Grass plastomes reveal unexpected paraphyly with endemic species of Micrairoideae from India and new haplotype markers in Arundinoideae 1

Transcription

1 RESEARCH ARTICLE AMERICAN JOURNAL OF BOTANY Grass plastomes reveal unexpected paraphyly with endemic species of Micrairoideae from India and new haplotype markers in Arundinoideae 1 Melvin R. Duvall 2,5, Shrirang R. Yadav 3, Sean V. Burke 2, and William P. Wysocki 2,4 PREMISE OF THE STUDY: We investigated the little-studied Arundinoideae/Micrairoideae clade of grasses with an innovative plastome phylogenomic approach. This method gives robust results for taxa of uncertain phylogenetic placement. Arundinoideae comprise ~45 species, although historically was much larger. Arundinoideae is notable for the widely invasive Phragmites australis. Micrairoideae comprise nine genera and ~200 species. Some are threatened with extinction, including Hubbardia, some Isachne spp., and Limnopoa. Two micrairoid genera, Eriachne and Pheidochloa, exhibit C 4 photosynthesis in this otherwise C 3 subfamily and represent an independent origin of the C 4 pathway among grasses. METHODS: Five new plastomes were sequenced with next-generation sequencing-by-synthesis methods. Plastomes were assembled by de novo methods and phylogenetically analyzed with eight other recently published arundinoid or micrairoid plastomes and 11 outgroup species. Stable carbon isotope ratios were determined for micrairoid and arundinoid species to investigate ambiguities in the proxy evidence for C 4 photosynthesis. KEY RESULTS: Phylogenomic analyses showed strong support for ingroup nodes in the Arundinoideae/Micrairoideae subtree, including a paraphyletic clade of Hubbardieae with Isachneae. Anatomical, biochemical, and positively selected sites data are ambiguous with regard to the photosynthetic pathways in Micrairoideae. Species of Hubbardia, Isachne, and Limnopoa were definitively shown by δ 13 C measurements to be C 3 and Eriachne to be C 4. CONCLUSIONS: Our plastome phylogenomic analyses for Micrairoideae are the first phylogenetic results to indicate paraphyly between Isachneae and Hubbardieae. The definitive δ 13 C data for four genera of Micrairoideae indicates the breadth of variation possible in the proxy evidence for photosynthetic pathways of both C 3 taxa. KEY WORDS Arundinoideae; C /C photosynthesis; Hubbardia ; Micrairoideae; Phragmites australis ; plastome phylogenomics; Poaceae; rbcl evolution 3 4 Grasses (Poaceae) comprise 12 well-supported lineages, which are recognized as subfamilies, ranging in size from 4 to 3900 species ( GPWG I, 2001 ; Soreng et al., 2015 ). In the crown grass clade (CGC), nine of the twelve subfamilies diverged into two major subgroups: BOP (Bambusoideae + Oryzoideae + Pooideae) and PACMAD (Panicoideae + Aristidoideae + Chloridoideae + Micrairoideae + Arundinoideae + Danthonioideae). Here we examine two closely related PACMAD lineages that comprise a clade of what are now recognized as the subfamilies Arundinoideae and Micrairoideae 1 Manuscript received 29 July 2016; revision accepted 6 January Biological Sciences, Northern Illinois University, 1425 W. Lincoln Hwy, DeKalb, Illinois USA; and 3 Department of Botany, Shivaji University, Kolhapur , India 4 Present address: Center for Data Intensive Sciences, University of Chicago, 5454 S. Shore Dr., Chicago, IL USA 5 Author for correspondence ( mel-duvall@niu.edu) doi: /ajb ( GPWG I, 2001 ; GPWG II, 2012 ). Grasses in this clade have been the subject of considerable taxonomic revisions and require the application of new tools to improve our understanding of their relationships. Since the 19th century species in these groups were variously classified in Bambusoideae, Eragrostoideae, Panicoideae, or Pooideae (see Sánchez-Ken and al., 2007 for a detailed history). Even recently, Arundinoideae were described as the most unresolved subfamily of grasses ( Hardion et al., 2012 ). Species of Eriachneae, Isachneae, and Micraireae were especially problematic and historically were classified in Arundinoideae or Panicoideae. The combination of characteristics in Eriachneae are so unusual that there was a suggestion to segregate this tribe in its own subfamily ( myspecies.info/ [accessed 5 May 2016]). In later molecular analyses, where representatives from two or three core genera were sampled, Arundinoideae and Micrairoideae were resolved with varying levels of support as monophyletic groups that were distinct from 286 AMERICAN JOURNAL OF BOTANY 104 (2): , 2017; Duvall et al. Published by the Botanical Society of America. This work is licensed under a Creative Commons Attribution License (CC-BY-NC).

2 FEBRUARY 2017, VOLUME 104 DUVALL ET AL. PLASTOME PHYLOGENOMICS OF ARUNDINOIDEAE/MICRAIROIDEAE 287 Panicoideae and other subfamilies, but not consistently placed in the grass phylogeny ( Duvall et al., 2007 ; Sánchez-Ken et al., 2007 ; Bouchenak-Khelladi et al., 2008 ; GPWG II, 2012 ). In analyses with balanced sampling between subfamilies and/or plastid genome (plastome) scale sampling of markers, Arundinoideae and Micrairoideae were retrieved as sisters ( Cotton et al., 2015 ; Duvall et al., 2016 ; GPWG II, 2012 ). These phylogenetic results supported the recircumscription of Arundinoideae to a monophyletic core group and the reinstatement of Micrairoideae about 50 yr after its original description, which are significant advances in grass systematics. However, morphological synapomorphies for either subfamily have yet to be discovered ( Kellogg, 2015 ; Sánchez-Ken et al., 2007 ). Arundinoideae Broad historical interpretations of this subfamily created a traditional Arundinoideae that was later judged to be polyphyletic. Early molecular studies ( Barker et al., 1995 ; Clark et al., 1995 ) identified components of the group that were unrelated, which were then placed in other subfamilies. In one other case, a species that was previously misclassified as Eragrostis walteri (a chloridoid genus) was decisively shown to be an arundinoid ( Ingram et al., 2010 ), although it has not been formally renamed. The present circumscription of Arundinoideae has been restricted to genera, most of which are monotypic, and species, reduced from the original 736 species ( Kellogg, 2015 ; Soreng et al., 2015 ). Two groups within the subfamily that were previously delimited on the basis of structural data were the crinipoids ( Linder et al., 1997 ) and an unnamed group of four genera ( Kellogg, 2015 ), but neither of these have been officially recognized. Soreng et al. (2015) divided the subfamily into Arundineae, a tribe of three genera, and Molinieae, which comprise the remainder. Most species of Arundinoideae have Old World distributions. However, Phragmites australis and Arundo donax are globally distributed as a result of widespread introductions, and both species are considered to be invasive ( Hardion et al., 2012 ; Lambert et al., 2010 ). The two species are superficially similar, but can be distinguished in having proximal florets that are either staminate or sterile ( P. australis ) or bisexual ( A. donax ) (Kellogg, 2015 ). Saltonstall (2002, 2003 ), working with P. australis, was able to discriminate distinct genetic lineages. These lineages showed significant geographic structure. Ten extant lineages in the North American native, P. australis subsp. americanus Saltonstall, P.M. Peterson & Soreng, which were identified on the basis of two noncoding plastid DNA markers, were mapped to specific continental regions. There are two other distinguishable extant haplotypes, one that extends from the Gulf Coast into Central America [ P. australis subsp. berlandieri (E.Fourn.) Saltonstall & Hauber] and a second presumed native to Eurasia ( Saltonstall and Hauber, 2007 ). The Eurasian haplotype was introduced into contemporary North American landscapes where it is replacing native haplotypes ( Saltonstall, 2003 ). The plastid locus haplotypes of P. australis have been extended to the determination of complete plastome sequences in two cases: the Eurasian haplotype, designated haplotype M and the most common of the native haplotypes, E (unpublished GenBank records). Micrairoideae Tribal delimitations in Micrairoideae are somewhat problematic because intergeneric relationships are not well understood. The nine described genera of Micrairoideae, as the subfamily is currently delimited, are grouped into four tribes ( Soreng et al., 2015 ; Watson and Dallwitz, 1992 ). The most diverse of these, Isachneae, includes two large genera, Coelachne R.Br. and Isachne R.Br., and three monotypic genera Heteranthoecia Stapf, Limnopoa C.E.Hubb., and Sphaerocaryum Nees ex Hook. f. Some species of Isachneae are extremely rare, largely because of habitat disturbance. One of these from moist forest habitats in western India, I. dimyloides Bor, was believed to be extinct until its recent rediscovery ( Devi and Bhattacharyya, 2015 ). Another species in this tribe, Limnopoa meeboldii (C.E.C.Fisch.) C.E.Hubb., has a floating unbranched habit that is unique among Micrairoideae. This species is an endemic of coastal lagoons in southwestern India. Limnopoa meeboldii is known from only four habitat fragments and is considered to be endangered ( Kumar et al., 2011 ; ). Eriachneae, with the two genera Eriachne R.Br. (40 48 spp.) and Pheidochloa S.T.Blake (two spp.), are defined as the only C 4 taxa in the subfamily. Photosynthetic pathway is the primary basis for distinguishing Eriachneae from the other tribes. Phylogenetic surveys suggest that Eriachneae was one of independent origins of C 4 lineages of PACMAD grasses ( Christin et al., 2008b ; GPWG II, 2012 ). Micraireae comprise a single genus of 13 species endemic to Australia. Species of Micraira F.Muell. uniquely produce leaf primordia in a spiral, rather than in a distichous arrangement. Paleas are also distinctively multinerved or divided into halves in the species of this tribe ( Clayton and Renvoize, 1986 ). This unusual combination of morphological characters was the basis for the original segregation of Micraira and initial recognition of this distinctive subfamily ( Kellogg, 2015 ). Hubbardieae also comprise a single genus. Hubbardia heptaneuron Bor, which grows in moist, shady, rock outcrops was on the verge of extinction until a successful species recovery initiative ( Chandore et al., 2012 ; Yadav et al., 2010 ). The other species, H. diandra Chandore, Gosavi & S.R.Yadav, is similarly rare, and both species are critically endangered. Hubbardia was originally classified in Isachneae, but later moved to a separate tribe. Hubbardieae are distinguished from Isachneae in reproductive morphology by an absence of paleas and in vegetative morphology by an absence of ligules ( Clayton and Renvoize, 1986 ). This taxonomic change might also have been partly motivated by the conservation status of Hubbardia. The separate monophylies of Isachneae and Hubbardieae have not been tested with phylogenetic methods. Whether the similarly rare but arguably more distinctive L. meeboldii also merits segregation into a separate tribe has not been assessed. Photosynthetic pathway is an issue of some relevance to intertribal and intergeneric relationships in Micrairoideae. The leaf anatomies of both C 3 micrairoid grasses are atypical. Both of the C 4 genera, Eriachne and Pheidochloa, have been biochemically typed as NADP-ME ( Dengler et al., 1994 ). The expected anatomy and associated ontogeny that is typical for NADP-ME species ( Hattersley and Watson, 1976 ) is found in neither genus of Eriachneae ( Prendergast et al., 1987 ). Both genera exhibit an unusual nonclassical anatomical characters more typical of NAD-ME or PCK subtypes of C 4 photosynthesis ( Dengler et al., 1994 ; Voznesenskaya et al., 2005 ). This inconsistency obscures the precise C 4 status of Eriachneae. Another confounding issue is that the species in two of the tribes that are presumed C 3, Isachneae and Hubbardieae, are indicated to have mesophyll radiate around the vascular bundles, similar to Kranz anatomy but the vein spacing distant ( Kellogg, 2015 ; p. 400). In other words, the leaf anatomies of species in these two tribes simultaneously show some features of C 3 grasses, although C C4 3 intermediacy has not been proposed as an explanation.

3 288 AMERICAN JOURNAL OF BOTANY Recent research by Christin et al. (2008a) suggests an alternative way to assess the photosynthetic pathway of plants. These researchers examined positively selected amino acids (AAs) in rbcl loci of C 3 grasses. Contrasting AAs were found at five sites with high probabilities depending on the photosynthetic pathway for a given species. For example, at AA position 101, V is the typical C 3 residue, which is usually encoded as GTA. However, in C 4 species of grasses and sedges, there is a high probability of a first position transition in codon 101 resulting in a change of AA from V to I. The alternative AA at five positively selected sites was suggested to represent functional trade-offs between substrate specificity and catalytic efficiency that are expected to be differentially selected in C 3 plants ( Christin et al., 2008a ). These five AAs can then be used as a sort of probability signature to indicate the type of photosynthetic pathway. Only one C 4 micrairoid was included in their analysis, an undetermined species of Eriachne. Unlike most other C 4 cyperoid and graminoid species, this Eriachne sp. showed four AA residues more typical of C 3 species and only one expected in a C 4 species, which is an unexpected observation. The ultimate resolution of questions regarding photosynthetic pathway requires analysis of stable carbon isotope ratios, but to our knowledge this type of measurement has not been performed on any Micrairoideae. One relatively new tool that can be applied to questions in plant systematics is plastome phylogenomic analyses. Extensive plastome phylogenomic analyses of CGC grasses are reported ( Wu et al., 2009 ; Wu and Ge, 2012 ; Cotton et al., 2015 ; Saarela et al., 2015 ; Wysocki et al., 2015 ; Burke et al., 2016 ; Duvall et al., 2016 ). In these studies, the substantial phylogenetic signal in the complete plastome improves resolution and raises support values across the phylogeny so that alternative hypotheses can be rigorously tested. Previous plastome phylogenomic studies have examined partitioning schemes to remove noise that conflicts with the principal phylogenetic signal. Partitioned approaches that separately analyze coding sequences or major plastome subregions have failed to show clear-cut advantages over the unpartitioned approach ( Zhang et al., 2011 ; Burke et al., 2012 ; Ma et al., 2014 ; Cotton et al., 2015 ; Saarela et al., 2015 ). Analysis of all of the unambiguously aligned coding and noncoding sequences in unpartitioned studies substantially increased the available phylogenetic information and uniformly raised support values across phylogenies. Here we apply similar methodologies to studying intergeneric relationships of the arundinoid/micrairoid complex. Four specific objectives were addressed in this study. (1) Selected Arundinoideae were investigated with plastome phylogenomics including a new plastome for Molinia Schrank. (2) Levels of nucleotide variation in whole plastomes of native and invasive haplotypes of P. australis were contrasted with those found between congeneric species. A new marker region was identified that clearly distinguished the three haplotypes of P. australis with unique patterns of indels. (3) Intertribal relationships of Micrairoideae were evaluated with plastome phylogenomics since there are contemporary differences in classifications at this level. Included were species of the previously unrepresented genera Hubbardia and Limnopoa. Note that at present, no parallel nuclear studies were conducted for arundinoid/ micrairoid grasses. (4) Micrairoid photosynthetic types (C 3 or C 4 ) were examined in a phylogenomic context. For the fourth objective, the evidence for photosynthetic pathway in the rbcl loci of presumed C 4 micrairoid species was examined by surveying positively selected AA positions (following Christin et al., 2008a ). Carbon isotope ratios were obtained for selected species as sampling permitted and compared against those from C 3 reference species. MATERIALS AND METHODS New sequence data Leaf tissues were obtained from representative species of Arundinoideae and Micrairoideae. The sampling in this complex was constrained by the fact that some species are quite rare or restricted to remote areas. Sources of tissues were either silica-dried leaves [ E. armitii F.Muell. ex Benth., H. diandra, L. meeboldii, and M. caerulea (L.) Moench] or, in one case, leaf fragments of a dried herbarium specimen ( Phragmites australis subsp. americanus ). The plastomes of two haplotypes of P. australis, an invasive (M) and a native (E) were previously sequenced. The plastome of a third, western North American haplotype H was sequenced here to allow further comparisons. The haplotype was inferred from the locality of this specimen in southern central California and confirmed with publicly available molecular markers (see below). Herbarium vouchers are indicated ( Table 1 ). Leaf tissue was homogenized in liquid nitrogen, and DNA was extracted with the DNeasy Plant Minikit (Qiagen, Valencia, California, USA) following the standard protocol issued by the manufacturer. Two library preparation methods from the same manufacturer were chosen to accommodate different starting quantities of DNA, so that the Nextera library preparation kit (Illumina, San Diego, CA, USA) was used for E. armitii, H. diandra, L. meeboldii, and M. cearulea. The Nextera XT kit was used for P. australis. Samples were quantified using the Qubit fluorometric system and diluted to 2.5 ng/μl in 20 μl (Nextera) or 0.2 ng/μl in 5 μl (Nextera XT). Library preparation, sequencing, and plastome assembly Libraries were prepared using the standard protocol of the respective Illumina library preparation kit. Dual index adaptors were used. Two Illumina platforms were used because of an equipment upgrade. Sequencing was performed on the HiSeq 3000 platform ( M. caerulea ), producing paired-end (PE) reads, or the HiSeq 2500 platform (other four species), producing single-end (SE) reads, at a commercial facility (Iowa State University DNA Sequencing Facility, Ames, Iowa, USA). Plastome assembly was performed using completely de novo methods following Wysocki et al. (2014). Briefly, contigs were assembled from Nextera sequence data with the Velvet software package ( Zerbino and Birney, 2008 ) in four steps, loading contigs from each previous step into the assembler with k -mer lengths increasing in steps of six from 19 to 85 bp. For the Nextera XT data, SPAdes v ( Bankevich et al., 2012 ; spades ) was used with k -mers as specified above. Assembled contigs for each species were scaffolded with the anchored conserved region extension method. Any remaining gaps between contigs in the scaffolds were resolved by mapping contigs or reads in Geneious Pro v8.0.2 (Biomatters, Auckland, New Zealand). The minimum acceptable overlap for resolving gaps between contigs was 20 identical base pairs. A final verification step was performed by mapping the quality-trimmed read pool against the draft de novo assembly in Geneious Pro to look for inconsistencies in base composition and overlap. Read depths were determined from this mapping. Plastome phylogenomics A total of 24 complete plastomes were analyzed. These included (1) the five newly sequenced plastomes;

4 FEBRUARY 2017, VOLUME 104 DUVALL ET AL. PLASTOME PHYLOGENOMICS OF ARUNDINOIDEAE/MICRAIROIDEAE 289 TABLE 1. Sources of material and GenBank accessions for plastomes from two Arundinoideae and three Micrairoideae newly sequenced in this study. Herbarium acronyms can be found at Species Origin, Voucher (Herbarium) GenBank accessions Eriachne armitii F.Muell. ex Benth. Canada, J. Saarela 1808 (CAN) KX Hubbardia diandra Chandore, Gosavi & S.R.Yadav India, S. Yadav s. n. (SUK) (see below) a Limnopoa meeboldii (C.E.C.Fisch.) C.E.Hubb. India, S. Yadav s. n. (SUK) (see below) a Molinia caerulea (L.) Moench USA, L. Clark 1684 (ISC) KX Phragmites australis subsp. americanus Saltonst., P.M.Peterson & Soreng USA, D. Bell 4664 (RSA) KX a Material handled under permit from the Indian National Biodiversity Act (NBA), permit #18/16/16-17/3175, for conducting noncommercial research outside India. Plastome sequences for the Indian endemics H. diandra and L. meeboldii were provisionally banked, but are now available on request to MRD to comply with NBA regulations. (2) five previously published arundinoid plastomes [ Elytrophorus spicatus (Willd.) A.Camus (GenBank accession NC_025233), Hakonechloa macra Makino (NC_025235), Monachather paradoxus Steud. (NC_025237), P. australis haplotype E (KJ825856), and P. australis haplotype M (NC_022958)]; (3) three previously published micrairoid plastomes [ E. stipacea F.Muell. (NC_025234), I. distichophylla Munro ex Hillebr. (NC_025236), and M. spiciforma Lazarides (KJ920234)]; and (4) 11 outgroup plastomes [ Bouteloua curtipendula (Michx.) Torr. (NC_029414), Centropodia glauca (Nees) Cope (NC_029411), Chaetobromus involucratus Nees subsp. dregeanus (Nees) Verboom (KJ920226), Chionochloa macra Zotov (NC_025230), Chloris barbata Sw. (NC_029893), Danthonia californica Bol. (NC_025232), Eragrostis tef (Zuccagni) Trotter (NC_029413), Hilaria rigida (Thurb.) Benth. ex Scribn. (NC_029896), Neyraudia reynaudiana (Kunth) Keng ex Hitchc. (NC_024262), Sporobolus michauxianus (Hitchc.) P.M.Peterson & Saarela (NC_029416), and Zoysia macrantha Desv. (NC_029418)]. One of the inverted repeat regions (IRa) was removed from each plastome to preclude double representation of these sites. Plastomes were aligned in Geneious Pro with the MAFFT v 6.814b plug-in ( Katoh et al., 2005 ) using the auto function for the algorithm and other default settings. All nucleotide positions with at least one gap were removed before phylogenomic analyses to eliminate regions of ambiguity. Models of character evolution were compared with the jmodel- Test software package v ( Guindon and Gascuel, 2003 ; Darriba et al., 2012 ) under the Akaike information criterion (AIC). The GTR + I + G was in the group of best-fit models and was used in subsequent plastome phylogenomic analyses. A maximum likelihood (ML) analysis was performed using the program RAxML v ( Stamatakis, 2014 ) at the CIPRES Science Gateway ( Miller et al., 2010 ) with the GTRGAMMA option and allowing RAxML to halt bootstrapping automatically based on the majority rule criterion: automre. The Consense function of the Phylip software package v 3.66 ( Felsenstein, 2005 ) at the CIPRES Science Gateway was used to produce a bootstrap consensus tree. A Bayesian inference (BI) analysis was performed using MrBayes in the program XSEDE v ( Ronquist et al., 2012 ) at the CIPRES Science Gateway, specifying six states and the inv.gamma model. The Markov chain Monte Carlo (MCMC) analysis was run for 2 20,000,000 generations with a designated burn-in of 25% and other parameters set at default values. Shimodaira Hasegawa (SH) test of alternative hypothesis Visual inspection of phylogenies obtained here indicated incongruence with the accepted taxonomy of the micrairoid subfamily as outlined in the introduction. To determine whether the incongruence was associated with a significant change in tree likelihood, an SH test was used to assess the log-likelihood difference between the ML tree and the alternative topology suggested by the current taxonomy ( Shimodaira and Hasegawa, 1999 ). A tree was inferred in RAxML in which micrairoid tribes were constrained to be monophyletic. In particular, this constraint forced L. meeboldii to be the immediate sister to I. distichophylla instead of H. diandra. An SH test that resampled estimated log likelihoods was conducted in RAxML with a specified GTR + I + G substitution model and 1000 bootstrap replicates. Survey of positively selected sites in rbcl Amino acid sites in rbcl, previously identified as being under positive selection due to photosynthetic pathway, were surveyed for published and newly sequenced loci. Sequences from 23 accessions of Micrairoideae were available in the GenBank database. These species (and number of accessions for each when more than one) included Coelachne japonica Hack., Eriachne aristidea F.Muell. (two), E. ciliata R.Br. (partial sequence), E. glauca R.Br. (partial sequence), E. helmsii (Domin) Hartley, E. mucronata R.Br., E. pulchella Domin, E. stipacea ; E. triodioides Domin, E. triseta Nees ex Steud., Isachne arundinacea Griseb., I. disperma Döll, I. distichophylla, I. globosa (Thunb.) Kuntze (three), I. mauritiana Kunth (two), Micraira adamsii Lazarides (two), M. spiciforma, and M. subulifolia F.Muell. Note that multiple accessions from four species allowed for assessment of intraspecific sequence variation. Four rbcl loci, extracted from the newly sequenced complete plastomes of E. armitii, H. diandra, L. meeboldii, and a C 3 reference, P. australis, were also included in the survey. One outgroup C 4 reference sequence, Bouteloua curtipendula, was also included. The rbcl sequences and GenBank accession numbers are given (Table 2 ). Nucleotide sequences were aligned as for full plastomes with manual adjustments to preserve codon boundaries and correctly position start and stop codons. Residues at previously identified positions in rbcl : 101, 258, 270, 281, and 309 were scored and compared against the expected C 3 or C 4 residue at that position ( Christin et al., 2008a ). Note that three other sites, 142, 145, and 328, which Christin et al. (2008a) also found to be under positive selection, were later determined to be under more general adaptations to variations in environmental xericity and warmness ( Christin et al., 2008a, p. 2365) and so were not examined here. Carbon isotope analysis Stable carbon isotope ratios were determined for P. australis and H. macra (Arundinoideae) as well as E. armitii, H. diandra, H. heptaneuron, and L. meeboldii. For each species, a minimum of 1 mg of dried tissue was homogenized to a fine powder in liquid nitrogen. Samples were sent to the Stable Isotope Facility, Geology and Environmental Geosciences, Northern Illinois University for isotope analysis on a DELTAplus Advantage Mass Spectrometer (ThermoFisher Scientific, Wilmington, DE, USA). The isotopic standard was Vienna Pee Dee Belemnite. Each measurement was replicated at least once.

5 290 AMERICAN JOURNAL OF BOTANY TABLE 2. Amino acids at five positively selected sites in previously banked rbcl loci from 23 accessions of Micrairoideae plus three new sequences and two outgroups. Amino acids and the specifying codons are indicated for Phragmites australis. Dots in the columns indicate the same AA as indicated for P. australis. The C 3 reference and the micrairoid species not in Eriachneae are in the upper section. Eriachne spp. and the C 4 reference are in the lower section. A dash indicates a missing AA due to a truncated sequence. Alternate AA that have a high probability in C 4 species and their specifying codons are indicated. AA with a low probability in C 4 species are shaded in gray.

6 FEBRUARY 2017, VOLUME 104 DUVALL ET AL. PLASTOME PHYLOGENOMICS OF ARUNDINOIDEAE/MICRAIROIDEAE 291 RESULTS Plastome sequencing The number of reads in the SE libraries ranged from 2.3 to 14.4 million, and these complete plastomes were assembled from 2 to 5 contigs. The number of reads in the PE library was 119 million, and the number of assembled contigs for this species was 18. Mean read coverage across the five new plastomes ranged from 42.4 to ( Table 3 ). De novo assembly of these read libraries yielded full plastomes for all five species (see Table 1 for GenBank accession numbers), which ranged from 134,735 to 137,612 bases in length ( Table 3 ). The identity of the P. australis subsp. americanus as the western haplotype of Saltonstall (2003) was confirmed by sequence comparisons with available haplotype markers (GenBank accessions AF and AY016333). Plastome phylogenomics Complete plastomes were aligned with one copy of the IR region excised. This alignment had a length of 125,995 sites. After exclusion of nucleotide sites with at least one gap introduced by the alignment, the matrix length was reduced to 96,300 unambiguously aligned bases. This aligned data matrix and the associated tree file are available at the TreeBase repository ( ). The ML analysis produced a fully resolved tree with a lnl = 269,652.2 ( Fig. 1 ). All the nodes on this tree were fully supported with ML bootstrap values (MLBV) = 100%. Bayesian inference analysis produced a fully congruent tree (not shown). All nodes in the BI tree were supported with posterior probabilities = 1.0. The monophylies of included subfamilies, Arundinoideae, Micrairoideae, Chloridoideae, and Danthonioideae, were supported at maximum values. Within Arundinoideae, M. paradoxus (Arundineae) was sister to the remaining arundinoid species (Molineae). Elytrophorus spicatus was sister to the remaining five Arundineae. Hakonechloa macra was sister to M. caerulea, which in turn was sister to the Phragmites clade (three accessions). The two native haplotypes of P. australis subsp. americanus were sister taxa, and this clade was sister to the Eurasian haplotype of P. australis. The arundinoid clade was sister to a clade of six species of Micrairoideae. Within Micrairoideae, M. spiciforma was sister to the remaining species. The two C 4 species (Eriachneae) were monophyletic. Eriachneae were sister to a clade in which Isachneae is paraphyletic with Hubbardieae. Hubbardia diandra and L. meeboldii (Hubbardieae and Isachneae, respectively) were sisters, and this clade was sister to I. distichophylla (Fig. 1 ). SH test of an alternative hypothesis When a tree search of the plastome matrix was constrained to force monophyly of Isachneae sensu Soreng et al. (2015), the tree likelihood score was significantly decreased ( P < 0.001) compared with that of the unconstrained ML topology ( Fig. 1 ). Survey of positively selected sites in rbcl Amino acids for sites that were determined to be positively selected for photosynthetic pathway by Christin et al. (2008a) are indicated in Table 2. Nucleotide identity was highly conserved across the five codons. For four of the five positively selected AA positions (101, 258, 270, and 309), all of the surveyed AA, which were V, R, L, and M, respectively, were those of high probability for C 3 grasses even though there were 11 accessions of Eriachne in 10 species in our survey. For position 281, all of the presumed C 3 species had the high probability residue (A) as did four of the C 4 species. The other four Eriachne species had the expected C 4 residue (S). The C 4 reference, Bouteloua curtipendula, had the high probability C 4 residues at three of the five sites with the exceptions of 258 and 309, which had AAs more typical of C 3 species, R and M, respectively. Carbon isotope analysis Mean carbon isotope ratios for each of six species are reported ( Table 4 ). For the three Micrairoideae previously characterized as C 3 species, the mean δ 13 C values ranged from ± to ± consistent with values previously determined for C 3 species (e.g., O Leary, 1981 ). For E. armitii, which was previously characterized as C 4, the mean δ 13 C value was ± consistent with values previously determined for C 4 species ( O Leary, 1981 ). DISCUSSION Plastome phylgenomic analyses In this study, complete plastomes were sequenced and assembled from five species in the arundinoid/ micrairoid clade using de novo methods. Our analyses had uniformly high support values across all nodes in the plastome phylogenomic trees. In parallel studies of other groups of grasses, similarly high levels of support were obtained ( Zhang et al., 2011 ; Wu and Ge, 2012 ; Jones et al., 2014 ; Cotton et al., 2015 ; Saarela et al., 2015 ; Wysocki et al., 2015 ; Burke et al., 2016 ; Duvall et al., 2016 ). Arundinoideae The arundinoid clade in the plastome phylogenomic analysis was sister to a clade of Micrairoideae, consistent with previous analyses (e.g., Cotton et al., 2015 ; Duvall et al., 2016 ; GPWG II, 2012 ). The seven arundinoid taxa formed a monophyletic group with maximum support values. While this topology is generally consistent with other studies that redefined a narrowly circumscribed Arundinoideae, the branch subtending this subfamily in our ML tree is the shortest internal backbone branch in the tree. In part, this short branch explains the difficulty in resolving this subfamily in single and multigene studies ( Clark et al., 1995 ; GPWG II, 2012 ). The length of this branch, , is considerably shorter than, for example, that subtending Micrairoideae, which is and is one of the shortest internal branches deep in the tree TABLE 3. Assembly statistics, lengths (bases) of plastome regions, and % AT for newly sequenced plastomes. Taxon No. of reads No. of scaffolded contigs Mean coverage Total length LSC SSC IR %AT Eriachne armitii 14,398, ,859 79,942 12,593 21, Hubbardia diandra 9,415, ,485 80,702 12,643 21, Limnopoa meeboldii 6,193, ,735 79,893 12,684 21, Molinia caerulea 119,108, ,574 82,303 12,741 21, Phragmites australis 2,329, ,612 82,381 12,693 21, Notes: IR = inverted repeat region; LSC = large single copy region; SSC = small single copy region.

7 292 AMERICAN JOURNAL OF BOTANY FIGURE 1 Maximum likelihood phylogram inferred from complete plastomes. Branch lengths are proportional to the number of substitutions per site along the branch. Arundinoideae and Micrairoideae are indicated, and micrairoid tribes are specified. Haplotype designations E, H, and M are indicated for the three accessions of Phragmites australis. Two independent origins of C 4 photosynthesis are noted. Note that a Bayesian inference analysis produced an identical topology. All nodes were supported with MLBV = 100% and with posterior probabilities of 1.0. Note that the strongly supported sister relationship between Hubbardia diandra and L. meeboldii places Hubbardieae in a paraphyletic arrangement with Isachneae (solid gray shading). (Fig. 1 ). This very short branch is suggestive of either a decrease in nucleotide substitution rates relative to diversification rates along the stem lineage of Arundinoideae or decreased extinctions in this subfamily compared with other major PACMAD lineages. Monachather paradoxus was sister to a monophyletic Molinieae consistent with the tribal classification of Soreng et al. (2015) for the subfamily. The two native North American haplotypes of P. australis subsp. americanus, E and H, were sister taxa, which is consistent with their historical biogeography. In separate, whole-plastome matrices aligned using the same methods described above, these TABLE 4. Mean δ 13 C measurements for arundinoid and micrairoid species. Species Mean δ 13 C ratio ( ±SE) ( ) Hubbardia diandra ± H. heptaneuron Bor ± Limnopoa meeboldii ± Phragmites australis subsp. americanus (C 3 reference) ± Hakonechloa macra Makino (C 3 reference) ± Eriachne armitii ± two plastomes differed by 308 nucleotide sites, which were more than the 243 nucleotide differences that were observed between the complete plastomes of the two species of Eriachne determined in this study. An even greater pairwise difference was observed between the North American and Eurasian haplotypes, which had a mean value of 535. Note that this level of genetic differentiation is not correlated with structural or morphological features as the haplotypes are not readily distinguished in field surveys ( Saltonstall et al., 2004 ). Acknowledgment of this genetic diversity is reflected in the recognition of P. australis subsp. americanus although other haplotypes are not similarly recognized. Two plastome loci have been used to typify haplotypes, the trnt-trnl and rbcl-psai intergenic spacers (IGS). Having sequences of complete plastome loci allows for the identification of additional loci useful in haplotype characterizations. For example, in an alignment of complete plastomes of the E, H, and M haplotypes of P. australis, we observed variation in the trnf-ndhj IGS, which is downstream of the trnt-trnl IGS. In this region, the M haplotype has two upstream insertions, a 27-base tandem repeat insert and a 4-base nontandem repeat. Neither of these are found in the native North American haplotypes. Also in the trnf-ndhj IGS, our newly sequenced H haplotype has a downstream

8 FEBRUARY 2017, VOLUME 104 DUVALL ET AL. PLASTOME PHYLOGENOMICS OF ARUNDINOIDEAE/MICRAIROIDEAE base tandem repeat insertion, again not found in the other two. So the E, H, and M haplotypes are uniquely defined by this single trnf-ndhj IGS. These and other distinguishing plastome loci could potentially lead to more efficient discrimination among the lineages of this species. Micrairoideae Within Micrairoideae, the one species of Micraira was sister to the remaining taxa, which is consistent with the results of GPWG II (2012), although with somewhat different sampling. The remaining micrairoids divided into two clades, one with two accessions of Eriachne and the other a paraphyletic grouping of Isachneae with Hubbardieae. In this latter subclade, the sister relationship between Hubbardia and Limnopoa was strongly supported (MLBV = 100% and PP = 1.0). This phylogeny is inconsistent with the classification in which Limnopoa is included in the same tribe as Isachne to the exclusion of Hubbardia, which is segregated into its own tribe ( Soreng et al., 2015 ). Moreover, the SH test of the alternate phylogenetic arrangement, in which Limnopoa was constrained to be the immediate sister to Isachne, indicated that the associated likelihood was significantly decreased ( P < 0.001) compared with that of the unconstrained ML topology. So in the context of our sampling, the existing tribal classification of Soreng et al. (2015) contradicts plastome phylogenomic data. Our results suggest one of three alternate taxonomic schemes for this clade. Under the first possibility, Hubbardieae would be combined with the currently recognized Isachneae into a single tribe, in agreement with a comment by Kellogg (2015) regarding the unnecessary division of Micrairoideae into tribes. The second possibility would be to combine Limnopoa with Hubbardia in the same tribe, supported by the results in our plastome phylogenomic analyses. A third possibility would be to segregate L. meeboldii into a new and separate tribe, thereby recognizing its unique habit in the subfamily. Further sampling of Coelachne, Heteranthoecia, Sphaerocaryum, and other species of Isachne in a plastome phylogenomic study would be useful to test these alternate classifications. Moreover, in our analysis of the Micrairoideae, we have included only the type species for Limnopoa, and it is necessary to sample the types of all other genera to address any questions regarding classification. Evaluation of photosynthetic pathway One of the 22 to 24 independent origins of C 4 photosynthesis in grasses occurs within Micrairoideae ( GPWG II, 2012 ). Because of this fact and that the photosynthetic pathway is a systematic character used to define Eriachneae, we re-evaluated the total evidence for this character among selected species. As noted above, Micrairoideae exhibit unusual combinations of characters for photosynthetic pathway in several respects. Most micrairoid genera have radiate chlorenchyma in the leaf that is characteristic of C 4 Kranz anatomy, but at the same time have more widely spaced veins, which is more typical of C 3 species ( Kellogg, 2015 ). Yet all of these taxa, except the two genera of Eriachneae are reputedly C 3. Eriachne and Pheidochloa have a combination of the NADP-ME decarboxylating enzyme, but an anatomical double sheath with a mestome sheath interior to the bundle sheath (technically an XyMS+ anatomy) ( Prendergast et al., 1987 ). The anatomy, but not the decarboxylating enzyme, is typically associated with plants of arid habitats ( Karbaschi, 2015 ), and no other NADP-ME grasses have been observed to have this type of anatomy. We observed a high degree of conservation across all micrairoid rbcl sequences at previously identified positively selected sites irrespective of the photosynthetic pathway. This observation contradicts expectations of consistent and contrasting differences between C 3 species. The study of Christin et al. (2008b) sampled 150 representative grasses including those in 17 separate C 4 lineages. Their study indicates high probabilities of specific adaptive amino acid changes among C 4 plants. However, the only variation among Micrairoideae we observed that did conform to expectations was in position 281 in four species. Five accessions of Eriachne showed no evidence of the C 4 AA signatures. Whereas leaf anatomy, biochemistry, and positively selected sites are generally reliable proxies for photosynthetic pathway, the decisive evidence is to be found in measurements of carbon isotope ratios. Here we present such measurements for four micrairoid species for the first time. The mean δ 13 C values for Hubbardia, Isachne, and Limnopoa were in the expected range for C 3 species and that for Eriachne was in the expected range for a C 4 species. The δ 13 C measurements of the C 3 references were also as expected. These results suggest that the ambiguities observed in the proxy evidence bearing on the photosynthetic pathway in micrairoid grasses is due to unidentified factors that are possibly not directly related to photosynthetic pathway. Eriachne is distributed in tropical China, Indomalaya, Australia, and New Guinea. Diverse species are variously found in arid, semiarid, and seasonally wet habitats, often in rocky or sandy substrates. Pheidochloa spp. are found in damp sandy heaths of Australia and New Guinea ( Watson and Dallwitz, 1992 ). The overall diversity of habitats does not present clear evidence of common selective pressures that might explain unexpected combinations of anatomical, physiological, and molecular characters in this or other micrairoid genera clearly bearing on either substrate specificity or catalytic efficiencies ( myspecies.info accessed 5 May 2016). Further study of the other C 4 specialists in the subfamily, Pheidochloa spp., would better show the full diversity of this unusual character suite in Micrairoideae. CONCLUSIONS The ready accessibility of high-throughput sequencing technology suggests that whole plastomes might be useful for haplotype discriminations among geographically segregated lineages of plants such as P. australis, A. donax, and other grasses. The high support found in plastome phylogenomic analyses suggests approaches for studies to resolve further questions of PACMAD evolution, such as the deep relationships in this group. The incongruence between the recent classification by Soreng et al. (2015) for tribal classifications in Micrairoideae and our plastome phylogenomic analyses suggests the need for reclassification, but only after further sampling among these sometimes rare and difficult to obtain taxa. Parallel nuclear phylogenomic analyses would also contribute information relevant to these taxonomic questions. The newly presented δ 13 C measurements for Micrairoideae suggest that further study is needed to address questions of why anatomical, physiological, and molecular characters are imperfectly correlated with photosynthetic pathways and what other selection pressures might be operating on these diverse grasses. ACKNOWLEDGEMENTS We thank L. Clark. J. Saarela, and M. Nazaire for material. We also thank M. Nazaire for the correct determination of one specimen

9 294 AMERICAN JOURNAL OF BOTANY that was misidentified as Arundo donax ( GBIF ). For technical assistance, we thank K. Murrell, L. Attigala, and A. Buczynska. We appreciate the helpful comments we received from Paul Peterson and one anonymous reviewer. This work was supported in part by the Plant Molecular and Bioinformatics Center and the Department of Biological Sciences at Northern Illinois University. Support is also acknowledged from grants from the National Science Foundation to M.R.D. (DEB and DEB ). Any opinions, findings, and conclusions or recommendations expressed in this material are those of the authors and do not necessarily reflect the views of the National Science Foundation. LITERATURE CITED Bankevich, A., S. Nurk, D. Antipov, A. A. Gurevich, M. Dvorkin, A. S. Kulikov, V. M. Lesin, et al SPAdes: A new genome assembly algorithm and its applications to single-cell sequencing. Journal of Computational Biology 19 : Barker, N. P., H. P. Linder, and E. H. Harley Polyphyly of Arundinoideae (Poaceae): Evidence from rbcl sequence data. Systematic Botany 20 : Bouchenak-Khelladi, Y., N. Salamin, V. Savolainen, F. Forest, M. van der Bank, M. W. Chase, and T. R. Hodkinson Large multi-gene phylogenetic trees of the grasses (Poaceae): Progress towards complete tribal and generic level sampling. Molecular Phylogenetics and Evolution 47 : Burke, S. V., C. P. Grennan, and M. R. Duvall Plastome sequences of two New World bamboos Arundinaria gigantea and Cryptochloa strictiflora (Poaceae) extend phylogenomic understanding of Bambusoideae. American Journal of Botany 99 : Burke, S. V., W. P. Wysocki, F. O. Zuloaga, J. M. Craine, J. C. Pires, P. P. Edger, D. Mayfield-Jones, et al Evolutionary relationships in panicoid grasses based on plastome phylogenomics (Panicoideae; Poaceae). BMC Plant Biology 16 : 140. Chandore, A. N., K. V. C. Gosavi, S. M. Gund, R. V. Gurav, and S. R. Yadav Hubbardia diandra, a new species of Poaceae from the northern Western Ghats with a note on tribe Hubbardieae. Kew Bulletin 67 : Christin, P. A., G. Besnard, E. Samaritani, M. R. Duvall, T. R. Hodkinson, V. Savolainen, and N. Salamin. 2008a. Oligocene CO2 decline promoted C 4 photosynthesis in grasses. Current Biology 18 : Christin, P. A., N. Salamin, A. M. Muasya, E. H. Roalson, F. Russier, and G. Besnard. 2008b. Evolutionary switch and genetic convergence on rbcl following the evolution of C 4 photosynthesis. Molecular Biology and Evolution 25 : Clark, L. G., W. Zhang, and J. F. Wendel A phylogeny of the grass family (Poaceae) based on ndhf sequence data. Systematic Botany 20 : Clayton, W. D. and S. A. Renvoize Genera graminum: Grasses of the world. Kew Bulletin Additional Series, vol. 13. Her Majesty s Stationary, London, UK. Cotton, J. L., W. P. Wysocki, L. G. Clark, S. A. Kelchner, J. C. Pires, P. P. Edger, D. Mayfield-Jones, and M. R. Duvall Resolving deep relationships of PACMAD grasses: A phylogenomic approach. BMC Plant Biology 15 : 178. Darriba, D., G. L. Taboada, R. Doallo, and D. Posada jmodeltest 2: More models, new heuristics and parallel computing. Nature Methods 9 : 772. Dengler, N. G., R. E. Dengler, P. M. Donnelly, and P. W. Hattersley Quantitative leaf anatomy of C 3 grasses (Poaceae): Bundle sheath and mesophyll surface area relationships. Annals of Botany 73 : Devi, M., and D. Bhattacharyya Rediscovery of the red-listed grass Isachne dimyloides (Poaceae: Micrairoideae) 130 years after type collection. Phytotaxa 221 : Duvall, M. R., J. I. Davis, L. G. Clark, J. D. Noll, D. H. Goldman, and J. G. Sánchez-Ken Phylogeny of the grasses (Poaceae) revisited. Aliso 23 : Duvall, M. R., A. E. Fisher, J. T. Columbus, A. L. Ingram, W. P. Wysocki, S. V. Burke, L. G. Clark, and S. A. Kelchner Phylogenomics and plastome evolution of the chloridoid grasses (Chloridoideae: Poaceae). International Journal of Plant Sciences 177 : Felsenstein, J PHYLIP (phylogeny inference package). Distributed by the author. Department of Genome Sciences, University of Washington, Seattle, Washington, USA. GPWG I [Grass Phylogeny Working Group I] Phylogeny and subfamilial classification of the grasses (Poaceae). Annals of the Missouri Botanical Garden 88 : GPWG II [Grass Phylogeny Working Group II] New grass phylogeny resolves deep evolutionary relationships and discovers C 4 origins. New Phytologist 193 : Guindon, S., and O. Gascuel A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. Systematic Biology 52 : Hardion, L., R. Verlaque, A. Baumel, M. Juin, and B. Vila Revised systematics of Mediterranean Arundo (Poaceae) based on AFLP fingerprints and morphology. Taxon 61 : Hattersley, P. W., and L. Watson C4 grasses: An anatomical criterion for distinguishing between NADP-malic enzyme species and PCK or NADmalic enzyme species. Australian Journal of Botany 24 : Ingram, A. L., P. A. Christin, and C. P. Osborne Molecular phylogenies disprove a hypothesized C4 reversion in Eragrostis walteri (Poaceae). Annals of Botany 107 : Jones, S. S., S. V. Burke, and M. R. Duvall Phylogenomics, molecular evolution, and estimated ages of lineages from the deep phylogeny of Poaceae. Plant Systematics and Evolution 300 : Karbaschi, M. R Structural, physiological and molecular characterisation of the Australian native resurrection grass Tripogon loliiformis (F.Meull) C.E. Hubb during dehydration and rehydration. Ph.D. dissertation, Queensland University of Technology, Brisbane, Australia. Katoh, K., K. I. Kuma, H. Toh, and T. Miyata MAFFT version 5: Improvement in accuracy of multiple sequence alignment. Nucleic Acids Research 33 : Kellogg, E. A Flowering plants. Monocots, Poaceae. In K. Kubitzki [ed.], The families and genera of vascular plants, vol. XIII. Springer International Publishing Switzerland, Cham, Switzerland. Kumar, R. B., K. Anitha, A. Watve, S. Mani, S. Rehel, and W. Arisdason The status and distribution of aquatic plants of the Western Ghats. In S. Molur, K. G. Smith, B. A. Daniel, and W. R. T. Darwall [compilers], The status and distribution of freshwater biodiversity in the Western Ghats, India, International Union for Conservation of Nature [IUCN], Cambridge, UK and Gland, Switzerland and Zoo Outreach Organization, Coimbatore, India. Lambert, A. M., T. L. Dudley, and K. Saltonstall Ecology and impacts of the large-statured invasive grasses Arundo donax and Phragmites australis in North America. Invasive Plant Science and Management 3 : Linder, H. P., G. A. Verboom, and N. P. Barker Phylogeny and evolution in the Crinipes group of grasses ( Arundinoideae : Poaceae ). Kew Bulletin 52 : Ma, P. F., Y. X. Zhang, C. X. Zeng, Z. H. Guo, and D. Z. Li Chloroplast phylogenomic analyses resolve deep-level relationships of an intractable bamboo tribe Arundinarieae (Poaceae). Systematic Biology 63 : Miller, M., W. Pfeiffer, and T. Schwartz Creating the CIPRES Science Gateway for inference of large phylogenetic trees. Gateway Computing Environments Workshop (GCE), New Orleans, Louisiana,, USA, 1 8 IEEE. O Leary, M. H Carbon isotope fractionation in plants. Phytochemistry 20 : Prendergast, H. D. V., P. W. Hattersley, and N. E. Stone New structural/ biochemical associations in leaf blades of C 4 grasses (Poaceae). Functional Plant Biology 14 : Ronquist, F., M. Teslenko, P. van der Mark, D. L. Ayres, A. Darling, S. Höhna, B. Larget, et al MrBayes 3.2: Efficient Bayesian phylogenetic inference and model choice across a large model space. Systematic Biology 61 : Saarela, J. M., W. P. Wysocki, C. F. Barrett, R. J. Soreng, J. I. Davis, L. G. Clark, S. A. Kelchner, et al Plastid phylogenomics of the cool-season grass subfamily: Clarification of relationships among early-diverging tribes. AoB Plants 7 : plv046. Saltonstall, K Cryptic invasion by a non-native genotype of the common reed, Phragmites australis, into North America. Proceedings of the National Academy of Sciences, USA 99 :