/ITS/ETS data partitions, three final parsimony analyses were run (Table 1). Analysis I included ITS and ETS data partitions and 65 OTUs. This included all OTUs for which we had ITS and ETS data that, Poa aside, were not significantly incompatible. Included here, but not in the following TLF/ITS/ETS analyses (II and III), were Bellardiochloa polychroa and one additional collection each of Arctagrostis latifolia (A14) and Nicoraepoa robusta (S7359) for which we had ITS and ETS data but not TLF data (Appendix 1). The putative hybrid taxa Arctopoa and Nicoraepoa pugionifolia were included, as were three collections of Aniselytron (four OTUs, see Results). We excluded Poa infirma and P. supina, two sister species on long branches whose positions conflicted considerably between ITS and ETS in exploratory analyses. Analysis II included TLF, ITS, and ETS data partitions and 49 OTUs. The three putative hybrid taxa (Arctopoa, Aniselytron, and Nico- raepoa pugionifolia) that in previous analyses had resolved in highly incongruent positions between TLF and ITS trees (Gillespie et al. 2008) were not included 1120_kap_18_32_6k.indd 593 07/06/10 21.23 594 Gillespie et al. A n al ys is T LF IT S E T S IT S/ E T S IT S/ E T S* IT S/ E T S* + h yb ri d I? ? Fi g. ?1 T LF /I T S/ E T S II ?? F ig .? 2 T LF /I T S/ E T S + h yb ri d II I? ? Fi g. ?3 O T U s 49 49 49 49 50 65 49 74 N o . o f ch ar ac te rs 19 72 62 9 51 7 114 6 114 6 114 6 31 18 31 18 N o . o f p ar si m o n y in fo rm at iv e ch ar . 16 1 14 7 18 4 33 1 29 8 30 3 49 2 51 1 % p ar si m o n y in fo rm at iv e ch ar . 8. 2% 23 .4 % 35 .6 % 28 .9 % 26 .0 % 26 .4 % 15 .8 % 16 .4 % T re e Le n gt h 47 7 54 5 69 4 12 58 10 99 113 8 17 61 18 21 N u m b er o f tr ee s 45 3 67 3 10 5 36 35 44 83 12 83 0 0 C I 0 .7 0 5 0 .4 75 0 .5 30 0 .4 96 0 .4 95 0 .4 89 0 .5 30 0 .5 24 R I 0 .8 48 0 .7 42 0 .7 45 0 .7 34 0 .7 63 0 .7 92 0 .7 48 0 .7 79 T ab le 1. S u m m ar y st at is ti cs fo r m ax im u m p ar si m o n y an al ys es o f s ep ar at e T LF , I T S, a n d E T S d at as et s, a n d c o m b in ed d at as et s (A n al ys es I? ?I II ). C o n si st en cy in d ex d o es n o t in cl u d e p ar si m o n y u n in fo rm at iv e ch ar ac te rs T h e IT S/ E T S* t ax o n s et w as t h e sa m e as t h at o f IT S/ E T S* + h yb ri d ( A n al ys is I ) b u t ex cl u d in g th e th re e p u ta ti ve h yb ri d t ax a. 1120_kap_18_32_6k.indd 594 07/06/10 21.23 Phylogeny and Reticulation in Poinae 595 in this analysis. Analysis III was the same as Analysis II, with the addition of the three putative hybrid taxa. TLF and ITS/ETS sequences of Arctopoa and N. pugionifolia were not combined; each accession was treated as two OTUs, one containing only the TLF sequence, the other the combined ITS/ETS sequence. Aniselytron OTUs are explained under Results. The total number of OTUs was 74, including three species of Arctopoa (eight collections, 17 OTUs), one species of Aniselytron (three collections, four OTUs), and N. pugionifolia (two collections, four OTUs). Maximum parsimony (MP) and bootstrap analyses were performed as in Gillespie et al. (2008). Complete heuristic searches were run in PAUP* 4.0b10 (Swofford 2002) for Analyses I and II, with MULPARS and TBR branch swap- ping. Two heuristic search strategies were used for Analysis III: 1) maximum tree setting of 90,000, no replication; and 2) 100 random addition replicates, 900 trees saved per replicate. Monophyly constraint analyses were performed in PAUP* using the constraints command and the monophyly option. Bootstrap analyses were performed with 1000 replicates, TBR swapping, and the MUL- TREES setting on for Analyses I and II, and with 1000 replicates, 10 addition sequences per replicate, TBR swapping, and the MULTREES setting turned off for Analysis III due to long search times (DeBry and Olmstead 2000). Follow- ing the suggestion of Starr et al. (2004), bootstrap support (BS) for clades was characterized as very poor (<55%), poor (55%?64%), moderate (65%?74%), good (75%?84%), very good (85%?94%), or strong (95%?100%). Consistency index (CI) is calculated on the basis of parsimony informative chaacters only. Clade sup- port refers to bootstrap support, unless otherwise indicated. We conducted Bayesian analyses using MrBayes v. 3.1 (Huelsenbeck and Ronquist 2001, Ronquist and Huelsenbeck 2003). To choose the DNA substitu- tion model we used the Akaike Information Criterion (AIC) as implemented in ModelTest ver. 3.7 (Posada and Crandall 1998). For datasets of all three final analyses the optimal model of evolution was GTR + ? + I. For each analysis, two parallel sets of four chains (three heated and one cold) were run simultaneously, with each Markov chains starting from a random tree. Runs were stopped after the average standard deviation of split frequencies (a convergence diagnostic) was less than 0.01; the number of generations run for each data partition was 1 million. Trees were sampled every 100 generations and the first 25% of the trees were discarded as burn-in. For each analysis trees from each run were pooled and a majority-rule consensus tree was computed in PAUP* to estimate posterior probabilities (pp.) of individual clades. Clades were considered well defined if their pp.?values were 95% or above. Results Sequence Characteristics and ILD Tests?Characteristics of the TLF, ITS, and ETS data matrices considered separately and combined are provided in Table?2. 1120_kap_18_32_6k.indd 595 07/06/10 21.23 596 Gillespie et al. For details on TLF and ITS sequences and alignments refer to Gillespie et al. (2008). The aligned ETS data matrix comprised a total of 642 characters. Eleven insertions (either unique or those shared by few OTUs and lacking parsimony informative nucleotide characters), one to 64 bases in length, were deleted from the data matrix prior to analysis, as was a 12 bp region that could not be unambiguously aligned, for a total of 517 analyzed base positions. Length of the sequenced ETS region for individual accessions ranged from ca. 500 bp (numerous species) to 567 bp (P. infirma). Poa infirma and P. supina were both characterized by numerous, mostly shared indels (the largest, a 64 bp insertion, was unique to P. infirma). In addition, these two species shared one very large insertion (222 bases in length in P. infirma, incomplete at 194 bases in P. supina) at the 3? end of ETS, 15 bases downstream of the 18S-R primer, which was located just outside the segment of ETS included in the final data matrix. Intra-sample variation was detected in some TLF, ITS, and ETS sequences, suggesting the presence of two (or possibly more) sequence variants within a sample (in the nrDNA data these variants may occur on different arrays, and are thus not polymorphisms in the strict sense). Single nucleotide differ- ences represented most of this variation, while single base indel variants were less frequent. ITS variants were discussed previously in Gillespie et al. (2008); one collection was coded differently here. Arctopoa tibetica (OL03-7) had both indel and base variants and was previously coded only as polymorphic for the base variants; the sample was reinterpreted here as including two variant ITS sequences, 1 and 2. One sample of Aniselytron treutleri (S5229) newly seqeunced here also contained two variant ITS sequences (discussed further below). A second collection of Nicoraepoa pugionifolia (P17128) had the same two ITS indel variants found in the sample previously analyzed (Gillespie et al. 2008: S7336). Indel variants in the TLF region were detected in only two species sampled here: Taxa constrained as monophyletic ITS ITS/ETS + hybrid TLF/ITS/ETS TLF/ITS/ETS + hybrids Analysis I II III OTUs 54 65 49 74 None Length 586 1138 1761 1821 Alopecurus + Beckmannia Length increase 1 3 4 4 Alopecurus + Beckmannia + Phleum Length increase 6 10 15 15 Table 2. Monophyly constraint analyses for subtribe Alopecurinae. For each of the four datasets number of operational taxonomic units (OTUs) and tree length (L) are given for the unconstrained analysis, analysis constraining Alopecurus and Beckmannia as monophyletic, and the analysis con- straining Alopecurus, Beckmannia, and Phleum as monophyletic. For the constrained analyses number of steps longer than the unconstrained analysis are given. 1120_kap_18_32_6k.indd 596 07/06/10 21.23 Phylogeny and Reticulation in Poinae 597 Aniselytron treutleri and Nicoraepoa pugionifolia. Both species (two collections each) had the same poly-A indel variant in the trnT-L section of the sequence. ETS sequences included only minor variants, and there was no evidence of highly divergent sequences within a sample in our data generated using the direct sequencing approach. Single 1 bp indel variants were present in ETS sequences of Aniselytron treutleri (both collections), Nicoraepoa pugionifolia (both collections), N. robusta, and Poa ligulata. The sample of Catabrosa werdermannii contained two indel and numerous nucleotide variants resulting in a messy but interpretable sequence; only the dominant sequence variant was coded. Based on ILD tests ITS and ETS data partitions were determined to be in- compatible for all 65 OTUs (49 taxa) included in Analysis I (p=0.1??0.2). Likewise, all combinations of the three data partitions were found to be incompatible for all 49 OTUs (45 taxa) included in Analysis II (p=0.1??0.2). Exploratory analyses and ILD tests showed that much of this incompatibility can be attributed to the Poa clade. Removing various combinations of Poa species (other than P. subg. Sylvestres species) resulted in data partitions that were not significantly incompatible for all partition combinations (p = 0.15??0.79). Removing seven of the 20 Poa species (P. diaphora, P. flabellata, P. infirma, P. kerguelensis, P. let- termanii, P. pseudoabbreviata, and P. supina) resulted in the highest p value for the ITS/ETS and TLF/ITS partitions (0.79 for each). Retaining only P. subg.. Sylvestres species (P. autumnalis, P. saltuensis, P. sylvestris, and P. wolfii) resulted in the highest p value for the TLF/ETS (0.51) and TLF/ITS/ETS partitions (0.15). Excluding the three outgroup taxa, in addition to all Poa except for P. subg. Sylvestres, gave an even higher p value (0.25) for the TLF/ITS/ETS partition, possibly due to exclusion of Deschampsia, which has previously been shown to be unstable in position between plastid and nuclear analyses (Quintanar et al. 2007; Soreng et al. 2007; Gillespie et al. 2008). There appeared to be no significant incompatibility between any combina- tion of the TLF, ITS, and ETS partitions for taxa in the Poinae (minus Poa) clade, or more broadly for taxa in the PPAM clade with the exception of Poa based on the above ILD tests. Given that we are primarily interested here in relationships among Poinae and Alopecurinae genera, and not in relationships within Poa, we feel that combining datasets is justified for the purposes of this paper. The significant incompatibilities between the two nrDNA datasets, ITS and ETS, when all taxa in our dataset were included was surprising, and is an issue that will be addressed in a future paper focusing on relationships within Poa. Previous analyses showed that the positions of Arctopoa, Aniselytron, and Nico- raepoa pugionifolia were highly incongruent between TLF and ITS trees (Gillespie et al. 2008). ILD tests performed on TLF/ITS/ETS, TLF/ITS, and TLF/ETS data partitions for the taxon set excluding Poa members except for P. subg. Sylvestres plus either Arctopoa or Nicoraepoa pugionifolia indicated that these partitions were significantly incompatible when either taxon was included (p=0.01). In 1120_kap_18_32_6k.indd 597 07/06/10 21.23 598 Gillespie et al. contrast, ITS/ETS data partitions for the same set of taxa were not significantly incompatible (p=0.20, 0.24, 0.12 for the same taxon set including Arctopoa, N. pugionifolia, and both taxa, respectively). Therefore ITS and ETS sequences were combined, but TLF and ITS/ETS sequences were not combined for Arctopoa and N. pugionifolia. Each accession of these two taxa was treated as two OTUs, one containing only the TLF sequence, the other the ITS/ETS sequence. The situation with Aniselytron was more complex than previously thought (Gillespie et al. 2008). Preliminary analyses suggested that TLF and ETS sequenc- es may be compatible, and the ILD test confirmed this (p=0.34 for the Analysis II taxon set including Aniselytron and excluding all Poa except for P. subg. Sylvestres). The one additional collection sequenced for ITS (S5229) comprised two divergent sequences, which differed in three indel variants and numerous base variants. Although indel variants result in garbled electropherogram traces downstream of the variant, by careful reading of multiple sequences obtained from end and internal primers, we were able to decipher two consensus sequences. The first variant sequence was identical to that previously obtained for S5264 (Gillespie et al. 2008), whereas the second appeared most similar to sequences of Poinae gen- era such as Nicoraepoa and Hookerochloa. ILD tests confirmed that the TLF/ITS/ ETS data partitions were not significantly incompatible when the combined S5229 sequence with ITS variant 2 was included (p=0.07 [borderline], 0.19 with taxon set excluding outgroups). In contrast, data partitions for the taxon set including the combined S5229 sequence with ITS variant 1 were significantly incompatible (p=0.01). Thus, for accession S5229 ITS variant 1 was analyzed as a separate OTU, while variant 2 was combined with TLF and ETS data. Analyses?Summary tree statistics for MP analyses of separate TLF, ITS, and ETS datasets and of combined ITS/ETS and TLF/ITS/ETS datasets are provided in Table 1. ETS, the shortest DNA region, had the highest number and percent- age of parsimony informative (PI) characters. Plastid TLF, the longest region, had by far the lowest percentage, but an intermediate number of PI characters, whereas ITS had the lowest number of PI characters. MP trees resulting from the separate TLF analysis had the highest consistency index (CI). ETS MP trees had a higher CI than ITS trees, a somewhat surprising result given the greater number of variable and PI characters in the ETS dataset. Consistency indices of trees resulting from analyses of combined datasets were all greater than those from the ITS analysis, and not substantially lower than the average of the CIs from the separate analyses. analysis I?Maximum parsimony analysis of the combined ITS/ETS data matrix with putative hybrid taxa resulted in 4483 trees, 1128 steps long, with a CI of 0.490 (Table 1). A phylogram depicting one of the most parsimonious trees, with branches found in the strict consensus tree shown by bold lines, is illustrated in Fig.?1. The Bayesian majority rule consensus tree was for the most part identical in structure to the MP tree shown in Fig.?1. The major differences were Milium and 1120_kap_18_32_6k.indd 598 07/06/10 21.23 Phylogeny and Reticulation in Poinae 599 Nicoraepoa chonotica Nicoraepoa robusta S7358 Nicoraepoa robusta S7359 Nicoraepoa andina Nicoraepoa subenervis subenervis Nicoraepoa subenervis spegazziniana P19186 Nicoraepoa subenervis spegazziniana S7155 Arctagrostis latifolia A157a Arctagrostis latifolia A145a Arctagrostis latifolia G6586 Hookerochloa eriopoda J9128 Hookerochloa eriopoda J9129 Hookerochloa hookeriana Aniselytron treutleri Ma140 ETS Aniselytron treutleri S5229 ITS1/ETS Arctopoa subfastigiata W61819 Arctopoa subfastigiata W61824 Arctopoa eminens S6022 Arctopoa eminens G7010-2 Arctopoa eminens G7010-3 Arctopoa tibetica OL03-7 ITS2/ETS Arctopoa tibetica S5481 Arctopoa tibetica S7626 Arctopoa tibetica OL03-7 ITS1/ETS Cinna latifolia Cinna arundinacea Dupontia fisheri G6589 Dupontia fisheri G6699 Arctophila fulva Alopecurus borealis Alopecurus gerardii Bellardiochloa variegata Belardiochloa polychroa Apera interrupta Beckmannia syzigachne A Poa ligulata Poa molinerii Poa badensis Poa pseudoabbreviata Poa lettermanii Nicoraepoa pugionifolia S7336 Nicoraepoa pugionifolia P17128 Poa flabellata Poa kerguelensis Poa dolosa Poa trichophylla Poa media Poa sibirica Poa chaixii Poa remota Poa diaphora Aniselytron treutleri S5264 ITS Aniselytron treutleri S5229 ITS2 Poa saltuensis Poa wolfii Poa autumnalis Poa sylvestris Catabrosa werdermannii Puccinellia vahliana Milium vernale Milium effusum Phleum pratense 5 changes 93 98 87 52 87 63 77 96 99 55 98 99 99 99 83 79 94 73 100 57* 84 63 100 100 100 100 100 100 100 54 83 85 80 90 83 95 72 80 56 85 99 65 70 86 94 52* Poa clade Poinae (minus Poa) clade Puccinelliinae clade A A A C C 98 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 88 100 100 100 100100 100 83 68 72 90 95 95 98 98 63 97 92 95 98 53 56 100 subg. Sylvestres clade sect. Parodiochloa clade - - - 100 1 2 Fig.?1. Phylogram with strict consensus tree shown in bold lines resulting from maximum parsimony (MP) analysis of combined ITS and ETS data for collections of subtribes Poinae, Alopecurinae, Cin- ninae, and Puccinelliinae (4483 trees, 1138 steps, CI = 0.495). MP bootstrap values >50% are given above branches, Bayesian posterior probability values below branches. Collection information is given only for species represented by multiple collections; ?ITS? and/or ?ETS? are provided only for OTUs having variant sequences or where only one of the two regions was sequenced; variant ITS sequences are indicated as ?ITS1? or ?ITS2?; ?A? indicates species belonging to subtribe Alopecurinae, ?C? those belonging to Cinninae. Outgroup taxa are not shown. 1120_kap_18_32_6k.indd 599 07/06/10 21.23 600 Gillespie et al. Phleum forming a clade with Poa with high posterior probability (pp.?=100, not shown on Fig.?1), and Beckmannia as sister to the Alopecurus-Arctophila-Dupontia clade (pp.?=68). Three major PPAM clades are inferred with good to strong support and high posterior probability: 1) Poa (including Nicoraepoa pugionifolia and some Anise- lytron ITS OTUs) (BS=90%, pp.?=100); 2) Poinae (excluding Poa) with Alopecurus, Beckmannia, and Cinna (referred to here as the Poinae (minus Poa) clade) (BS=79, pp.?=100); and 3) Puccinelliinae (BS=99, pp.?=100). Relationships among these three clades and the two separate Milium and Phleum lineages are not supported (BS<50). Support for the monophyly of genera ranges from poor (Cinna; BS=57), moderate (Bellardiochloa, Dupontia; BS=65??72), very good (Poa; BS=90), to strong (Alopecurus, Arctagrostis, Arctopoa, Hookerochloa, Nicoraepoa [excluding N. pugioni- folia]; BS=96??100). Within the Poinae (minus Poa) clade the Apera-Bellardiochloa subclade is strongly supported (BS=100), while the Arctophila-Alopecurus-Dupontia subclade (BS=85) and the large subclade comprising Arctagrostis, Arctopoa, Anise- lytron, Cinna, Hookerochloa, and Nicoraepoa (subclade 1: BS=83) are well supported. Relationships among these clades plus Beckmannia receive no support (BS<50). Within the Poa clade subgenus Sylvestres resolves with very good support as a clade (BS=85) and with good support as sister to a clade comprising all other Poa species included in the analysis (BS=80). Aniselytron resolved in two different positions, with P. subg. Sylvestres in the Poa clade (two ITS sequences) and in subclade 1 of the Poinae (minus Poa) clade (one ETS and one combined ETS/ ITS sequence). analysis II?Parsimony analysis of the combined TLF/ITS/ETS data matrix resulted in 12 trees, 1761 steps long, with a CI of 0.530 (Table 1; Fig.?2). The Bayesian majority rule consensus tree was identical in branching structure to the MP phylogram shown in Fig.?2, and thus slightly more resolved than the MP strict consensus tree (shown by bold lines in Fig.?2). However, the two clades that resolved in the Bayesian tree, but not in the MP strict consensus tree (monophyly and position of Cinna as shown in Fig.?2), were present in the bootstrap tree with moderate to good support. Several large clades had high posterior probablity in the Bayesian tree, but low support in the MP tree, e.g., Alopecurus-Apera-Arctophila-Bellardiochloa-Beckmannia-Dupontia (clade 2, pp.?=100 versus BS=58) and Milium-Phleum-Poa (pp.?=100, BS=65). The TLF/ITS/ETS analysis strongly supports the same three major PPAM clades as in Analysis I: 1) Poa (BS=99, pp.?=100); 2) Poinae (excluding Poa) plus Alo- pecurus, Beckmannia, and Cinna (BS=96, pp.?=100); and 3) Puccinelliinae (BS=100, pp.?=100). As in Analysis 1 Milium and Phleum form two additional separate line- ages; here they resolve together in a clade with Poa (BS=65, pp.?=100). There is strong support (BS=98??100) for the monophyly of all genera (where more than one species was included), with the exception of Dupontia (BS=89) and Cinna (BS=60). The Puccinelliinae clade resolves as sister to all other PPAM members with good support (BS=83). 1120_kap_18_32_6k.indd 600 07/06/10 21.23 Phylogeny and Reticulation in Poinae 601 Poa dolosa Poa media Poa trichophylla Poa remota Poa sibirica Poa chaixii Poa diaphora Poa pseudoabbreviata Poa lettermanii Poa badensis Poa molinerii Poa ligulata Poa supina Poa infirma Poa flabellata Poa kerguelensis Poa autumnalis Poa sylvestris Poa saltuensis Poa wolfii Milium vernale Milium effusum Phleum pratense Nicoraepoa chonotica Nicoraepoa robusta Nicoraepoa andina Nicoraepoa subenervis spegazziniana P19186 Nicoraepoa subenervis spegazziniana S7155 Nicoraepoa subenervis subenervis Arctagrostis latifolia G6586 Arctagrostis latifolia A157a Hookerochloa eriopoda J9128 Hookerochloa eriopoda J9129 Hookerochloa hookeriana Dupontia fisheri G6589 Dupontia fisheri G6699 Arctophila fulva Alopecurus borealis Alopecurus gerardii Bellardiochloa variegata Apera interrupta Beckmannia syzigachne Cinna arundinacea Cinna latifolia Catabrosa werdermannii Puccinellia vahliana 10 changes 100 100 100 100 100 100 100 100 100 100 81 99 99 96 99 95 94 96 96 91 57 68 56 59 65 83 84 100 100 75 58 99 58 74 60 62 100 100 100 100 100 100 100 100 100100 100 100 100 100 100 100 100 99 74 100 100 100 100 100 95 53 100 100 100 100 100 53 80 88 78 100 81 100 100 100 100 89 Poa clade Poinae (minus Poa) clade Puccinelliinae clade A A C C A 100 P. subg. Sylvestres P. subg. Poa P. subg. Stenopoa P. subg. Pseudopoa P. subg. Ochlopoa 99 100 99 100 99 100 A * * * * * * * * * * * 1 2 Fig.?2. Phylogram with strict consensus tree shown in bold lines resulting from MP analysis of com- bined TLF, ITS and ETS data for subtribe Poinae, excluding putative hybrid taxa, plus representative members of subtribes Alopecurinae, Cinninae, and Puccinelliinae (12 trees, 1761 steps, CI = 0.530). MP bootstrap values >50% are given above branches, Bayesian posterior probability values below branches. Collection information is given only for species represented by multiple collections; ?A? indi- cates species belonging to subtribe Alopecurinae, ?C? those belonging to Cinninae; ?*? indicates taxa with one-flowered spikelets, all others have multi-flowered spikelets. Outgroup taxa are not shown. 1120_kap_18_32_6k.indd 601 07/06/10 21.23 602 Gillespie et al. Within the Poa clade, subgenus Sylvestres resolves with strong support as a clade (BS=94) and as sister to a clade comprising all other Poa species included in the analysis (BS=96). Species in this latter clade form two well supported clades corresponding for the most part to subgenus Ochlopoa (Asch. & Graebn.) Hyl. (BS=75), and subgenera Poa (P. chaixii, P. remota, P. siberica), Pseudopoa (K. Koch.) Stapf (P. diaphora), and Stenopoa (Dumort.) Soreng & L.J. Gillespie (P. badensis, P. ligulata, P. molinerii) (BS=81); the exception is P. media (Ochlopoa). This species, P dolosa, and P. trichophylla are discussed in Soreng et al. (2010). Strongly supported clades in the Poinae (minus Poa) clade include: Arctagrostis and Nicoraepoa (BS=100); Arctagrostis, Hookerochloa, and Nicoraepoa (BS=99); Apera and Bellardiochloa (BS=100); and Arctophila and Dupontia (BS=100). Alopecurus is sister to Arctophila-Dupontia (BS=74), and Cinna is sister to Arctagrostis-Hookero- chloa-Nicoraepoa (BS=81), with moderate to good support, respectively. analysis III?Parsimony analysis of the combined TLF/ITS/ETS data matrix including putative hybrid taxa resulted in 8300 trees, 1821 steps long, with a CI of 0.524 (Table 1, Fig.?3). The Bayesian majority rule consensus tree was very similar to the MP tree shown in Fig.?3, differing mostly in being somewhat more resolved. The additional clades in the Bayesian tree had mostly low posterior probablity, and the only one of interest here was an Aniselytron (ITS OTUs)- Arctopoa (TLF OTUs) clade (pp.?=72). As in analysis II several large clades had high posterior probability, but low bootstrap support (pp.?=99??100 versus BS<65), while one clade strongly supported in the MP tree was not detected in the Bayesian tree (e.g., Arctopoa TLF OTUs, BS=95). Tree structure is congruent with that in Analysis II, but with less resolution among Arctagrostis, Hookerochloa, and Nicoraepoa members. Although not present in the MP strict consensus tree, bootstrap analysis provided moderate to good support for the following clades: Arctagrostis and Nicoraepoa (excluding N. pugioni- folia) (BS=81); and this clade plus Hookerochloa and N. pugionifolia (BS=74). The putative hybrid taxa each resolved as two clades, plastid and nrDNA (except see below for partition of Aniselytron OTUs), with very good (N. pugionifolia, BS=85) to strong support (Arctopoa, Aniselytron, BS=95??96). Arctopoa TLF OTUs resolved together (BS=95) in the Poa clade (BS=93) with subg. Sylvestres (BS=62), whereas ITS/ETS OTUs resolved in the Poinae clade (minus Poa) (BS=89) within sub- clade 1 (BS=58). Nicoraepoa pugionifolia TLF OTUs resolved in the same Poinae (minus Poa) subclade, but ITS/ETS OTUs resolved in the Poa subg. Ochlopoa sect. Parodiochloa clade (BS=100). Two ITS OTUs of Aniselytron resolved in the Poa subg. Sylvestres clade (S5229 ITS variant 1 and S5264). The single complete TLF/ ITS/ETS sequence (S5229 including ITS variant 2) and the TLF/ETS sequence (Ma 140; ITS not successfully sequenced) of Aniselytron formed a clade (BS=96) in subclade 1 in the Poinae (minus Poa) clade. alopecurinae monophyly analyses?Constraint analyses testing for the monophyly of subtribe Alopecurinae are summarized in Table 2. None of the three Alopecurinae genera examined here, Alopecurus, Beckmannia, and Phleum, 1120_kap_18_32_6k.indd 602 07/06/10 21.23 Phylogeny and Reticulation in Poinae 603 Poa badensis Poa molinerii Poa ligulata Poa supina Poa infirma Nicoraepoa pugionifolia S7336 ITS/ETS Nicoraepoa pugionifolia P17128 ITS/ETS Poa kerguelensis Poa flabellata Poa dolosa Poa media Poa trichophylla Poa remota Poa sibirica Poa chaixii Poa diaphora Poa pseudoabbreviata Poa lettermanii Arctopoa eminens 7010-3 TLF Arctopoa eminens S6022 TLF Arctopoa eminens 7010-2 TLF Arctopoa subfastigiata W61819 TLF Arctopoa subfastigiata W61824 TLF Arctopoa tibetica S5481 TLF Arctopoa tibetica OL03-7 TLF Arctopoa tibetica S7626 TLF Poa autumnalis Poa sylvestris Poa saltuensis Poa wolfii Aniselytron treutleri S5264 ITS Aniselytron treutleri S5229 ITS1 Milium vernale Milium effusum Phleum pratense Nicoraepoa chonotica Nicoraepoa robusta Nicoraepoa andina Nicoraepoa subenervis spegazziniana P19186 Nicoraepoa subenervis spegazziniana S7155 Nicoraepoa subenervis subenervis Arctagrostis latifolia G6586 Arctagrostis latifolia A157a Hookerochloa eriopoda J9128 Hookerochloa eriopoda J9129 Hookerochloa hookeriana Nicoraepoa pugionifolia S7336 TLF Nicoraepoa pugionifolia P17128 TLF Aniselytron treutleri Ma140 TLF/ETS Aniselytron treutleri S5229 TLF/ITS2/ETS Arctopoa subfastigiata W61819 ITS/ETS Arctopoa subfastigiata W61824 ITS/ETS Arctopoa eminens S6022 ITS/ETS Arctopoa eminens 7010-2 ITS/ETS Arctopoa tibetica OL03-7 ITS2/ETS Arctopoa eminens 7010-3 ITS/ETS Arctopoa tibetica S5481 ITS/ETS Arctopoa tibetica S7626 ITS/ETS Arctopoa tibetica OL03-7 ITS1/ETS Cinna latifolia Dupontia fisheri G6589 Dupontia fisheri G6699 Arctophila fulva Alopecurus borealis Alopecurus gerardii Bellardiochloa variegata Apera interrupta Beckmannia syzigachne Catabrosa werdermannii Puccinellia vahliana 10 changes 100 100 100 100 100 100 100 100 72 58 63 59 74 74 70 85 89 81 87 88 85 96 99 99 99 96 95 92 93 91 85 89 96 95 96 90 60 100 84 58 100 100 62 82 62 58 84 57 57 80 55 A A C C A Cinna arundinacea A Poa clade Poinae (minus Poa) clade Puccinelliinae clade P. sect. Parodiochloa clade P. subg. Sylvestres clade 100 100 100 100 100 100 100 100 78 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 100 95 100 53 87 100 78 96 67 88 89 95 55 67 100 87 85 90 96 55 99 - 51 - 71 99 99 97 100 100 57 * * * * * ** * * ** * * ** 1 2 Fig.?3. Phylogram with strict consensus tree shown in bold lines resulting from parsimony analysis of combined TLF, ITS and ETS data for subtribe Poinae, including putative hybrid taxa, plus representa- tive members of subtribes Alopecurinae, Cinninae, and Puccinelliinae (8300 trees, 1821 steps, CI = 0.524). MP bootstrap values >50% are given above branches, Bayesian posterior probability values below branches. Collection information is given only for species represented by multiple collections; variant ITS sequences are indicated as ?ITS1? or ?ITS2?; ?A? indicates species belonging to subtribe Alopecurinae, ?C? those belonging to Cinninae; ?*? indicates taxa with one-flowered spikelets, all others have multi-flowered spikelets. Putative hybrid taxa are indicated in bold. TLF and ITS/ETS data were not combined for Arctopoa and Nicoraepoa pugionifolia; each accession is represented by two OTUs indicated as ?TLF? and ?ITS/ETS?. Aniselytron OTUs are indicated by their constituent DNA regions described in detail in the text. Outgroup taxa are not shown. 1120_kap_18_32_6k.indd 603 07/06/10 21.23 604 Gillespie et al. formed a clade in TLF, ITS, ITS/ETS, or TLF/ITS/ETS unconstrained analyses. However, Alopecurus and Beckmannia did form a clade in the ETS analysis (not shown), sister to the Arctophila-Dupontia clade in the Poinae (minus Poa) major clade, the same position that Alopecurus occupied in the ITS (not shown), ITS/ ETS (Fig.? 1) and TLF/ITS/ETS trees (Figs.?2-3). Constraining Alopecurus (two species and collections) and Beckmannia (one collection) as monophyletic in the combined TLF/ITS/ETS datasets resulted in MP trees only four steps longer than trees from the unconstrained analyses (Figs.?2-3), whereas constraining all three genera, Alopecurus, Beckmannia, and Phleum (one collection), resulted in MP trees 15 steps longer. Strict consensus trees were identical to those in the unconstrained analyses, except for the positions of Beckmannia and Phleum, and the more or less resolved position of Cinna (e.g., more resolved in TLF/ITS/ ETS as sister to the Arctagrostis-Hookerochloa-Nicoraepoa clade). In all analyses Beckmannia and Phleum (where constrained) resolved with Alopecurus, in a clade sister to the Arctophila-Dupontia clade. Since five additional ITS sequences were available for Alopecurinae (Gillespie et al. 2008), we also ran a constraint analysis only on the ITS data matrix. In the unconstrained analysis each of the three genera were monophyletic: Alopecurus (four species, five collections: A. geniculatus and A. vaginatus added); Beckmannia (two species and collections: B. eruciformis added); and Phleum (two species and collections: P. phleoides added). Constrain- ing Alopecurus and Beckmannia as monophyletic in the ITS analysis resulted in MP trees one step longer than those from the unconstrained analysis, whereas constraining all three genera, Alopecurus, Beckmannia, and Phleum, resulted in a tree seven steps longer. In both analyses the constrained genera resolved as a clade sister to the Arctophila-Dupontia clade, as in the TLF/ITS/ETS constrained analysis. Discussion Analyses of combined plastid TLF and nrDNA ITS and ETS data resulted in a much more robust phylogenetic hypothesis for subtribes Poinae and related Alopecurinae, Cinninae, Miliinae, and Puccinelliinae (PPAM clade) (Figs.?2-3) compared with those obtained from previous separate analyses of plastid and ITS data (Gillespie et al. 2007, 2008). Likewise combining ETS and ITS data resulted in a more strongly supported nrDNA phylogeny (Fig.? 1) than that obtained with ITS data alone. TLF, ITS, and ETS data partitions were for the most part congruent. Although all three data partitions were incongruent for all taxa, this conflict was determined to lie in the Poa clade. Data partitions for all other genera, with only the Poa subg. Sylvestres clade included to represent Poa, were not significantly incongruent. Surprisingly, conflict within the Poa clade between ETS and ITS partitions was as high as between either of these nrDNA partitions and the plastid parti- tion. Positions of several diploid species of Poa (in particular P. infirma, P. let- 1120_kap_18_32_6k.indd 604 07/06/10 21.23 Phylogeny and Reticulation in Poinae 605 termanii, P. pseudoabbreviata, and P. supina) were determined to be incongruent between ETS and ITS trees. This finding will be explored in more detail in a future paper focusing on the phylogeny of Poa. Incongruence between ITS and ETS data is unusual, but has been recorded previously for several other taxa (Bena et al. 1998; Cubas et al. 2006; Mitsui et al. 2008) and may be due to concerted evolution operating at different rates and/or with different outcomes in the ETS and ITS regions. Nucleotide and indel sequence variants within a sample were detected in both diploid and polyploid Poinae taxa, in putative hybrid and non-hybrid taxa, and in all three DNA regions. The more complex variants, including the only case of divergent paralogous sequences, were found in the polyploid taxa Aniselytron and Nicoraepoa pugionifolia, providing further evidence for their putative hybrid status. PPAM?major clades?The phylogeny of the PPAM clade based on combined plastid and nrDNA data is mostly consistent with, but more strongly supported than, trees from previous separate analyses of plastid and ITS data (Gillespie et al. 2007, 2008). Likewise, the combined ITS/ETS phylogeny is mostly consist- ent with and more strongly supported than those based only on ITS or ETS data (Gillespie et al. 2008; and analyses performed here). All of these analyses focusing on the PPAM clade resolve three major clades: Poa; subtribe Poinae (minus Poa) including Alopecurus, Beckmannia (both Alopecurinae) and Cinna; and subtribe Puccinelliinae; plus two separate lineages, Milium and Phleum. The three major clades have good to strong support in the combined trees presented here (Figs.?1-3), whereas the Poa and Poinae (minus Poa) clades had bootstrap support less than 55% in ITS analyses and moderate to very good support in TLF analyses (Gillespie et al. 2007, 2008). The Puccinelliinae clade was well supported as sister to all remaining members of the PPAM clade in the combined TLF/ITS/ETS and ITS/ETS trees, consistent with previous trees based only on plastid data, but not with ITS alone. This supports recognition of a PAM clade (first recognized and named in Gillespie et al. 2007) compris- ing subtribes Alopecurinae, Miliinae, and Poinae as detected in several earlier analyses of plastid data alone (Soreng and Davis 2000 [note: Puccinellia stricta in error]; Davis and Soreng 2007; D?ring et al. 2007; Gillespie et al. 2007, 2008; Soreng et al. 2007). Our revised classification of the PPAM clade is presented in Table 3 and modifications are discussed below. Poa resolves as a strongly supported clade in combined ITS/ETS and TLF/ ITS/ETS trees, compared to only poorly to moderately supported in previous separate TLF and ITS analyses (Gillespie and Soreng 2005; Gillespie et al. 2007, 2008). We now have strong support for the genus as monophyletic. Recent taxo- nomic and nomenclatural changes based on molecular and morphological evi- dence have fine tuned its circumscription to exclude some species (e.g., P. subg. Andinae = Nicoraepoa) and submerge genera that resolved within it (reviewed in Introduction). Poa subg. Sylvestres is very well supported as a clade compared to 1120_kap_18_32_6k.indd 605 07/06/10 21.23 606 Gillespie et al. being only poorly supported in separate analyses, and its position as sister to all other Poa species is now strongly supported. Relationships within Poa based on combined TLF/ITS/ETS analyses will be explored in a separate paper. We re- main in doubt as to the sister group to Poa. TLF/ITS/ETS analyses suggest that Milium and Phleum may form the sister clade, but support for this clade is poor and its sister relationship to Poa has only poor to moderate bootstrap support (but high posterior probability in the Bayesian analyses). ITS/ETS MP analyses are even less conclusive, with relationships among the three major clades, Mil- ium, and Phleum all having no bootstrap support although a Milium-Phleum-Poa clade does have high posterior probability in the Bayesian analysis. Our results indicate that subtribe Poinae, as currently circumscribed, is paraphyletic. While the majority of Poinae genera form a strongly supported clade, two of three Alopecurinae genera (Alopecurus and Beckmannia) and the single genus of Cinninae examined here are firmly embedded in this clade. A second Cinninae genus, Cyathopus, also resolved with very good support in a clade equivalent to our Poinae (minus Poa) clade in the plastid matK analy- sis of D?ring et al. (2007). Even if Cinninae is submerged within Poinae (as recommended by Gillespie et al. 2008), Alopecurinae are sufficiently distinct morphologically to tentatively retain as a separate subtribe. Our results also suggest that the Poinae (minus Poa) clade and the Poa clade may not be sister 1. Subtribe Alopecurinae: Alopecurus L., Beckmannia Host, Cornucopiae L.^, Limnas Trin.^ 2. Subtribe Miliinae: Colpodium Trin., Milium L., Zingeria P.A. Smirn. 3. Subtribe Phleinae: Phleum L., Pseudophleum Dogan^, Rhizocephalus Boiss.^ 4. Subtribe Poinae: i. Poa clade: Poa (including Aphanelytrum Hack., Dissanthelium Trin., and Tovarochloa T. D. Macfarl. & But, plus genera now subsumed within, see Introduction) ii. Poinae (minus Poa)?clade: Aniselytron Merr.H, Apera Adans., Arctagrostis Griseb., Arctophila, (Rupr.) Anderss., Arctopoa (Griseb.) Prob.H, Bellardiochloa Chiov., Cinna L.C, Cyathopus StapfC, Dupontia R. Br., Gaudiniopsis (Boiss.) Eig, Hookerochloa E. B. Alexeev, Nicoraepoa Soreng & L. J. Gillespie, Ventenata Koeler iii. genera lacking DNA data: Libyella Pamp.^, Limnodea DeweyC^, Lindbergella Bor^, Nephelochloa Boiss.^ 5. Subtribe Puccinelliinae: Catabrosa P. Beauv., Catabrosella (Tzvelev) Tzvelev, Hyalopoa (Tzvelev) Tzvelev, Oreopoa H. Scholz & Parolly^, Paracolpodium (Tzvelev) Tzvelev, Phippsia (Trin.) R. Br., Pseudosclerochloa Tzvelev^, Puccinellia Parl., Sclerochloa P. Beauv. Table 3. Modified classification of subtribes and genera in the PPAM clade.C= subtribe Cinninae genera now placed in Poinae; H= genera of putative hybrid origin; ^= no DNA data. 1120_kap_18_32_6k.indd 606 07/06/10 21.23 Phylogeny and Reticulation in Poinae 607 taxa; their precise relationship remains unclear due to the equivocal positions of Phleum and Milium. While major changes in the circumscription of subtribe Poinae will undoubtedly be necessary to align the classification and molecular phylogeny, we feel that increased sampling and stronger support for relation- ships among major clades are necessary to support further taxonomic change. Both molecular results and morphology support the inclusion of Cinninae within subtribe Poinae as suggested by Gillespie et al. (2008). All three Cinninae genera (Cinna, Cyathopus, Limnodea) have single-flowered spikelets that disar- ticulate below the glumes, features that previously were used to place them in tribe Aveneae (Clayton and Renvoise 1986). While the majority of Poinae genera are characterized by multiple-flowered spikelets that disarticulate above the glumes, Apera, Aniselytron, Arctagrostis, and a few species of Poa (Tovarochloa and some New Guinea species) also have single-flowered spikelets, and Ventenata has spikelets that disarticulate below the glumes. Alopecurinae genera are also characterized by single-flowered spikelets that disarticulate below the glumes. When all of these characteristics are considered, Cinninae appears little differ- entiated morphologically from Poinae. Soreng et al. (2007) discussed homoplasy in these characteristics across tribe Poeae. The precise affinities of Milium and Phleum remain unclear. While well sup- ported as members of the PAM clade in combined plastid and nrDNA trees, they appear as two separate lineages distinct from the Poa and the Poinae (minus Poa) clades in all analyses focusing on the PPAM clade (Figs.?1-3; Gillespie et al. 2007, 2008). In the majority of analyses Milium and Phleum resolve as a clade (except in the ITS analysis), but this relationship has low support. Likewise, their positions within the PPAM clade have been poorly supported in most analyses. The combined plastid and nrDNA tree (Fig.?2) suggests a sister relationship with Poa with moderate bootstrap support and high posterior probability. Milium, in particular, is characterized by a high rate of nucleotide substitution in all three DNA regions analyzed, which may be a contributing factor in its equivo- cal position, possibly obscuring its true affinities. Our previous finding based on ITS data that Milium may be part of a separate Miliinae s.l. clade including Colpodium and Zingeria (Gillespie et al. 2008) needs to be tested with additional samples and data from other DNA regions. Poinae (minus Poa) clade?All Poinae genera analyzed, with the exception of Poa, are included within the strongly supported Poinae (minus Poa) clade. Also included here are two of three Alopecurinae examined and Cinna. In pre- vious plastid analyses relationships within this clade were largely unresolved or poorly supported (Gillespie et al. 2008: Fig.?1); only Apera-Bellardiochloa and Arctophila-Dupontia clades were strongly supported. ITS analyses were somewhat more resolved, having, in addition, a very well supported Arctagrostis-Hookerochloa (incl. Festucella)-Nicoraepoa clade (Gillespie et al. 2008: Fig.?2a). Combining ETS with ITS resulted in a nrDNA tree with increased support for relationships in the Poinae (minus Poa) clade, and analysis of combined plastid and nrDNA 1120_kap_18_32_6k.indd 607 07/06/10 21.23 608 Gillespie et al. provided even greater support. Structure within this clade was congruent and mostly strongly supported in trees resulting from both sets of analyses, sug- gesting that the hypothesis of relationships presented here is robust. Both ITS/ETS and TLF/ITS/ETS analyses resolved two major sister clades in the Poinae (minus Poa) clade. Arctagrostis-Hookerochloa-Nicoraepoa-Cinna (clade 1) and Alopecurus-Apera-Arctophila-Beckmannia-Bellardiochloa-Dupontia (clade 2) each have a high posterior probability in the Bayesian analysis, but are only poorly to well supported in the MP analysis. Both clades contained several strongly supported subclades. In clade 1, the following subclades are strongly supported and of biogeographically interest: Nicoraepoa (southern South American) and Arctagrostis (arctic), and these sister to Hookerochloa (Australia). Clade 2 included subclades: Apera-Bellardiochloa (both Eurasia), Arctophila-Dupontia (both arctic), and the latter sister to Alopecurus (distribution widespread) (the first two sub- clades strongly supported, the third with moderate to very good support). Subtribe Alopecurinae: Monophyletic, Paraphyletic or Polyphyletic??The three genera analyzed of subtribe Alopecurinae, Alopecurus, Beckmannia, and Phleum, did not resolve as monophyletic in separate TLF and ITS analyses or in combined analyses (Gillespie et al. 2007, 2008; results presented here). Alopecurus and Beckmannia did resolve as a clade in the ETS analysis; in other analyses nodes separating these two genera have at most poor to moderate sup- port (e.g., Figs.?2-3). Constraining Alopecurus and Beckmannia as monophyletic resulted in trees not considerably longer than the unconstrained analysis (Table 2, 1??4 steps). Thus, there appears to be no strong evidence contradicting the monophyly of Alopecurus and Beckmannia. In contrast, Phleum resolved as a distinct lineage distant from Alopecurus and Beckmannia in all analyses. Includ- ing Phleum in the monophyly constrained analyses substantially increased tree length (Table 2, 6-15 steps). Our results suggest that Alopecurinae, as currently treated, is polyphyletic and support splitting the subtribe into two or three lineages following Tzvelev (1976). Tzvelev recognized three subtribes in his tribe Phleeae: Alopecurinae, Beckmanniinae, and Phleinae. Our data strongly support recognizing Phleum as a distinct subtribe Phleinae (Table 3 or Appendix 2). In contrast, we tentatively leave Beckmannia within Alopecurinae since we have no strong evidence contradicting this relationship. The constrained Alopecurinae clade resolved as sister to the Arctophila-Dupon- tia clade in the Poinae (minus Poa) clade in all monophyly constrained analyses performed. Analyzing different data partitions, constraining only Alopecurus and Beckmannia versus all three genera (i.e., including Phleum), and including additional species (nine in total) in the ITS analysis did not change its posi- tion. The position of the constrained Alopecurinae clade is the same as that occupied by only Alopecurus in ITS, ITS/ETS, and TLF/ITS/ETS analyses, and by both Alopecurus and Beckmannia in the ETS analysis. Reticulate Evolution?? Cases of Intergeneric Hybridization??Combined plas- tid and nrDNA analyses presented here strongly support the hypothesis that 1120_kap_18_32_6k.indd 608 07/06/10 21.23 Phylogeny and Reticulation in Poinae 609 hybridization has played a role in the origin and evolution of genera in Poinae (Gillespie et al. 2008). ILD tests confirmed that plastid and nrDNA data for Arc- topoa and Nicoraepoa pugionifolia were highly incongruent, whereas Aniselytron was determined to have two highly divergent ITS paralogues. Including the putative hybrids as separate plastid and nrDNA OTUs (in the case of Aniselytron: sepa- rate TLF/ETS and ITS OTUs, and one combined TLF/ITS/ETS OTU) reduced resolution and support in the trees, particularly in the Arctagrostis-Hookerochloa- Nicoraepoa-Cinna clade where relationships among genera were all unresolved. Different levels of plastid versus nrDNA variation and the presence of missing data in the plastid only and nrDNA only OTUs likely contribute to this lower resolution and support. Recombination of two or more divergent repeats to form chimeric ITS and/or ETS sequences may be an additional contribution factor (Alvarez and Wendel 2003), although levels of recombination would pre- sumably be low here if this has occurred. Considerable incongruence between plastid and nrDNA data was also detected within Poa, and will be discussed in Soreng et al. (2010) and future papers focusing on hybridization in Poa. arctopoa ?Arctopoa species are tetraploid or hexaploid and occur in ripar- ian and littoral habitats of central to far eastern Asian mountains, steppes and coasts, and along seashores in boreal North America. Eight collections representing three of five species have now been examined for both plastid and nrDNA data (Gillespie et al. 2008; data presented here). All collections consistently resolve in two divergent positions: plastid OTUs in the Poa subg. Sylvestres clade, nrDNA OTUs in the Poinae (minus Poa) clade (Fig.?3). Addi- tional species and collections previously examined for only plastid or nrDNA resolve in positions consistent with our results, including A. eminens (plastid restriction site data, Soreng 1990) and A. schischkinii (Tzvelev) Prob. Tzvelev (ITS sequence data, Rodionov et al. 2005). These results provide strong evidence that Arctopoa is likely of intergeneric, and presumably ancient, hybrid origin. Our evidence suggests a member of the Poa subg. Sylvestres clade as a maternal parent (i.e., which donated a plastid genome), and a member of the Poinae Arctagrostis-Hookerochloa-Nicoraepoa-Cinna clade as a paternal parent. In both cases, no extant species group or genus can be readily identified as possible parents; further sampling of Asian Poa and Poinae may identify putative par- ents, although these may now be long extinct. Data from additional nuclear linkage groups could potentially provide further insight. Poa subg. Sylvestres, as currently understood, is endemic to temperate mesic woodlands in North America. The wetland genus Arctopoa bears little resemblance morphologically and shares no morphological synapomorphies with this subgenus. The pos- sibility of a sister relationship with Cinna as inferred by the nrDNA trees and the Bayesian combined plastid and nrDNA tree needs to be further explored, but again we know of no morphological synapomorphies for this set. Neither plastid nor nrDNA data support the current subdivision of Arctopoa into two sections, as currently circumscribed: sects. Arctopoa (A. eminens) and 1120_kap_18_32_6k.indd 609 07/06/10 21.23 610 Gillespie et al. Aphydris (A. schischkinii, A. subfastigiata, A. tibetica, and A. trautvetteri (Tzvelev) Prob.) (Probatova 1974; Tzvelev 1976). Both datasets suggest that A. subfastigiata and A. eminens, the two most salt tolerant species, are most closely related to each other (e.g., identical nucleotide characters in TLF), than either is to A. tibetica. This suggests that the distinctive features of A. eminens may be derived in this genus. One of the three collections examined of Arctopoa tibetica is suggested to be an infrageneric hybrid (also suggested in Gillespie et al. 2007, where it was included as a single polymorphic sequence). The sample contained two variant ITS sequences, one identical to the other two A. tibetica collections sampled, the other closer to A. eminens and A. subfastigiata. An alternative explanation might be incomplete lineage sorting and retention of two ancestral variants. aniselytron?Aniselytron includes two species, one widespread in south-east Asia (one hexaploid count), the other apparently endemic to Taiwan and the Philippines (ploidy unknown) (Lu and Phillips 2006). The genus was tradition- ally placed in Aveneae, near, or even included within, Calamagrostis Adans. (Clay- ton and Renvoize 1986), but recent molecular data on the widespread species placed it in the PAM clade (Soreng et al. 2007, with Poinae and allied subtribes in the ?Poeae type? plastid group; Gillespie et al. 2008). While the majority of sequence data (plastid, ETS, and one ITS sequence) place A. treutleri with the Poinae genera Arctagrostis, Hookerochloa, Nicoraepoa, and Cinna, a divergent ITS paralogue (two different collections) resolves in the P. subg. Sylvestres clade. This suggests that either hybridization was involved in the origin of Aniselytron or there has been more recent hybridization with Poa. As with Arctopoa, no extant genus or species group, respectively, have yet been identified as putative parents, but further sampling of co-occurring south-east Asian Poa and the Cinninae genus Cyathopus (the latter sampled only for plastid matK; D?ring et al. 2007) would be appropriate. Aniselytron shares with Cyathopus the character of single flowered spikelets, distinct from the typical multi-flowered spiketlets of Poa. The floret of Aniselytron is very similar to a typical Poa floret, but differs in its short stiff hairs on the front of the callus and on lower lemma margins (as oc- curs in Agrostis L.). Curiously, the positions occupied by Arctopoa and Aniselytron in both the Poa and Poinae (minus Poa) clades are very similar, but with plastid and ITS (except S5229 variant 2) OTUs reversed. Previous sampling also detected a more divergent ITS paralogue in Aniselytron that was not readily alignable (Gillespie et al. 2008). nicoraepoa ?N. pugionifolia, of the southern South American genus Nicorae- poa (ploidy unknown), was previously suggested to be of intergeneric hybrid ori- gin based on previous incongruent TLF and ITS data (Gillespie 2008). A second sample from a different location and our new ETS results confirm these two incongruent positions in the Poa clade (plastid OTUs) and the Poinae (minus Poa) clade (nrDNA OTUs). The species resolves in the Arctagrostis-Hookerochloa- 1120_kap_18_32_6k.indd 610 07/06/10 21.23 Phylogeny and Reticulation in Poinae 611 Nicoraepoa clade based on its plastid data, close to, but not in a clade with, other Nicoraepoa members (Fig.?3). The lower resolution and levels of support in this clade, including Nicoraepoa not resolving as monophyletic in the MP strict consensus tree, compared to the analysis without hybrid taxa (Fig.?2) are likely due, at least in part, to the low levels of plastid variation in this clade along with the absence of ITS/ETS data in these N. pugionifolia OTUs. nrDNA data provides strong evidence for a member of Poa sect. Parodiochloa as the paternal parent of N. pugionifolia. Poa flabellata is the only geographically-likely parental candidate in Poa, as it is the only member of this section occurring in South America (all other species are found on sub-Antarctic islands). Evolution involving hybridization is also suggested for Aniselytron and N. pugionifolia by the relatively high levels of infra-sample indel and base variation in sequences of all three DNA regions analyzed. Indel variants in the TLF da- taset were detected only in these two species, and in both collections of each sequenced. Divergent ITS sequences were detected in Aniselytron, the only case of divergent paralogues found in this study by direct sequencing of PCR amplified products, whereas N. pugionifolia had multiple indel variants in ITS, together accounting for the most complex patterns of within sample indel variation in ITS sequences so far detected in the PAM clade. Incomplete concerted evolution in nrDNA and the presence of indel variants in plastid sequences may possibly suggest a relatively recent origin for A. treutleri and N. pugionifolia. Conclusions This study advances our understanding of the phylogeny of the large sub- tribe Poinae and of the relationships of the smaller subtribes Alopecurinae, Cinninae, and Miliinae. Based primarily on our molecular results we recognize Cinninae within Poinae and subtribe Phleinae as separate from Alopecurinae. Among Poinae genera, Arctopoa is an old intergeneric hybrid, and Aniselytron and Nicoraepoa pugionifolia are interpreted as putative intergeneric hybrids or as resulting from more recent hybridization, with Poa involved in all three cases. Poinae as currently circumscribed is paraphyletic as Poa is not part of the main Poinae clade whereas Alopecurinae is nested within. Additional research is needed prior to taking further decisions on taxonomic changes in the circumscription of Poinae and the status of Alopecurinae. We plan to increase taxon sampling, particularly in Alopecurinae, Miliinae and Phleae, and sample additional DNA regions, along with an in-depth re-examination of morphological characters in a phylogenetic context, to advance our understanding of evolution in this large, taxonomically complex clade. acknowledgements. We thank the following people for providing silica-gel- dried leaf samples: Annie Archambault, Pilar Catal?n, Fran?oise Hennion, Surrey Jacobs, Ma Hai-Ying, Marina Olonova, Paul Peterson, and Sierra Stoneberg. We 1120_kap_18_32_6k.indd 611 07/06/10 21.23 612 Gillespie et al. gratefully acknowledge the French Polar Institute (IPEV) programme n?136, headed by Marc Lebouvier, for material of the sub-Antarctic island Poa kerguel- ensis. We thank Jeff Saarela for assisting with the Bayesian analysis and provided helpful comments on the manuscript; Julian Starr for help with primer design; and Travis Columbus and Frank Blattner for their helpful reviews of the manu- script. Molecular and field research of Gillespie was funded by grants from the Canadian Museum of Nature; part of the fieldwork of Soreng was funded by grants from the Smithsonian Institution. Literature Cited Acevedo-Rosas, R., K. Cameron, V. Sosa, and S. Pell. 2004. A molecular phylogenetic study of Graptopetalum (Crassulaceae) based on ETS, ITS, RPL16, and TRNL-F nucleotide sequences. American Journal of Botany 91: 1099-1104. Alvarez, I. and J. F. Wendel. 2003. Ribosomal ITS sequences and plant phylogenetic inference. Molecular Phylogenetics and Evolution 29: 417-434. Baldwin, B. G. and S. Markos. 1998. Phylogenetic utility of external transcribed spacer (ETS) of 18S-26S rDNA: congruence of ETS and ITS trees of Calycadenia (Compositae). Molecular Phylogenetics and Evolution 10: 449-463. Bena, G., J.-M Prosperi, B. Lejeune, and I. Olivieri. 1998. Evolution of annual species of the genus Medicago: a molecular phylogenetic approach. Molecular Phylogenetics and Evolution 9: 552-559. Clayton, W. D. and S. A. Renvoize. 1986. Genera Graminum: Grasses of the world. Kew Bulletin Additional Series 13: 1??389. Cubas, P.,?C. Pardo,?and H. Tahiri. 2006. Morphological convergence or lineage sorting? The case of Cytisus purgans auct. (Leguminosae). Taxon 55: 695-704. Davis, J. I. and R. J. Soreng. 2007. A preliminary phylogenetic analysis of the grass sub- family Pooideae (Poaceae), with attention to structural features of the plastid and nuclear genomes, including an intron loss in GBSSI. Aliso 23: 335-348. DeBry, R. W. and R. G. Olmstead. 2000. A simulation study of reduced tree-search effort in bootstrap resampling analysis. Systematic Biology 49: 171-179. D?ring, E., J. Albrecht, K. W. Hilu, and M. R?ser. 2007. Phylogenetic relationships in the Aveneae/Poeae complex (Pooideae, Poaceae). Kew Bulletin 62: 407-424. Farris, J. S., M. K?llersj?, A. G. Kluge, and C. Bult. 1995. Testing significance of incongru- ence. Cladistics 10: 315-319. Gillespie, L. J. and R. J. Soreng. 2005. A phylogenetic analysis of the bluegrass genus Poa L. (Poaceae) based on cpDNA restriction site data. Systematic Botany 30: 84-105. Gillespie, L. J., A. Archambault, and R. J. Soreng. 2007. Phylogeny of Poa (Poaceae) based on trnT-trnF sequence data: major clades and basal relationships. Aliso 23: 420-434. Gillespie, L. J., R. J. Soreng, R. D. Bull, S. W. L. Jacobs, and N. F. Refulio-Rodriguez. 2008. Phylogenetic relationships in subtribe Poinae (Poaceae, Poeae) based on nuclear ITS and plastid trnT-trnL-trnF sequences. Botany 86: 938-967. Gillespie, L. J., R. J. Soreng, and S. W. L. Jacobs. 2009. Phylogenetic relationships of Australian Poa (Poaceae: Poinae) including molecular evidence for two new genera, Saxipoa and Sylvipoa. Australian Systematic Botany 22: 413-436. 1120_kap_18_32_6k.indd 612 07/06/10 21.23 Phylogeny and Reticulation in Poinae 613 Grebenstein, B., M. R?ser, W. Sauer, and V. Hemleben. 1998. Molecular phylogenetic relationships in Aveneae (Poaceae) species and other grasses as inferred from ITS1 and ITS2 rDNA sequences. Plant Systematics and Evolution 213: 233-250. Hall, T. A. 1999. BioEdit: a user-friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. Nucleic Acids Symposium Series 41: 95-98. Holmgren, P. K. and N. H. Holmgren. 1998??. Index Herbariorum: A global directory of public herbaria and associated staff. New York Botanical Garden Virtual Herbarium. http: //sweetgum.nybg.org/ih/ Huelsenbeck, J. P. and F. Ronquist. 2001. MRBAYES: Bayesian inference of phylogeny. Bioinformatics 17: 754??755. Hunter, A. M., D. A. Orlovich, K. M. Lloyd, W. G. Lee, and D. J. Murphy. 2004. The generic position of Austrofestuca littoralis and the reinstatement of Hookerochloa and Festucella (Poaceae) based on evidence from nuclear (ITS) and chloroplast (trnL?trnF) DNA sequences. New Zealand Journal of Botany 42: 253-26Jacobs, S. W. L., L. J. Gillespie, and R. J. Soreng. 2008. New combinations in Hookerochloa and Poa (Gramineae). Telopea 12: 273-278. Jeanmougin, F., J. D. Thompson, M. Gouy, D. G. Higgins, and T. J. Gibson. 1998. Mul- tiple sequence alignment with Clustal X. Trends in Biochemical Sciences 23: 403-405. Li, J., J. H. Alexander, and D. Zhang. 2002. Paraphyletic Syringa (Oleaceae): evidence from sequences of nuclear ribosomal DNA ITS and ETS regions. Systematic Botany 27: 592-597. Lu, S.-L. and S. M. Phillips. 2006. Aniselytron Merr. p. 310??311. in Flora of China, vol. 22: Poaceae, eds. W. Zhengyi, P. H. Raven, and H. Deyuan. Beijing and St. Louis: Science Press and Missouri Botanical Garden Press. Markos, S. and B. G. Baldwin. 2001. Higher-level relationships and major lineages of Lessingia (Compositae, Astereae) based on nuclear rDNA internal and external tran- scribed spacer (ITS and ETS) sequences. Systematic Botany 26: 168-183. Mitsui Y., S.-T. Chen, Z.-K. Zhou, C.-I. Peng, Y.-F. Deng, and H. Setoguchi. 2008. Phy- logeny and Biogeography of the Genus Ainsliaea (Asteraceae) in the Sino-Japanese Region based on Nuclear rDNA and Plastid DNA Sequence Data. Annals of Botany 101: 111??124. Posada, D. and K. A. Crandall. 1998. ModelTest: Testing the model of DNA substitution. Bioinformatics 14: 817??818. Probatova, N. S., 1974. O novom rode Arctopoa (Griseb.) Probat. (Poaceae). Novosti Sis- tematiki Vysshchikh Rastinii (Leningrad) 11: 44??54. Quintanar, A., S. Castroviejo, and P. Catal?n. 2007. Phylogeny of the tribe Aveneae (Pooideae, Poaceae) inferred from plastid trnT-F and nuclear ITS sequences. American Journal of Botany 94: 1554-1569. Razafimandimbisona, S. G, J. Moog, H. Lantza, U. Maschwitz, and B. Bremer. 2005. Re- assessment of monophyly, evolution of myrmecophytism, and rapid radiation in Neonauclea s.s. (Rubiaceae). Molecular Phylogenetics and Evolution 34: 334-354. Refulio-Rodriguez, N. F. 2007. Systematics of Dissanthelium Trin. (Poaceae: Pooideae). Ph.D. Dissertation. Claremont: Claremont Graduate University, Rodionov, A. V., E. S.?Kim, E. O. Punina, E. M. Machs, N. B. Tyupa, and N. N. Nosov. 2007. Evolution of chromosome numbers in the tribes Aveneae and Poeae inferred 1120_kap_18_32_6k.indd 613 07/06/10 21.23 614 Gillespie et al. from the comparative analysis of the internal transcribed spacers ITS1 and ITS2 of nuclear 45S rRNA genes. Botanicheskii?Zhurnal 92: 57??71. [in Russian] Ronquist, F. and J. P. Huelsenbeck. 2003. MRBAYES 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19: 1572??1574. Soreng, R. J. 1990. Chloroplast-DNA phylogenetics and biogeography in a reticulating group: study in Poa. American Journal of Botany 77: 1383-1400. Soreng, R. J. and J. I. Davis. 2000. Phylogenetic structure in Poaceae subfamily Pooideae as inferred from molecular and morphological characters: misclassification versus reticulation. Pp.?. 61-74 in Grasses: systematics and evolution, eds. S. W. L.Jacobs and J. Everett. Melbourne: CSIRO Publishing. Soreng, R. J. and L. J. Gillespie. 2007. Nicoraepoa (Poaceae, Poeae), a new South American genus based on Poa subgenus Andinae, and emendation of Poa section Parodiochloa of the sub-Antarctic islands. Annals of the Missouri Botanical Garden 94: 821-849. Soreng, R. J., J. I. Davis, and J. J. Doyle. 1990. A phylogenetic analysis of chloroplast DNA restriction site variation in Poaceae subfamily Pooideae. Plant Systematics and Evolution 172: 83??97. Soreng, R. J., G. Davidse, P. M. Peterson, F. O. Zuloaga, E. J. Judziewicz, T. S. Filgueiras, and O. Morrone. 2003. Catalogue of New World grasses (Poaceae): IV. Subfamily Pooideae. Contributions from the United States National Herbarium 48: 1-730. [http: //mobot.mobot.org/W3T/Search/nwgclass.html] Soreng, R. J., J. I. Davis, and M. A. Voionmaa. 2007. A phylogenetic analysis of Poaceae tribe Poeae s.l. based on morphological characters and sequence data from three chloroplast-encoded genes: evidence for reticulation, and a new classification for the tribe. Kew Bulletin 62: 425-454. Soreng, R. J., R. D. Bull, and L. J. Gillespie. 2010. Phylogeny and reticulation in Poa L. based on plastid trnTLF and nrITS sequences with attention to diploids. Proceed- ings of the Fifth International Symposium on Grass Systematics and Evolution. Copenhagen, Denmark. August 11??15, 2008. (this volume) Starr, J. R., S. A. Harris, and D. A. Simpson. 2003. Potential of the 5? and 3? ends of the intergenic spacer (IGS) of rDNA in the Cyperaceae: new sequences for lower-level phylogenies in sedges with an example from Uncinia Pers. International Journal of Plant Sciences 164: 213-227. Starr, J. R., S. A. Harris, and D. A. Simpson. 2004. Phylogeny of the unispicate taxa in Cyperaceae tribe Cariceae I: Generic relationships and evolutionary scenarios. Sys- tematic Botany 29: 528-544. Subbotin, S. A., E. L. Krall, I. T. Riley, V. N. Chizhov, A. Staelens, M. De Loose, and M. Moens. 2004. Evolution of the gall-forming plant parasitic nematodes (Tylenchida: Anguinidae) and their relationships with hosts as inferred from Internal Transcribed Spacer sequences of nuclear ribosomal DNA. Molecular Phylogenetics and Evolution 30: 226??235. Swofford, D. L. 2002. PAUP*: phylogenetic analysis using parsimony (*and other methods), vers. 4b10. Sunderland: Sinauer Associates. Tzvelev, N. N. 1976. Zlaki SSSR. Leningrad: Nauka Publishers [English translation: 1983. Grasses of the Soviet Union, Vol. 1 and 2. New Delhi: Oxonian Press] Urbatsch, E. L., R. P. Roberts, and V. Karaman. 2003. Phylogenetic evaluation of Xylotha- 1120_kap_18_32_6k.indd 614 07/06/10 21.23 Phylogeny and Reticulation in Poinae 615 mia, Gundlachia, and related genera (Asteraceae, Astereae) based on ETS and ITS nrDNA sequence data. American Journal of Botany 90: 634-649. Appendix 1 Species and collections of tribe Poeae subtribes Alopecurinae, Cinninae, Milii- nae, Poinae, and Puccinelliinae (PPAM clade) and outgroup taxa sampled for plastid trnT-trnL-trnF and nrDNA ITS and ETS sequence data. Ploidy of species (where known), voucher information, and GenBank accession numbers of TLF, ITS, and ETS sequences, respectively, are provided. Sequences downloaded from GenBank are indicated by ?GB?, and literature reference where first published and place of origin, where known, are given. New sequences are indicated by * or, for those also included in Gillespie et al. 2009 (submitted after this paper), by ^; all other sequences are from Gillespie et al. (2007, 2008). Herbarium acro- nyms follow Holmgren and Holmgren (1998). * * * subtribe alopecurinae Alopecurus borealis Trin., 2n = 56: Canada, Nunavut, Gillespie et al. 6576 (CAN), DQ353966, EU792345, GQ324237^. Alopecurus geniculatus L.: GB: Quintanar et al. 2007, Spain, ?, DQ539571, ?. Alopecurus gerardii Vill., 2n = 14: Greece, Soreng et al. 7494 (US), EU792432, EU792344, GQ324238^. Alopecurus vaginatus (Willd.) Pall. ex Kunth: GB: Grebenstein et al. 1998, Caucasus, ?, Z96920/Z96921, ?; GB: Grebenstein et al. 1998, Caucasus, ?, Z96922/Z96923, ?. Beckmannia eru- ciformis (L.) Host: GB: Grebenstein, unpubl., ?, AJ389163, ?. Beckmannia syzi- gachne (Steud.) Fernald, 2n = 14, 28: USA, Wyoming, Soreng 3513 (US), DQ353965, EU792342, GQ324255^. Phleum phleoides (L.) Karsten., 2n = 14, 28+: GB: Subbotin et al. 2004, ?, AF498396, ?. Phleum pratense L., 2n = (14) 42, 56, 63, 70, 84: USA, New York, cultivated, no voucher (contaminant in USDA 20228 seed acc.), DQ353964, EU792341, GQ324284^. subtribe cinninae Cinna arundinacea L., 2n = 28: USA, West Virginia, Soreng & Olonova 7462 (US), EU792436, EU792343, GQ324260^. Cinna latifolia (Trevir. ex G?pp.) Griseb.: USA, California, Peterson et al. 19769 (US), ?, ?, GQ324261^; GB: Quintanar et al. 2007, Finland, DQ631498/DQ631432, DQ539569, ?. subtribe miliinae Milium effusum L., 2n = 14, 28: Canada, Quebec, Gillespie 7422 (CAN), ?, GQ324477^, GQ324273^; GB: Quintanar et al. 2007, Finland, DQ631501/DQ631435, ?, ?. Milium vernale M. Bieb., 2n = 14, 28: Greece, Soreng et al. 3748 (US), DQ353963, EU792340, GQ324274^. 1120_kap_18_32_6k.indd 615 07/06/10 21.23 616 Gillespie et al. subtribe poinae Aniselytron treutleri (Kunze) Soj?k, 2n = 42: China, Ma 140 (KUN), GQ324394*, ?, GQ324274*; China, Soreng et al. 5229 (US), GQ324395^, EU792441, GQ324469^(1) GQ324468*(2), GQ324240^; China, Soreng et al. 5264 (US), ?, EU792373, ?. Apera interrupta (L.) P. Beauv., 2n = 14: Argentina, Peterson et al. 19173 (US), EU792439, EU792364, GQ324242^. Arctagrostis latifolia (R. Br.) Griseb., 2n = 56: Canada, Nunavut, Archambault 145 (CAN), ?, EU792352, GQ324243*; Canada, Nunavut, Ar- chambault 157 (CAN), EU792434, EU792353, GQ324244^; Canada, Nunavut, Gillespie et al. 6586 (CAN), DQ353969, EU792351, GQ324245^. Arctophila fulva (Trin.) Rupr., 2n = 42: Canada, Northwest Territories, Aiken 99-230 (CAN), DQ354058, EU792347, GQ324246^. Arctopoa eminens (J. Presl) Prob., 2n = 28, 42, 62: Canada, Labrador, Gillespie 7010-2 (CAN), EU792446, GQ324470*, GQ324247*; Canada, Labrador, Gillespie 7010-3 (CAN), EU792447, EU792367, GQ324248*; USA, Alaska, Soreng & Soreng 6022 (US), DQ353977, EU792366, GQ324249*. Arctopoa subfastigiata (Trin.) Prob., 2n = 28, 42, 91, 97, Mongolia, W6 18199 W94096 (GRIN ID#), EU792448, EU792372, GQ324251*; Mongolia, W6 18244 W94096 (GRIN ID#), EU792449, EU792371, GQ324250*. Arctopoa tibetica (Stapf) Prob., 2n = 42, Russia, Olonova 2003-07 (CAN), EU792444, GQ324471*(1) GQ324472*(2), GQ324252*; China, Tibet, Soreng & Peterson 5481 (US), DQ353976, EU792368, GQ324253*; Kyrgyz Republic, Soreng et al. 7626 (US), EU792445, EU792370, GQ324254*. Bellardiochloa polychroa (Trautv.) Roshev.: Turkey, Soreng & Guney 4191 (US), ?, EU792363, GQ324256^. Bellardiochloa variegata (Lam.) Kerguelen, 2n = 14: Greece, Soreng et al. 7519-1 (US), EU792438, EU792361, GQ324257^. Dupontia fisheri R. Br., 2n = 44, 88: Canada, Nunavut, Gillespie et al. 6589 (CAN), DQ353967, EU792346, GQ324266^; Canada, Nunavut, Gillespie et al. 6699 (CAN), DQ353968, GQ324475^, GQ324267^. Hooke- rochloa eriopoda (Vickery) S.W.L. Jacobs: Australia, Jacobs 9128 (NSW), EU792433, EU792349, GQ324270^; Australia, Jacobs 9129 (NSW), GQ324397^, EU792350, GQ324271^. Hookerochloa hookeriana (F. Muell. ex Hook.f.) E.B. Alexeev: Aus- tralia, Jacobs 9127 (NSW), EU792435, EU792348, GQ324272^. Nicoraepoa andina (Trin.) Soreng & L.J. Gillespie: Chile, Soreng & Soreng 7182 (US), DQ353971, EU792354, GQ324275^. Nicoraepoa chonotica (Phil.) Soreng & L.J. Gillespie: Chile, Soreng & Soreng 7309 (US), DQ353974, EU792355, GQ324276^. Nicoraepoa pugionifolia (Speg.) Soreng & L.J. Gillespie: Argentina, Peterson et al. 17128 (US), GQ324398*, GQ324478*, GQ324277*; Chile, Soreng & Soreng 7336 (US), DQ353973, EU792360, GQ324278*. Nicoraepoa robusta (Steud.) Soreng & L.J. Gillespie: Chile, Soreng & Soreng 7358 (US), DQ353975, EU792357, GQ324279^; Chile, Soreng & Soreng 7359 (US), ?, EU792356, GQ324280*. Nicoraepoa subenervis (Hack.) Soreng & L.J. Gillespie subsp. subenervis: Chile, Soreng & Soreng 7334 (US), DQ353972, EU792359, GQ324283^. Nicoraepoa subenervis subsp. spegazziniana (Nicora) Soreng & L.J. Gillespie: Argentina, Peterson et al. 19186 (US), EU792443, EU792358, GQ324281^; Chile, Soreng 7155 (US), EU792442, GQ324479^, GQ324282^. Poa alsodes A. Gray: Canada, Quebec, Gillespie 6467 (CAN), DQ353981, EU792374, GQ324288^. Poa autumnalis Elliott, 2n = 28: USA, Maryland, Soreng 4680 (US), 1120_kap_18_32_6k.indd 616 07/06/10 21.23 DQ353979, EU792379, GQ324294^. Poa badensis Haenke ex Willd., 2n = 14, 18?21, 28: Bulgaria, Hajkova 2004-12 (US), GQ324402^, GQ324490^, GQ324295^. Poa chaixii Vill., 2n = 14: Russia, Soreng 4677 (US), EU854590, EU792404, GQ324299^. Poa diaphora Trin., 2n = 14, 28, 42: Turkey, Soreng & G?ney 4165 (US), DQ353988, EU792400, GQ324311^. Poa dolosa Boiss. & Heldr., 2n = 14: Greece, Soreng et al. 7495-1 (US), GQ324402^, GQ324502^, GQ324312^. Poa flabellata (Lam.) Ra- spail, 2n = 28: Falkland Islands, Wright 4NCD (not vouchered), DQ353982, EU792380, GQ324320^. Poa infirma Kunth, 2n = 14: Spain, Catalan 3-2000 (UZ), GQ324427^, GQ324516^, GQ324334^. Poa kerguelensis (Hook. f.) Steud: Sub- antarctic Islands, Kerguelen Islands, Hennion Gen5 (P), EU792457, EU792385, GQ324336^. Poa lettermanii Vasey, 2n = 14, USA, Colorado, Soreng & Soreng 7434 (US), GQ324431^, GQ324521^, GQ324345^. Poa ligulata Boiss., 2n = 14, Spain, (JACA 166095), GQ324432^, GQ324522^, GQ324346^. Poa media (L.) Cav., 2n = 14, Europe, Stoneberg SH17 (US), GQ324437^, GQ324527^, GQ324352^. Poa mo- linerii Balb., 2n = 14, 28: Slovakia, Stoneberg SH13 (CAN), DQ354036/DQ354037, EU792389, GQ324354^. Poa remota Forselles, 2n = 14: Kyrgyz Republic, Soreng et al. 7540 (US), GQ324452^, GQ324545^, GQ324372^. Poa saltuensis Fernald & Wiegand, 2n = 28: Canada, Ontario, Gillespie 7043 (CAN), EU792451, EU792378, GQ324374^. Poa sibirica Roshev. subsp. sibirica, 2n = 14: Russia, Khakasia, Olonova 2002-1 (CAN), DQ354044/DQ354045, EU792401, GQ324376^. Poa supina Schrad., 2n = 14: USA, cultivated, Soreng & Cayouette 5950-2 (US), DQ353984, EU792387, GQ324383^. Poa sylvestris A. Gray, 2n = 28: USA, Maryland, Soreng 4678-3 (US), DQ353980, EU792375, GQ324384^. Poa trichophylla Heldr. & Sart. ex Boiss., 2n = 14: Greece, Soreng et al. 7508 (US), GQ324461^, GQ324554^, GQ324386^. Poa wolfii Scribn., 2n = 28: USA, Missouri, Soreng 5800 (US), DQ354032/DQ354033, EU792377, GQ324389^. subtribe puccinelliinae Catabrosa werdermannii (Pilg.) Nicora & R?golo: Argentina, Peterson et al. 19371 (US), EU792431, EU792333, GQ324259^. Puccinellia vahliana (Leibm.) Scribn., 2n = 14: Gillespie 5808 (CAN), Canada, Nunavut, EU854591, EU792336, GQ324285^. Outgroups: aveneae subtribe aveninae Helictotrichon sempervirens (Vill.) Pilg., 2n = 42: Soreng 4622 (US), USA, New York, cultivated, DQ353955, EU792325, GQ324269^. poeae subtribe loliinae Festuca baffinensis Polunin, 2n = 28: Gillespie & Consaul 6920 (CAN), Canada, Northwest Territories, DQ353951/DQ353952, GQ324476^, GQ324268^. poeae subtribe airinae Deschampsia brevifolia R. Br., 2n = 26, 27, 28, 50, 52: Gillespie & Consaul 6810b (CAN), Canada, Northwest Territories, DQ353962, EU792328, GQ324262^ 1120_kap_18_32_6k.indd 617 07/06/10 21.23