PHYLOGENY AND CLASSIFICATION OF CAREX SECTION OVALES (CYPERACEAE)

Andrew L. Hipp,1,* Anton A. Reznicek,y Paul E. Rothrock,z and Jaime A. Weber§

*Department of Botany, 430 Lincoln Drive, University of Wisconsin, Madison, Wisconsin 53706, U.S.A.;yUniversity of Michigan Herbarium, 3600 Varsity Drive, Ann Arbor, Michigan 48108, U.S.A.;zRandall Environmental Studies Center, Taylor University, Upland, Indiana 46989, U.S.A.;

and §Morton Arboretum, 4100 Illinois Route 53, Lisle, Illinois 60532, U.S.A.

Section Ovales is the most species-rich section of the sedge genus Carex in the New World. Phylogeneticanalyses of molecular data recover a predominantly New World clade as sister to a solitary east Asian species,C. maackii. Nuclear ribosomal DNA are congruent in the placement of all taxa within the section, with asolitary exception: incongruence between ITS and ETS data in the placement of C. bonplandii and C.roraimensis suggests a hybrid origin for this lineage. Biogeography correlates strongly with phylogeny in thesection, but there have been at least two instances of long-range dispersal, one from an eastern North Americanclade to western North America and one from the New World to Eurasia. Morphological characters studiedare all homoplastic. Developing a comprehensive infrasectional classification with a phylogenetic basis wouldbe complicated by the fact that most of the novel morphological characters in the section have evolved withinrelatively small, independent clades.

Keywords: Carex subgenus Vignea, Carex section Ovales, Carex section Stellulatae, hybrid speciation,ancestral character state reconstruction, nuclear ribosomal DNA.

Introduction

The sedge family (Cyperaceae) numbers ca. 5000 speciesworldwide, making it the third largest family of monocots(Goetghebeur 1998). The genus Carex L. of tribe Cariceaecomprises roughly 40% of the family by species, making itone of the largest genera of angiosperms (Reznicek 1990;Mabberley 1997). Carex species are ecologically importantmembers of floodplain forests, dry prairies, alpine meadows,peat lands, swamp forests, sedge meadows, and a wide rangeof other communities (Reznicek 1990). Molecular work inCyperaceae tribe Cariceae has mostly focused on generic cir-cumscription or relationships among the subgenera and sec-tions that make up the genus Carex (Starr et al. 1999, 2003,2004; Yen and Olmstead 2000; Roalson et al. 2001;Hendrichs et al. 2004a, 2004b). There has been little workon fine-scale patterns of morphological diversification withinthe genus (though see Roalson and Friar 2004).Section Ovales Kunth is the largest section in Carex subge-

nus Vignea (P. Beauv. ex T. Lestib.) Peterm., containing 72North American species (Mastrogiuseppe et al. 2002), 15 ad-ditional described species endemic to South or Central Amer-ica, and three species endemic to Europe and Asia (Reznicek1993), for a total of 90 species worldwide. The section re-flects much of the ecological breadth of the entire genus.Morphologically, however, the section is extremely cohesive,

marked by vegetative shoots with nodes (‘‘vegetative culms’’sensu Reznicek and Catling 1986), gynecandrous spikes, andperigynia bearing marginal, epidermal ‘‘wings’’ (Reznicek1993). This combination of characters makes the section eas-ily recognizable, though distinguishing the species is often dif-ficult. Monophyly of section Ovales has been demonstratedusing sequence data from the internal transcribed spacer re-gions (ITS1 and ITS2) and 5.8S nuclear ribosomal gene(Hipp, forthcoming).Kenneth K. Mackenzie (1931–1935) divided section

Ovales into 11 informal species groups based on charactersof the perigynia, pistillate scales, leaf sheaths, and vegetativeculms (fig. 1). Although some of the species groups havestrong ecological, geographical, and morphological homoge-neity (e.g., the ‘‘Tribuloideae’’ and ‘‘Alatae’’), most do not,and the groups are not generally viewed as natural (Reznicek1993). The groups were never described formally. Conse-quently, Mackenzie’s species group names are in quotationmarks throughout this article.In this study, we combine ITS data with sequences from

the 59 end of the external transcribed spacer (ETS) region ofnrDNA to evaluate biogeographic shifts and patterns of mor-phological evolution associated with the diversification ofCarex section Ovales. We circumscribe major lineages withinthe section and identify the sister group to the North Ameri-can clade that makes up the majority of the section. We usetests of topological incongruence to evaluate Mackenzie’s in-frasectional classification and the strength of support for al-ternative placements of key taxa in the ETS and ITS datapartitions, providing preliminary phylogenetic and chromo-somal evidence for a possible allopolyploid origin of the

1 Author for correspondence; current address: Morton Arboretum,

4100 Illinois Route 53, Lisle, Illinois 60532-1293, U.S.A.; e-mail

ahipp@mortonarb.org.

Manuscript received September 2005; revised manuscript received April 2006.

1029

Int. J. Plant Sci. 167(5):1029–1048. 2006.

� 2006 by The University of Chicago. All rights reserved.

1058-5893/2006/16705-0012$15.00

Latin American C. bonplandii complex. Finally, we use maxi-mum likelihood to reconstruct the evolution of morphologi-cal characters and evaluate the phylogenetic consistency ofMackenzie’s infrasectional classification.

Material and Methods

Taxon Sampling

Sequences were obtained from 121 individuals representing18 of ca. 26 recognized sections of subgenus Vignea (whichincludes sections Ovales and Cyperoideae) and three sectionsof subgenus Carex (appendix table A1). Sampling includes75 species and two varieties of Carex section Ovales, repre-senting all continents, in addition to four individuals of inde-terminate identity: C. cf. microptera, C. cf. lagunensis, C. cf.brevior from Mexico, and a plant from Arkansas with affini-ties to C. molesta that is referred to in this article as ‘‘BuffaloRiver.’’ Sampling outside of section Ovales includes threespecies of subgenus Vignea known either to reproduce exten-sively from vegetative shoots (C. chordorrhiza and C. pseu-

docuraica) or to produce true vegetative culms (C.sartwellii), three non-Ovales species with winged perigynia(C. planata, C. brizoides, and C. siccata), and both speciesfrom section Cyperoideae (C. bohemica and C. sychnocephala),which has been variously treated as a separate section or aspart of section Ovales. Three outgroups are included fromsubgenus Carex based on their placement in a previous study(Roalson et al. 2001).We follow the taxonomy presented in Flora of North

America North of Mexico (FNA; Ball and Reznicek 2002),with two exceptions: (1) we provisionally treat C. tenerifor-mis Mack., which is treated as a synonym of C. subfuscaBoott in FNA, as taxonomically distinct based on sequenceresults and field observations; and (2) we refer to section Cy-peroideae G. Don as a species group within section Ovales,following Mackenzie (1931–1935) and previous molecularstudy in the section (Hipp, forthcoming). The placement of35 taxa was confirmed using sequences from DNA extractedfrom an additional individual of each species (table A1).Most ITS sequences were generated for a previous study(Hipp, forthcoming); the remaining ITS and all ETS se-quences were generated for this study.

Fig. 1 Overview of K. K. Mackenzie’s (1940) infrasectional classification of Carex section Ovales, including the six major key characters thatMackenzie used to define the species groups. Biogeographic regions are approximate, and there is overlap in the distribution for several of the

species groups indicated as occurring in either western or eastern North America. Perigynium beak cross-sectional shape and perigynium body

shape were excluded from analyses presented in this article because herbarium study suggests that they are continuous. Not all key characters are

coded for analyses presented in this article as Mackenzie reported them (cf. fig. 5) because of discrepancies between Mackenzie’s keys and morerecent studies (Mastrogiuseppe et al. 2002) as well as our own herbarium observations. Per. bk ¼ perigynium beak cross-section; Per.body ¼ perigynium body shape in outline; Pist. scale ¼ pistillate scale; Bract ¼ lowest bract of the entire inflorescence; Veg. culm ¼ vegetative

culm; Lf sheath ¼ inner face of the leaf sheath. Illustrations by H. C. Creutzberg, reprinted by permission from the New York Botanical Garden,Bronx, NY (Mackenzie 1940).

1030 INTERNATIONAL JOURNAL OF PLANT SCIENCES

DNA Extraction, PCR, and Sequencing

DNA was extracted from live, silica-dried, frozen, or her-barium tissue of single individuals using a modified 6XCTAB method (Doyle and Doyle 1987) and DNeasy kits(QIAGEN, Valencia, CA). The ITS region was amplified us-ing the primers ITS-I (Urbatsch et al. 2000) and ITS4 (White

et al. 1990), and the ETS region was amplified using primers

ETS-1F and 18S-R (Starr et al. 2003). PCR was conducted

on 50-mL reactions containing 5 mL MgCl2 at 25 mM, 5 mL

10X MgCl2-free Taq buffer, 2 mg BSA, 1.5 mL DMSO, 0.5 mL

of each primer at 20 mM, 0.25 mL Taq DNA polymerase

(1.25 units), and 1 mL genomic DNA template. PCR conditions

Table 1

Uncorrected Number of Nucleotide Differences (Minimum and Maximum)between nrDNA Clones for Seven Taxa

ETS ITS þ 5.8S

Taxon

No.

clones

No.

unique

sequences

Min.

diff.

Max.

diff.

No.

clones

No.

unique

sequences

Min.

diff.

Max.

diff.

Carex bonplandii complex:C. bonplandii 10 8 1 5 9 8 2 15

C. roraimensis 10 4 1 4 9 6 1 7

Section Ovales:C. straminea 10 4 1 3 9 2 2 2C. vexans 9 4 1 6 10 6 1 7

Section Stellulatae:C. echinata 10 4 1 6 9 6 1 4C. exilis 9 7 1 5 10 7 1 6

C. interior 10 5 1 6 10 5 1 4

All interspecific 68 36 1 39 66 40 3 40

Note. Individuals cloned are the same individuals sequenced for phylogenetic study (table A1). Nine

to 10 clones were sequenced per individual. The number of unique sequences recovered for each taxon

is closely correlated between ETS and ITSþ 5:8S. Carex bonplandii displays a great deal of interclonal

divergence in the ITSþ 5:8S region; sequencing of additional clones would be needed to determinewhether sequence types exhibiting this same degree of divergence are present at low levels in all taxa or

only in C. bonplandii.

Fig. 2 Unrooted neighbor-joining (NJ) trees of unique clone sequences for seven taxa based on two nrDNA regions: ITSþ 5:8S and ETS. NJwas conducted on GTR distances in PAUP* 4.0b10 (Swofford 2002), with all other settings at default values. NJ bootstraps (1000 replicates) are

shown above the branches.

1031HIPP ET AL.—PHYLOGENY OF CAREX SECTION OVALES

for most runs were initial denaturation at 94�C for 5 min; 30cycles of DNA denaturation at 94�C for 30 s, primer anneal-ing at 48�C for 1 min, and extension at 72�C for 1.5 min;and a 7-min extension at 72�C. Double-stranded PCR pro-ducts were quantified on 0.8% EtBr-stained agarose andcleaned using Millipore spin columns, QIAquick cleanup kits(QIAGEN), or magnetic beads.Homogenization of nrDNA regions was evaluated by clon-

ing three individuals from section Stellulatae, two from sec-tion Ovales, and two members of the C. bonplandii complex(C. roraimensis and two individuals of C. bonplandii). PCRwas conducted for these individuals both with and withoutDMSO, which has been shown to have an effect on the pro-portion of functional nrDNA sequences recovered in PCR(Buckler et al. 1997). PCR bands were excised and cleaned us-ing the centrifugation protocol of the Wizard SV Gel and PCRClean-Up System (Promega, Madison, WI). Each cleaned PCRwas cloned using the pGEM-T Easy Vector System (Promega).The manufacturer’s protocol was followed except that someligation reactions were cut in half, and the correspondingtransformation reactions used 4 mL ligation reaction, 25 mLcompetent cells, and 500 mL SOC medium. Five clones perPCR per individual per nrDNA region were selected and puri-fied using the centrifugation protocol of the Wizard Plus SVMinipreps DNA Purification System (Promega).Cleaned PCR was cycle sequenced in half-reactions

(10 mL) using BigDye reaction kits and the primers employedin PCR. Additional ITS sequences from internal primersITS3B (Baum et al. 1998) and ITS2 (White et al. 1990) wereobtained for taxa in which ITS-I and ITS4 sequences did notprovide double coverage of a substantial portion of the ITSregions. Plasmid DNA purified from clones was cycle se-quenced using primers complementary to the T7 and SP6promoters. Cycle sequencing products were precipitated in75% ethanol or cleaned with magnetic beads and sequencedon ABI 377 or ABI 3100 automated sequencers at the Uni-versity of Wisconsin–Madison Biotechnology Center’s DNAFacility. Clone sequences were analyzed on an ABI 3730 inthe Field Museum’s Pritzker Lab (Chicago, IL).

Phylogenetic Analysis

Sequences were edited and assembled in Sequencher 3.0(GeneCodes, 1991–1995) and aligned manually in BioEdit5.0.9 (Hall 1999). Sequences are deposited in GenBank (tableA1), as are all unique clone sequences (ITS: DQ461091–DQ461130; ETS: DQ461055–DQ461090). Sequence align-ments were unambiguous except for a few single base pairindels; alternate alignments at these points made no differencein the topology or support levels of final trees. Gaps werecoded using the simple indel coding method (Simmons and

Ochoterena 2000). Indels were analyzed alone to determinewhether they recovered topologies that were congruent withsequence data before combined analysis. Ragged ends (74aligned nucleotides, including seven potentially informativesites) were included in parsimony analyses, where they had apositive effect on resolution of the strict consensus tree, butwere excluded from likelihood and Bayesian analyses. Muta-tional saturation for each sequence partition was estimated byplotting pairwise transition/transversion (Ti/Tv) ratios againstJukes-Cantor distances. A flat or inverse relationship betweenTi/Tv ratio and genetic distance was interpreted as evidence ofmutational saturation (Halanych et al. 1999; Lee 2000).Maximum parsimony trees were recovered in heuristic

searches on equally weighted characters in PAUP* 4.0b10(Swofford 2002). Searches were conducted using 1000 rep-licates of random-sequence addition to detect multiple is-lands of most parsimonious (MP) trees (Maddison 1991),with 100 MP trees saved for each replicate (chuckscore ¼1, nchuck ¼ 100, MULTREES ¼ yes) and tree-bisection-reconnection (TBR) branch swapping. The strict consensuswas used as a reverse constraint in a second heuristic searchunder the same parameters, saving only trees not compatiblewith the consensus. This method is used to ensure that noshorter trees exist and that the strict consensus recovered inthis study represents the set of all most parsimonious trees(Catalan et al. 1997). Branch support was estimated by non-parametric bootstrapping, using 1000 heuristic bootstrapreplicates of 1000 random-sequence addition replicates each,with 10 trees held at each step during stepwise addition(chuckscore ¼ 1, nchuck ¼ 10, MULTREES ¼ yes) and TBRbranch swapping.Data were also analyzed using Metropolis-coupled Markov

chain Monte Carlo in MrBayes 3.0b4 (Huelsenbeck andRonquist 2001). The three nucleotide partitions (ITS, 5.8S,and ETS) were modeled separately, using models selectedbased on Akaike information criterion (AIC) values calculatedin Modeltest 3.6 (Posada and Crandall 1998). Indels for ITSand ETS were modeled separately as two-state characters(data type ¼ standard), with a correction for the fact thatonly variable characters were scored (coding ¼ variable). Forcombined data analyses, all parameters except for branchlengths and topology were unlinked across data partitions.Three independent runs of four linked chains were each runfor 5 million generations, using the default priors and tem-perature parameter. Following inspection of run results, treesfrom the initial 1 million generations were eliminated fromanalysis to ensure that topologies and support levels repre-sent an unbiased estimate of the posterior probability distri-bution. Convergence was assessed by comparing post-burn-intree likelihood and consensus topology for the independent

Fig. 3 Majority-rule consensus trees resulting from separate Markov chain Monte Carlo analyses of each of the nrDNA data sets:

ITSþ 5:8Sþ ITS indels and ETSþ ETS indels. Topology, branch lengths, and posterior probabilities were estimated from runs of 5 milliongenerations, with the first 1 million generations discarded to ensure that all parameters were estimated while the chain was at stationarity.

Numbers above branches are posterior probabilities estimated from 40,000 trees sampled from the post-burn-in trees. Outgroups are excluded

from the figure to save space; their resolution provides no new information on relationships in the genus. Position of the Carex bonplandiicomplex is indicated by the unlabeled boxes.

1033HIPP ET AL.—PHYLOGENY OF CAREX SECTION OVALES

Fig. 4 Majority-rule consensus of Bayesian trees resulting from Markov chain Monte Carlo analysis of all data. Topology, branch lengths, and

posterior probabilities were estimated from a run of 5 million generations, with the first 1 million generations discarded. Substitution model

parameters were allowed to vary independently across data partitions. Numbers above branches are posterior probabilities estimated from 40,000trees sampled from the post-burn-in trees (before the slash) and maximum parsimony (MP) bootstraps (after the slash). The strict consensus of MP

runs, but trees and analyses presented in this article are re-sults of a single run.

Tests of Data Partition Congruence and TopologicalTests of Phylogenetic Hypotheses

The ITS and ETS data partitions were evaluated for congru-ence using the incongruence length difference (ILD) test (Far-ris et al. 1994, 1995) in PAUP* 4.0b10. A preliminary set ofILD tests was conducted under a range of relative weightingconditions to estimate the point of balance between data par-titions in number of informative characters and overallstrength of evidence (Dowton and Austin 2002; Hipp et al.2004). Indels, 5.8S sequences, and uninformative sites wereexcluded from ILD analyses to limit the known effects of dif-ferences in substitution parameters and data decisiveness onILD P values (Barker and Lutzoni 2002; Darlu and Lecointre2002). ILD tests were conducted on 499 randomly partitioneddata sets from which indels, 5.8S, and all uninformative siteshad been excluded, with heuristic searches conducted using10 random-sequence addition replicates, with 10 trees held ateach replicate (chuckscore ¼ 1, nchuck ¼ 10). The ILD testwas then conducted again for the relative weighting of thetwo partitions that produced the lowest P value. This test wasconducted with 25 random-sequence addition replicates, with20 trees held at each replicate.Alternative phylogenetic hypotheses were evaluated

using the Shimodaira-Hasegawa (SH) test and Bonferroni-corrected, one-tailed Kishino-Hasegawa (KH) test in PAUP*.Null distribution of the test statistic was simulated using10,000 bootstrap replicates, likelihoods approximated usingthe resampling estimated log likelihood (RELL) method ofKishino et al. (1990). Because of the relatively large numberof searches required for the topology tests, likelihood searcheswere conducted using five random-sequence addition repli-cates, with a limit of 100,000 rearrangements per replicate. Inmore than 50% of replicates, the rearrangement limit was notreached, and the optimal tree was recovered in at least tworeplicates in most searches.

Analysis of Morphological and Biogeographic Data

Ancestral character states of morphological characterswere reconstructed using maximum likelihood. Five charac-ters used by Mackenzie were assessed in the herbarium bymeasuring individuals from each of 19 taxa that represent abroad morphological and phylogenetic range of the section(table A1). Mean measurements from three individuals pertaxon were plotted along with the maximum and minimum,and discontinuities were assessed visually (data availablefrom A. L. Hipp). Of these, two characters were found to be

continuous and were excluded from analyses presented inthis study. A sixth character (production of prominent, leafyvegetative culms) could not be assessed in the herbarium be-cause collections of vegetative shoots are rarely sufficient.This character is nonetheless highly distinctive (Eaton 1959;Reznicek and Catling 1986) and was scored for this studyfollowing Mackenzie.Excluding the two continuous characters, the remaining

four characters were coded as categorical with two states:scales as long as and effectively concealing the perigynium(1) or perigynia exposed (0); inner face of the leaf sheathsgreen and nerved to the apex (1) or hyaline, unnerved (0);vegetative culms prominent throughout the year, leavesspreading and distributed along the entire length of the shoot(1) or vegetative culms inconspicuous or prominent mainlylate in the season, leaf blades mostly clustered near the apex(0); and inflorescence bracts leaflike and/or conspicuouslyexceeding the head (1) or setaceous and typically not as longas the head (0). Character states were based primarily onliterature reports (Reznicek 1993; Mastrogiuseppe 2002;Mastrogiuseppe et al. 2002), with adjustments based on ob-servation of herbarium material where literature accountswere inadequate or polymorphisms were reported. Characterevolution was reconstructed in Mesquite 1.06 (Maddisonand Maddison 2005) under the Mk1 (one-parameter) modelon both the fully resolved Bayesian ‘‘allcompat’’ tree and asubset of 1000 trees from the Bayesian analysis. Because theplacement of C. maackii as sister to the New World clade isstrongly supported, while the sister to section Ovales as awhole could not be determined, reconstructions in this studywere limited to the New World clade. Biogeography wascoded as a four-state character (0: Central and South America;1: western North America [Rocky Mountains and westward];2: eastern North America [east of the Rocky Mountains]; 3:Old World) and, because many taxa are polymorphic, ana-lyzed using parsimony.

Cytology

Chromosome analyses of a single C. roraimensis plant(Reznicek 11054) followed the technique of Cooperrider andMorrison (1967), as described by Rothrock and Reznicek(1996). In brief, immature spikes were collected in June andpreserved in methanol, chloroform, and propionic acid(6 : 3 : 2). Anthers were dissected from spikes and squashedin 2% lactic-acetic-orcein and viewed using phase contrast at31000. Seven pollen mother cells (PMCs) were inspected atfirst meiotic interphase. Drawings, photographs, and voucherspecimen have been deposited at the University of MichiganHerbarium (MICH).

trees is almost identical in topology; branches that collapse in the MP strict consensus are indicated with a dash. Geographic distributions followthe most current published accounts. Outgroups are not pictured because the placement of section Ovales within subgenus Vignea is uncertain.

Support for the branch that connects Carex maackii to the remainder of the section is based on analysis with outgroups included. Node labels

correspond to clades discussed in the text and character reconstructions (table 2). Arrowheads indicate the long-distance dispersal (LDD) eventsrequired to explain the geographic distribution of species. LDD events indicated with a question mark are in question either because vicariance

cannot be ruled out (in the case of C. interjecta) or because phylogenetic uncertainty makes biogeographic reconstruction at the base of the tree

problematic (in the case of C. bohemica).

1035HIPP ET AL.—PHYLOGENY OF CAREX SECTION OVALES

Results

ITS þ 5:8S Data Matrix

Aligning the ITS data set required inserting 36 gaps, includ-ing a variable-length gap of 2–5 base pairs long at the 39 endof a poly-A region. No indels were located in the 5.8S region.The aligned ITSþ 5:8S data matrix contains 637 nucleotidepositions; the first three nucleotides at the 59 end of ITS1 wereexcluded from all taxa because of inconsistent sequencing atthat end. The ITS1 region ranges from 216 to 222 nucleotideslong (excluding the first three nucleotides), ITS2 from 219 to227 nucleotides. The 5.8S region is 164 nucleotides in alltaxa. Across the entire data set, 155 nucleotide positions arepotentially parsimony informative; this number drops to 146when outgroup taxa are excluded and to 63 when all taxa areremoved except for members of sections Ovales, Cyperoideae,and Chordorrhizae. Sequences in the ITS regions are63.291% Gþ C, and homogeneity of base frequencies acrosstaxa cannot be rejected (x2 ¼ 77:029, df ¼ 366, P ¼ 1:000).There is no evidence of mutational saturation within subgenusVignea, but comparisons between subgenus Vignea and theoutgroup suggest some degree of saturation.For Ovales and outgroups, the K80 model is selected for

the 5.8S region using both the default hierarchical likelihoodratio test (hLRT, a ¼ 0:01) and AIC in Modeltest 3.6. TheAIC weight for this model, however, is only 0.2536, with 21models needed to come up to 0.95 cumulative AIC weight,and the five models >0.05 AIC weight have only one or twofree parameters. This probably reflects the fact that there isvery little information in the 5.8S sequences. The GTRþ IþGmodel is the best fit to the ITS regions using AIC (weight ¼

0:6371); the cumulative AIC weight of GTRþ IþG andGTRþG is 0.9766, suggesting a good deal of confidence inthe model selected. The model selected under the defaulthLRT (a ¼ 0:01) is TrNþG, which has an AIC weight ofonly 0.0002. The GTRþ IþG model would be selected usingthe default hLRTat a > 0:097552.Sequences were obtained from nine to 10 clones of each of

seven taxa (table 1). The largest number of unique sequences(eight) was found in Carex bonplandii and the lowest numberin C. straminea (two). The largest uncorrected interclonal diver-gence (15 nucleotides) is within C. bonplandii. Clones for eachindividual are monophyletic except in C. bonplandii, the clonesof which are paraphyletic with regard to C. roraimensis (fig. 2).

ETS Data Matrix

Aligning the ETS data set required inserting 32 gaps. Un-aligned sequences range from 601 to 609 nucleotides long,and the aligned data matrix contains 624 nucleotide posi-tions. Across the entire data set, 224 nucleotide positions arepotentially informative. This number drops to 184 withoutoutgroups and to 73 when only Ovales is included (excludingC. illota). Excluding C. roraimensis and C. bonplandii (see‘‘Congruence and Analysis of Combined Data’’) leaves 57 po-tentially informative characters. GC content is 52.931%, andhomogeneity of base frequencies across taxa cannot be re-jected (x2 ¼ 77:247, df ¼ 366, P ¼ 1:000). Genetic distanceand Ti/Tv are positively correlated for pairwise comparisonswithin section Ovales and subgenus Vignea. A nearly flat re-lationship between Ti/Tv and genetic distance in comparisonsbetween the outgroup and all taxa within subgenus Vignea,combined with Ti/Tv ratio not exceeding 1.6 and averaging

Table 2

Likelihood Reconstructions of Ancestral Character States

Major clades within section Ovales Minor clades within section Ovales

New World root

(n ¼ 80)

WNA internal

(n ¼ 53)

ENA root

(n ¼ 39)

LFS I

(n¼3)

‘‘Ath.’’ þ ‘‘Cyp.’’

(n ¼ 3)

ENA II

(n ¼ 11)

‘‘Trib.’’

(n ¼ 3)

Scales concealing perigynia:

Single-tree reconstructiona 0 (0.7591; ns) 0 (0.9529) 0 (0.9864) 1 (0.9796) 0 (0.9983) 0 (1.000) 0 (1.000)

1000-tree reconstructionb 0: 0.459; 1: 0.007 0: 0.713; 1: 0.001 0: 0.937 1: 0.996 0: 1.000 0: 1.000 0: 1.000Lowest bracts foliose:

Single-tree reconstructiona 0 (0.9999) 0 (1.000) 0 (1.000) 0 (0.9998) 1 (0.9933) 0 (1.000) 0 (1.000)

1000-tree reconstructionb 0: 0.992 0: 1.000 0: 1.000 0: 1.000 1 (0.997) 0: 1.000 0: 1.000

Leaf sheaths green (inner face):Single-tree reconstructiona 0 (0.9963) 0 (0.9753) 0 (0.6823; ns) 0 (0.9976) 0 (0.9985) 1 (0.8128; ns) 1 (0.9622)

1000-tree reconstructionb 0: 0.995 0: 0.914 0: 0.047; 1: 0.052 0: 1.000 0: 1.000 1: 0.352 1: 0.842

Vegetative culms prominent:Single-tree reconstructiona 0 (1.000) 0 (1.000) 0 (1.000) 0 (1.000) 0 (1.000) 0 (1.000) 1 (0.9944)

1000-tree reconstruction 0: 1.000 0: 1.000 0: 1.000 0: 1.000 0: 1.000 0: 1.000 1: 0.989

Note. WNA ¼ western North American; ENA ¼ eastern North American. Other abbreviations correspond to clades labeled in figures 3and 4. Ancestral character states were reconstructed using a one-parameter model (Mk1) in Mesquite 1.06. Underlined text indicates the recon-

struction that is least common for each character among the nodes tested. The major clades are nested and encompass three nodes along the

spine of the tree supported at posterior probability ðPPÞ �95%. The minor clades are smaller clades independent of one another and roughly

corresponding in size to Mackenzie’s species groups.a Proportional likelihood of the higher-likelihood reconstruction on the Bayesian ‘‘allcompat’’ topology (the fully resolved topology that is

compatible with the greatest number of trees encountered after the Markov chain Monte Carlo burn-in). Reconstructions that differ by <2.0 in

log likelihood from the lower-likelihood reconstruction are indicated as ‘‘ns’’ (not significant).b Proportion of 1000 Bayes trees for which the indicated reconstruction is significantly more likely than the less likely reconstruction; propor-

tions in many cases do not sum to 1.000 because equivocal (nonsignificant) reconstructions are not reported.

1036 INTERNATIONAL JOURNAL OF PLANT SCIENCES

1.19, suggests that a significant percentage of base pair sub-stitutions between the outgroup and ingroup are saturated(Halanych et al. 1999).For Ovales and outgroups, the TVMþ IþG model has

the highest AIC weight (0.2450), with another nine modelsrequired to come up to 0.95 cumulative AIC weight. The de-fault hLRT selects the HKYþG model at a ¼ 0:01 (AICweight ¼ 0:0217, which falls within the 95% confidence in-terval). The order of model selection in the default hLRT pre-cludes reaching the TVMþ IþG model for this data set. Forthe full data set, the GTRþG and GTRþ IþG modelshave nearly equal AIC weights (0.3927 and 0.3817, respec-tively), with five models making up 95% of the cumulativeAIC weight. The default hLRT (a ¼ 0:01) selects theTrNþ IþG model, which has an AIC weight of 0.0126 andfalls just outside the 99% confidence interval. The defaulthLRT selects GTRþ IþG at a > 0:072453.Sequences were obtained from nine to 10 clones of each

of seven taxa (table 1). The largest number of unique se-quences (eight) was found in C. bonplandii. The lowest num-ber (four) is found in four taxa, including C. roraimensisand members of sections Ovales and Stellulatae. Interclonaluncorrected distances within individuals are similar acrosstaxa (minimum one, maximum three to six). Clones foreach individual are monophyletic except in C. interior, theclones of which are paraphyletic with regard to C. echinata(fig. 2).

Congruence and Analysis of Combined Data

The ILD test was conducted across a symmetric gradientof relative data partition weights (5 : 1 through 1 : 1), andstrongest incongruence was found when partitions areweighted 1 : 1. At that weighting, the ILD P value for the entiredata set ¼ 0:002, strongly rejecting the null hypothesis thatthe data are drawn from a homogeneous pool of characters.Excluding all taxa outside of sections Ovales and Cyperoi-deae increases ILD P value to 0.040. Excluding C. roraimen-sis and C. bonplandii in addition to taxa outside of sectionsOvales and Cyperoideae increases ILD P value to 0.136, sug-gesting that these two taxa are a major source of incongru-ence in the data and that aside from the resolution of thesetwo taxa, there is little if any incongruence between ITS andETS in the resolution of section Ovales.To evaluate whether the topological difference between

ITS and ETS in the position of C. maackii is significant, anadditional test was performed with C. maackii and all taxaoutside of sections Ovales and Cyperoideae excluded fromanalysis, but with C. bonplandii and C. roraimensis retained.This test is also significant (P ¼ 0:030), suggesting thatC. maackii is not a significant source of incongruence be-tween the data partitions. Because only the ETS data fail torecover a monophyletic section Ovales, the KH and SH testswere used to investigate which specific hypotheses arestrongly rejected by the ETS data. Monophyly of sectionOvales including C. maackii but excluding C. bonplandii andC. roraimensis is not rejected (SH P ¼ 0:6969). However, thehypothesis that section Ovales is monophyletic includingC. bonplandii and C. roraimensis, with C. maackii con-strained to be sister to the remainder of the section, is re-jected by the ETS data (SH P ¼ 0:0138).

For purposes of evaluating the support for alternative hy-potheses on the combined data using SH and KH tests inPAUP* (which does not currently support separate modelsfor different data partitions), a single nucleotide substitutionmodel was selected for the full data set. The GTRþ IþGmodel was selected (AIC weight ¼ 0:9953), though the de-fault hLRT selects the TrNþ IþG model at a ¼ 0:01 (AICweight ¼ 2:433 10�5).

ITS Phylogenetic Results: All Taxa

Heuristic searches of the full ITS data matrix recovered41,977 unique most parsimonious trees of 658 steps (consis-tency index [CI] ¼ 0:517, retention index [RI] ¼ 0:754). Be-cause this search was limited to swapping on 1000 trees ateach random-addition replicate, many more MP trees arelikely to exist, and island structure was not investigated. Thestrict MP consensus (not shown) resolves somewhat fewernodes than the Bayesian majority-rule tree (fig. 3), but thereare no major points of incongruence between the trees recov-ered in the MP and Bayesian analyses.Excluding C. illota, monophyly of section Ovales is

strongly supported: posterior probability ðPPÞ ¼ 1:00, parsi-mony bootstrap ðPBÞ ¼ 80%. Carex chordorrhiza, C. pseu-docuraica, and C. brunnescens form a strongly supportedclade that is weakly supported as sister to Ovales (PP ¼ 0:94,PB < 50%). The largest well-supported clade within sectionOvales is an eastern North American (ENA I) clade made upprimarily of the eastern members of the ‘‘Festucaceae’’ andincluding the ‘‘Tribuloideae’’ and two of the ‘‘Alatae’’ (fig. 3,ENA I; PP ¼ 1:00, PB ¼ 69%). This clade derives from anear polytomy of mostly western North American species.Carex maackii resolves as sister to the remainder of sectionOvales (PP ¼ 0:99, PB ¼ 61%).

ETS Phylogenetic Results

Heuristic parsimony searches of the ETSþ indels data setrecovered 72,872 unique trees of 744 steps, CI ¼ 0:569,RI ¼ 0:812. Topology of the strict consensus (tree notshown) is essentially identical to that of the Bayesian analysis(fig. 3), though it is less resolved. Monophyly of sectionOvales, with the exception of C. maackii, C. bonplandii, andC. roraimensis, is strongly supported (PP ¼ 1:00, PB ¼ 97%).Carex maackii falls sister to a clade composed of Ovales andseveral other sections in the Bayesian analysis and sister to C.bromoides (section Deweyanae) in the strict consensus of MPtrees but is without strong statistical support for separating itfrom Ovales in either case. Carex bonplandii and C. rorai-mensis are placed within section Stellulatae with high sup-port (PP ¼ 1:00, PB ¼ 97%).

Combined Analysis

The two nrDNA data partitions support different resolu-tions for the sister group to section Ovales (fig. 3). Removalof all taxa outside of section Ovales results in a 20-fold de-crease in ILD significance (from P ¼ 0:002 to P ¼ 0:040).Combined analysis in this study was conducted usingmembers of section Glareosae and Chordorrhizae as out-groups, but trial analyses using other sections as outgroupsrecover the same rooting for the section. Carex maackii

1037HIPP ET AL.—PHYLOGENY OF CAREX SECTION OVALES

is conclusively placed as sister to the remainder of thesection.Although ITS and ETS recover different topologies, the

ILD and KH tests provide no strong evidence of incongru-ence between the data partitions regarding relationshipswithin Ovales. Likewise, the data partitions are consistent re-garding circumscription of the section, with two exceptions.First, the ETS data set, including all taxa, does not placeC. maackii at the base of Ovales, as the ITS data set does.However, placement of C. maackii is weakly supported inthe ETS data set but relatively strongly supported in the ITSdata set. In combination, the position of C. maackii as sisterto the remainder of section Ovales is strongly supported.Consequently, this does not appear to be a case of incongru-ence but rather of inadequate data in the ETS data partitionalone to resolve the position of C. maackii. Thus, the onlypoint of real topological incongruence between ITS and ETSregarding the circumscription of section Ovales appears to bethe position of C. bonplandii and C. roraimensis.Combined analysis of section Ovales (excluding C. bon-

plandii, C. roraimensis, and C. illota, and using C. chordor-rhiza and C. brunnescens as outgroups) recovers a phylogenythat is largely resolved within section Ovales, with parsi-mony and Bayesian analysis (fig. 4) and maximum likelihood(ML; not shown) supporting the same well-supported rela-tionships. Maximum parsimony recovers 78,494 unique treesof 459 steps (CI ¼ 0:691, RI ¼ 0:819). The shortest tree re-covered using the strict consensus as a reverse constraint is460 steps (97,900 trees recovered under the same searchparameters), suggesting that additional search effort wouldbe highly unlikely to affect the strict consensus. ML analysisrecovers a tree of log likelihood ðln LÞ ¼ �4204:82504 (treenot shown). Average ln L of three independent Markov chainMonte Carlo (MCMC) runs is �4464:06716 0:3864 stan-dard error. The difference in likelihood scores in the Bayesianand ML searches reflects in part the difference in modelsused for phylogenetic reconstruction and the inclusion of in-dels in the Bayesian analyses. Under the GTRþ IþG modelwith parameters fixed as in the ML search, ln L of the fullyresolved allcompat consensus tree is �4208.23427, whichdiffers insignificantly from the ML topology (one-tailed KHP ¼ 0:1628 under the GTRþ IþG model, using 10,000RELL bootstrap replicates).Two major groupings are recovered within the section (fig.

4): a western North American (WNA) grade that containsmore than half of the taxa in the section and an easternNorth American clade that is subdivided into two smallerclades (ENA I and ENA II). The Asian C. maackii resolves assister to the remainder of the section.

Character Evolution and Monophyly ofMackenzie’s Species Groups

The morphological characters studied, a subset of thoseused by Mackenzie to define species groups within Ovales,

are all homoplastic (table 2; fig. 5). Moreover, the ancestorsto the larger clades that originate along the spine of the phy-logeny share the same most likely set of character states(table 2). The character traits that define the section’s moredistinctive species groups—leafy, reproductive vegetativeculms, foliose inflorescence bracts, green-nerved inner leafsheaths, and pistillate scales concealing the perigynia—allevolved at the base of smaller clades embedded within thesection.Mackenzie’s species groups are polyphyletic except for

the ‘‘Athrostachyae,’’ the two species of which fall sister toC. sychnocephala (fig. 4). The SH test and Bonferroni-corrected KH test both indicate that the data fail to reject mono-phyly of the ‘‘Tribuloideae,’’ ‘‘Cyperoideae,’’ and ‘‘Festivae’’circumscribed narrowly (table 3). The SH and KH tests alsosuggest that a section Ovales that excludes C. sychnocephalaand C. bohemica (of section Cyperoideae) is not stronglyrejected (GTRþ IþG SH P ¼ 0:2574, corrected KHP ¼ 0:0533). However, optimizing branch lengths on theconstrained topology produces a zero-length branch at theroot of the section and an exceedingly long branch leading toC. maackii, approximately five times as long as the next lon-gest branch in the section. Trees recovered with C. maackiiexcluded have branches nearly the same length as those ofthe unconstrained tree. Consequently, the constrained treeseems at odds with a reasonable phylogenetic reconstructionfor the section, despite the fact that it is not rejected in topol-ogy tests. Monophyly of the remainder of the Mackenzianspecies groups is rejected by both tests (P < 0:05).

Cytology of Carex roraimensis

Observations of PMCs of C. roraimensis at first meiotic in-terphase show no indication of monovalents or trivalents(fig. 6), which would suggest aneuploidy as a consequence ofchromosome fission. None of the PMCs showed evidence oftetravalents either, which are expected in autopolyploids dueto homology between duplicated chromosomes. A chromo-some count of n¼62 (2n¼124) was confirmed throughinspection of seven PMCs. This is the highest euploid chro-mosome number known in the genus, excluding aberrant tet-raploid cells observed in C. aquatilis 3 C: paleacea (2n¼74;Cayouette and Morisset 1985).

Discussion

Data Congruence and the Origin of theCarex bonplandii Complex

The ILD, KH, and SH tests suggest that there are twosources of incongruence in the data set presented here: reso-lution of taxa outside of section Ovales and the placementof the C. roraimensis–C. bonplandii clade (referred to here-after in this article as the C. bonplandii complex, followingReznicek 1996). There seems to be no incongruence in the

Fig. 5 Maximum likelihood reconstructions of morphological characters on Bayesian ‘‘allcompat’’ consensus (fully resolved tree that issupported by the largest subset of trees sampled using Markov chain Monte Carlo). Circles at the nodes are shaded according to the proportional

likelihood of each character state (black denotes character state as described; white denotes the alternative state). Likelihoods at labeled nodes are

summarized in table 2. Node abbreviations are the same as in table 2.

1039HIPP ET AL.—PHYLOGENY OF CAREX SECTION OVALES

placement of C. maackii. Rather, its placement outside ofsection Ovales in the ETS data set is due to lack of support,as evidenced by the fact that trees in which C. maackii fallssister to the remainder of section Ovales are supported byboth data sets. Notably, the ETS data fail to reject treesconstrained so that section Ovales is monophyletic with theinclusion of the C. bonplandii complex and C. maackii.This resolution presumably is not rejected because it doesnot necessitate breaking up section Ovales; the constrainedtree resolves section Stellulatae (including C. roraimensisand C. bonplandii) as paraphyletic with respect to sectionOvales, with C. maackii sister to the remainder of sectionOvales. However, when the ITS resolution is enforced—C.maackii sister to section Ovales and the C. bonplandii com-plex embedded within Ovales—the ETS data strongly rejectthe tree.Incongruence in the placement of the C. bonplandii com-

plex, while strongly supported by all tests employed, appearsnot to be reflected in intraspecific divergence among ITS se-quences or ETS sequences (fig. 2). The raw sequences alsoshow very few double peaks, minimal length polymorphism,almost no variability in the 5.8S region, and no evidence ofheterogeneity in base composition between sequences, any ofwhich might suggest that divergent paralogs or pseudogenesare posing a potential problem for phylogenetic reconstruc-tion (Buckler et al. 1997). The clones sequenced in seventaxa recover no pseudogenes and suggest that nrDNA para-logs coalesce subsequent to the origin of the species studied,though paraphyly of ITSþ 5:8S clones in C. bonplandii andETS clones in C. interior (fig. 2) may be due to retention of

ancestral polymorphisms or ongoing gene flow between closerelatives. Laboratory error is not a likely explanation for in-congruence between ITS and ETS because both C. bonplandiiand C. roraimensis have been sequenced multiple times forboth data partitions (separate sequences not shown), andlargely unambiguous sequence alignment within both regionsrules out the possibility that the difference is an artifact ofalignment. A strong positive relationship between Ti/Tv ratioand corrected pairwise genetic distances within subgenus Vig-nea suggests that substitutional saturation is not an excessivesource of noise that could potentially affect both the accu-racy of phylogenetic reconstruction and results of the ILDtest.The data thus appear to support a hybrid origin for the

C. bonplandii complex, with ETS homogenizing to one pa-rental type (section Stellulatae) and ITS to the other (sectionOvales) (Okuyama et al. 2005). This hypothesis is consistentwith the morphology of the C. bonplandii complex, themembers of which are characterized by narrowly and bluntlywinged perigynia that resemble those of sections Deweyanae,Remotae, or Stellulatae. In fact, while taxonomists workingin the twentieth century have for the most part recognizedthe C. bonplandii complex as part of section Ovales(Mackenzie 1931–1935; Reznicek 1993), Kukenthal (1909)considered C. bonplandii to be part of section ElongataeKunth, a conglomerate section that has subsequently beensplit up into sections Deweyanae, Remotae, Stellulatae, andGlareosae in part (Hipp 2004). Hybridization also is consis-tent with the habitats and biogeographic distribution of thecomplex. Carex bonplandii is a conspicuously rhizomatous

Table 3

Topology Tests of Alternative Hypotheses

�ln L Diff. �ln LaKH

P value

KH

P value,

Bonferroni

corrected

SH

P value

Unconstrained (ML tree) 4204.82504 . . . . . . . . . . . .Ovales s.s. without ‘‘Cyperoidea’’ 4238.56830 33.74326 0.0031 0.0403 0.2050

‘‘Alatae’’ 4291.49387 86.66883 0.0000 0.0000 0.0012

Cyperoideae 4225.28575 20.46071 0.0186 0.2418 0.5231‘‘Festivae’’ broad 4273.44220 68.61716 0.0000 0.0000 0.0057

‘‘Festivae’’ narrow 4244.55044 39.72540 0.0040 0.0520 0.1244

‘‘Festucaceae’’ broad 4370.61901 165.79397 0.0000 0.0000 0.0000

‘‘Festucaceae’’ narrow 4345.64365 140.81861 0.0000 0.0000 0.0000‘‘Fetae’’ 4258.44594 53.62090 0.0007 0.0091 0.0288

‘‘Foeneae’’ 4290.31199 85.48695 0.0000 0.0000 0.0015

‘‘Foeneae’’ without C. arapahoensis 4266.07341 61.24837 0.0002 0.0026 0.0178

‘‘Leporinae’’ 4300.22860 95.40356 0.0000 0.0000 0.0000‘‘Specificae’’ 4306.42435 101.59931 0.0000 0.0000 0.0000

‘‘Tribuloideae’’ 4211.55587 6.73083 0.1613 1.0000 0.8558

Note. Kishino-Hasegawa (KH) and Shimodaira-Hasegawa (SH) tests were conducted with the unconstrainedmaximum likelihood topology and with maximum likelihood trees recovered with constraints indicated. The

Bonferroni-corrected P value (P 3 13) is based on the total number of tree comparisons (n � 1 ¼ 13; here

n ¼ number of trees being compared). Tests were conducted with all taxa indicated in figures 4 and 5. ‘‘ ‘Festi-vae’ broad’’ includes all taxa labeled ‘‘Festivae’’ and ‘‘Festivae or Festucaceae’’ in table A1; ‘‘ ‘Festivae’ narrow’’

excludes the latter group. Likewise, ‘‘ ‘Festucaceae’ broad’’ includes the taxa in ‘‘Festivae or Festucaceae,’’ while

‘‘ ‘Festucaceae’ narrow’’ excludes them.a Difference in log likelihood between each tree and the unconstrained (ML) tree.

1040 INTERNATIONAL JOURNAL OF PLANT SCIENCES

species of paramo and wet meadows at elevations of 1800 to>4500 m, ranging from Bolivia to Mexico. Most species ofsection Stellulatae occupy wet, sunny, peaty, or mineral soils,the kinds of habitats where C. bonplandii grows, and Mexi-co, Central America, and northern South America harborseveral Stellulatae taxa. While intersectional hybrids are un-common in the genus, they are not unknown (Cayouette andMorisset 1992), and the combination of molecular and mor-phological data for a hybrid origin for the C. bonplandiicomplex is compelling.An intriguing possibility is that the C. bonplandii complex

may represent an allopolyploid lineage. Carex roraimensishas one of the highest chromosome counts known in Carex(n¼62; fig. 6), approximately the sum of mean counts in sec-tions Ovales and Stellulatae. Moreover, chromosome figuresinspected at first meiotic interphase in numerous PMCs ofthe C. roraimensis individual studied show only bivalents(fig. 6), suggesting that autopolyploidy is not a likely expla-nation for the high number found. If confirmed, this wouldbe the first documented case of allopolyploidy in the genusCarex. Ongoing work aims at testing this hypothesis usingadditional molecular and chromosomal data.

Diversification and the Sister Group to Section Ovales

A sister relationship between the New World clade thatmakes up the core of section Ovales and the east Asian spe-

cies C. maackii is strongly supported. Assuming that at-tempts to sample all close relatives of the section have beensuccessful, this finding suggests a high relative rate of diversi-fication within the New World Ovales (assuming 89 taxawithin the New World clade and one taxon sister tothat clade; P ¼ 0:011 under Slowinski and Guyer’s [1993]test of relative diversification). Increased diversification inthis clade is probably not explained by appeal to the distinc-tive synapomorphies of the section as key innovations, forneither winged perigynia nor vegetative culms have resultedin increased diversification in the other clades in which theyoccur (Hipp, forthcoming). The role of chromosomal evolu-tion in diversification in the section may be a more reason-able explanation and is under further investigation (Hippet al. 2006).The sister group to section Ovales as a whole is unresolved

in this study (fig. 3). The ITS data recover a clade composedof C. chordorrhiza, C. pseudocuraica, and C. brunnescens(PP ¼ 1:00, PB ¼ 70%) as sister to Ovales (PP ¼ 0:94,PB < 50%), while the ETS data set recovers a clade com-posed of a wide range of sections (including Chordorrhizae,Phaestoglochin, Remotae, Glareosae, Stellulatae, Deweya-nae, Ammoglochin, Holarrhenae, Divisae, and others;PP ¼ 0:96, PB < 50%) as sister to Ovales (PP ¼ 0:96,PB < 50%). This latter clade is also recovered in combinedanalysis (Ford et al. 2006), but the relationship collapses instrict consensus. Neither of the scenarios supports Savile andCalder’s (1953) hypothesis that section Ovales is sister to sec-tion Ammoglochin or Egorova’s (1999) placement of Ovalesand Cyperoideae in a lineage with section Boernera (repre-sented in this study by C. duriuscula). Also, none of the non-Ovales taxa with winged perigynia (C. remota, C. planata,C. arenaria, C. siccata) fall sister to Ovales under either sce-nario. Identification of the sister group to section Ovales willrequire additional sampling of gene regions and an adequateaccount of the causes of incongruence between the nrDNAregions.While placement of C. illota is also unresolved, the

best estimate of its position based on the data presentedhere is near section Glareosae; this placement is alsosupported by cpDNA data (A. L. Hipp, unpublished data).Kukenthal (1909) placed C. illota into section ElongataeKunth. Morphologically, C. illota might be expected tofall most naturally in one of those sections (Whitkus1988). While inconclusive about the precise placementof the species, the data presented here do not supportMackenzie’s (1931–1935) argument that C. illota’s ‘‘truerelationship seems to be with the Ovales, of which itmay be regarded as one of the most primitive species’’ (p.131).

Biogeography

Geographic ranges correlate closely with phylogenetic rela-tionships in section Ovales (fig. 4), supporting recent findingsin other sections of the genus (Roalson and Friar 2004;Dragon and Barrington, forthcoming). Phylogenetic uncertaintyin the placement of C. foenea (widespread northward),C. bohemica (Eurasia), the clade containing C. argyrantha(northeast and northern Great Lakes), and section Ovales

Fig. 6 Meiotic figure from one pollen mother cell (PMC) ofCarex roraimensis (Reznicek 11054 [MICH]). Based on inspection of

seven PMCs from the individual vouchered, the figures shown are

interpreted as small bivalents. No tetravalents or other irregular pairing

relationships were detected in any of the PMCs inspected, as would beexpected in an autopolyploid. Note that chromosome groups on the

lower edge of the nucleus are difficult to view in this image, in which

abutting bivalents may resemble trivalents. These resolve clearly withthrough-focusing in this and other cells.

1041HIPP ET AL.—PHYLOGENY OF CAREX SECTION OVALES

among the other sections of subgenus Vignea preclude defini-tive statements about the geographic origins of the section.However, concentration of the western North American taxain a paraphyletic grade that reaches to the base of the sec-tion, accompanied by strong Bayesian support for the easternNorth American clade (PP ¼ 1:00, PB ¼ 0:54%), suggeststhat section Ovales in North America originated in thewest and spread eastward. Moreover, parsimony reconstruc-tion of biogeographic regions on the Bayes 1000-tree sub-sample recovers western North America as the ancestralrange for the New World clade at the root of North Americanclade in 808 trees (eastern North America is reconstructedat the base in 283 trees and the Old World in 680; these sumto >1000 because the figures include equivocal reconstruc-tions). Whether the position of C. maackii as sister to the re-mainder of the section reflects a Eurasian ancestry for thesection or long-distance dispersal from a western NorthAmerican progenitor cannot be ascertained without an accu-rate assessment of the section’s relationship to the remainderof the subgenus.In addition to diversification of lineages in situ (e.g., the

western North American clades labeled on fig. 4 andclades ENA I and ENA II), there appear to have been atleast two cases of more recent long-distance dispersalinvolved in the diversification of section Ovales: dispersalfrom eastern North America to the West Coast givingrise to C. feta and the West Coast populations of C. ovalisand dispersal to Eurasia giving rise to the Old Worldpopulations of C. ovalis (fig. 4). Additional long-distancedispersal events may be needed to explain C. interjecta(Mexico), C. bohemica (Eurasia), and C. maackii (eastAsia), but phylogenetic uncertainty at the base of the treemakes biogeographic history difficult to infer for the lattertwo taxa, and vicariance cannot be ruled out for C. inter-jecta.

Character Evolution and the Status ofMackenzie’s Species Groups

The morphological characters studied demonstrate a his-tory of parallel evolution in the section, with some morpho-logical innovations (e.g., elongate bracts) arising rarelybut others (e.g., pistillate scales that cover the perigynia,green-veined leaf sheaths) evolving repeatedly (fig. 5). Thispattern points to the problems with Mackenzie’s infrasectionalclassification and the difficulties of erecting an alternativesystem.Despite the fact that none of the morphological characters

investigated map cleanly onto the phylogeny, the easternNorth American clade and the western North Americangrade are morphologically distinctive. Indeed, two separatekeys are presented in the recent Flora of North AmericaNorth of Mexico section Ovales treatment for specieseast of the Rocky Mountains and species of the RockyMountains and westward (Mastrogiuseppe et al. 2002), cor-responding closely to the western North American gradeand the eastern North American clade presented in thisstudy. The eastern and western groups appear not to bedefined by a single character but, with some exceptions, dohave distinctive appearances, such that species can be placed

with some confidence. Eastern members of the sectiontypically have elongate inflorescences with relatively pale(hyaline to reddish brown) pistillate and staminate scales andflattened, serrulate margined beak apices. Western speciesfrequently have compact headlike inflorescences and darkerpistillate and staminate scales, and frequently the apicalportion of the beak is terete and smooth. While no singlecharacter may turn out to diagnose the eastern NorthAmerican clade, the look is distinctive enough to suggest thatwith additional work, characters may be found to distinguishthe two.

The Eastern North American Clade:ENA Clade I and ENA Clade II

Relationships within clade ENA I based on the nrDNAdata are poorly resolved, and several of the sister speciesrelationships implied by the nrDNA data seem improb-able in light of the morphology. This could result if radia-tion in the clade had proceeded at a higher rate thanmutation, leaving an unreliable imprint of phylogeneticrelationships in the nrDNA sequence data. Phylogeny ofclade ENA I has been addressed in detail in a separatestudy using AFLP data, which provide strong supportregarding species relationships in the clade (Hipp et al.2006).The other major eastern North American clade (ENA II)

is composed of members of three species groups: the major-ity of the ‘‘Alatae,’’ characterized by obovate perigyniumbodies and leaf sheaths that are green and veined on theinner face nearly to the apex; the eastern North Ameri-can members of the ‘‘Fetae,’’ characterized by perigyniumbodies that are widest at or below the middle and leafsheaths green and veined on the inner face nearly to the apex(like the ‘‘Alatae’’); and C. scoparia of the ‘‘Festucaceae,’’ ahighly polymorphic species that ranges across much ofnorthern North America but is much more common east ofthe Rocky Mountains than it is westward. The close relation-ship between the ‘‘Alatae’’ and ‘‘Fetae’’ is not unexpected(Rothrock et al. 1997): the two share the characteristic ofgreen-veined inner leaf sheaths, several species in each havespikes widely spaced in erect to arching inflorescences, andthe distinction between obovate and ovate perigynium bod-ies is challenging, at best, in several of the taxa. One taxonin this clade (C. interjecta Reznicek) is known only from thetype locality, a moist meadow in Morelos (Mexico), andwas not known to Mackenzie. The plant is similar in ap-pearance to C. longii, with elliptical to obovate perigyniaand leafy vegetative culms, but it has the hyaline inner leafsheaths, narrow leaf blades, and dark pistillate scales typicalof the other Mexican members of the section (Reznicek1993). The inclusion of C. scoparia within clade ENA II issomewhat surprising because it more closely resembles sev-eral taxa within clade ENA I. The exclusion of C. cumulataof the ‘‘Alatae,’’ with its green-nerved inner leaf sheaths andobovate perigynia, is also unexpected. These results are sup-ported, however, by AFLP data (Hipp et al. 2006) as wellas by sequences from additional individuals (data notshown).

1042 INTERNATIONAL JOURNAL OF PLANT SCIENCES

The Western North American Grade

The major species groups of the western North Americangrade (the‘‘Festivae,’’ ‘‘Leporinae,’’ ‘‘Foeneae,’’ ‘‘Specificae,’’and ‘‘Cyperoideae’’) are all polyphyletic. Well-supported cladeswithin the western grade fall into two classes: single-species-group clades and clades composed of two to three morpho-logically similar species groups. In the latter falls the cladecontaining the ‘‘Athrostachyae’’ and C. sychnocephala, whichwas discussed in conjunction with monophyly of sectionOvales.With ca. 20 recognized species, the ‘‘Festivae’’ is the second

largest of Mackenzie’s species groups (after the ‘‘Festuca-ceae’’). The ‘‘Festivae’’ breaks apart along biogeographic linesinto two major clades: one distributed primarily in thecoastal ranges and Sierra Nevadas, with two outliers in theRocky Mountains (‘‘Festivae II’’), and one distributed primar-ily in the southern Rocky Mountains (‘‘Festivae I’’). ‘‘FestivaeII’’ includes the widespread C. macloviana, a bipolar disjunctthat bears strong morphological similarity to C. haydenianaof ‘‘Festivae I.’’ It is surprising to find C. macloviana separat-ing from C. haydeniana and C. microptera in this study,though the result has been verified using two accessions, bothfrom the Rocky Mountains. Similarly, C. stenoptila is a dis-tinctive species with narrow, distinctly veined perigynia thatis allied morphologically to C. microptera, also of ‘‘FestivaeI.’’ Why these species should be related to the coastal/Sierrantaxa of ‘‘Festivae II’’ is unclear. These taxa may be productsof hybridization that is not revealed in incongruence betweenthe nrDNA data sets. The ITS data alone recover a cladecomposed of the coastal range species (C. gracilior, C. sub-bracteata, C. pachystachya, and C. harfordii; PP ¼ 0:91).The ETS data alone recover these taxa as a grade giving riseto the C. macloviana–C. mariposana–C. stenoptila clade. Inthe combined data, C. harfordii is recovered as part of thisclade at PP ; 0:50, sufficiently close to the 50% thresholdthat it is excluded from the 50% majority-rule tree shown infigure 4.‘‘Festivae I’’ is part of a larger, well-supported southern

Rocky Mountain clade that includes C. wootonii (‘‘Specificae’’)and C. egglestonii (‘‘Festucaceae’’). Two species in thisclade, C. microptera and C. haydeniana, extend well into thenorthern Rocky Mountains and intergrade in portions oftheir range (Whitkus and Packer 1984). As is the case in‘‘Festivae I,’’ there is a close relationship between the RockyMountain clade and a group of predominantly Sierra Neva-dan/Coastal range species, including C. abrupta, C. subfusca,C. integra, C. teneriformis, C. specifica, C. straminiformis,and C. multicostata. These Sierran/Coastal species form aparaphyletic grade from which the Rocky Mountain clade(‘‘Festivae I’’ and allies) arises, suggesting that at least thesouthern Rocky Mountain ‘‘Festivae’’ may have a Sierranorigin.Most of the remaining western taxa fall into three species

groups: the ‘‘Specificae,’’ a polyphyletic grouping made upof most of the western species with perigynia >6 mm long;the ‘‘Leporinae,’’ made up of the species with terete perigy-nium beaks smooth to the tip and pistillate scales that con-ceal the perigynia; and the ‘‘Foeneae,’’ comprising specieswith flat, often minutely serrulate perigynium beaks and

pistillate scales that conceal the perigynia. The tendency to-ward elongate, appressed perigynia nearly concealed by thepistillate scales gives these species a rather sleek look, forwant of a better term, and the two clades containing themajority of them (‘‘Leporinae-Foeneae-Specificae’’ [LFS]clades LFS I and LFS II; fig. 4) are morphologically cohe-sive. The singular exception is C. preslii, a species of thenorthern Rocky Mountains to Sierra Nevadas and Coastalranges, with perigynia not concealed by the pistillate scalesand perigynia rather ‘‘Festucaceae’’-like in appearance (notslender or terete tipped). The placement of this species ispeculiar enough that it bears verification with additionalspecimens.One of the more interesting finds within the LFS clades is

the placement of a specimen identified as C. constanceanain this study, a species previously known only from its typecollection (Stacey 1938) and represented in this study by aNevada County, California, population more than 425miles from the type locality (Mount Adams, Washington).The specimen is clearly not an accession of C. davyi, towhich it falls sister at a patristic distance of 0.00677 meanchanges per aligned nucleotide, exceeding, for instance, thegenetic distance between C. wootonii and C. peucophila(0.004261) or C. longii and C. vexans (0.003474). Whilegenetic divergence is not sufficient for recognizing separatepopulations as different species, the combination of geneticdistance and morphological divergence suggest that C. con-stanceana may well be an under-recognized and under-collected member of the western North American flora. Anequally plausible alternative is that the collection may rep-resent a local endemic distinct from C. constanceana andperhaps as restricted in distribution as that species. Furtherinvestigation of collected material is warranted, as is col-lecting aimed at finding additional populations of thisplant.

Acknowledgments

We thank Theodore S. Cochrane (WIS), Phillip E. Hyatt(USFS), and Elizabeth H. Zimmerman (WIS) for providingplant material. We also thank the curators of MICH, MO,RM, RSA, and WIS for allowing us to sample herbariummaterial. Technical assistance and advice were provided byEric Roalson, and Julian Starr generously provided ETSprimer sequences before their publication. Comments fromboth Eric and Julian and from Larry Hufford on the firstsubmission of this manuscript improved the paper consider-ably. The manuscript also benefited from comments by A. L.Hipp’s graduate committee: Paul E. Berry (chair), David A.Baum, Thomas J. Givnish, Carol E. Lee, and Kenneth J.Sytsma. Support for this work was provided by a LawrenceMemorial Fellowship from the Hunt Botanical Institute, anE. K. Allen Fellowship through the University of Wisconsin–Madison Department of Botany, a Karling Graduate StudentResearch Award from the Botanical Society of America, anAmerican Society of Plant Taxonomists research award, andNational Science Foundation Dissertation ImprovementGrant 0308975 to A. L. Hipp and P. E. Berry.

1043HIPP ET AL.—PHYLOGENY OF CAREX SECTION OVALES

Appendix

Table A1

Taxa Included in the Study

Section/species group Species Location Collector ITS ETS

Subgenus Carex L.:Acrocystis Dumort. pensylvanica Lam. WI (Dane) Hipp 513 (WIS) AY779137 DQ461013Hymenochlaenae Drej. ex L.H. Bailey gracillima Schwein. WI (Dane) Hipp 505 (WIS) AY779103 DQ460978Phacocystis Dumort. haydenii Dewey WI (Dane) Hipp 501 (WIS) AY779106 DQ460982

Subgenus Vignea (P. Beauv. ex T. Lestib.)Peterm. excluding Ovales:

Ammoglochin Dumort. aarenaria L. USSR Pobedimova 5113 (WIS) DQ461131 DQ460939brizoides L. Czechoslovakia Bohuslavek 694 (MICH) AY779076 DQ460948siccata Dewey NM (Taos) Hipp 2314 (WIS) DQ461148 DQ461034

Chordorrhizae Meinsh. achordorrhiza Ehrh. ex L.f. WI (Ashland) Judziewicz 11790 (WIS) AY779087 DQ460958pseudocuraica Fr. Schmidt China Lin 668 (MO) AY779148 DQ461024

Deweyanae Tuckerm. bolanderi Olney CA (Humboldt) Hipp 480 (WIS) DQ461132 DQ460945bromoides Schkuhr ex Willd. WI (Monroe) Hipp s.n. (WIS) DQ461133 DQ460949adeweyana Schwein. subsp. deweyana WI (Menominee) DeJoode 1543 (WIS) AY779094 DQ460966

Divisae Christ ex Kuk. praegracilis W. Boott CA (Santa Barbara) Hipp 216 (WIS) AY779143 DQ461019duriuscula C.A. Mey. MN (Norman) McNeilus 93-927 (WIS) DQ461136 DQ460967

Foetidae (Tuck. ex L.H. Bailey) Kuk. avernacula L.H. Bailey CA (El Dorado) Hipp & Clifton 680 (WIS) AY779178 DQ461050Glareosae G. Don abrunnescens (Pers.) Poir. WI (Forest) Cochrane et al. 6579 (WIS) DQ461134 DQ460950

acanescens L. WI (Jackson) Hipp et al. 587 (WIS) AY779078 DQ460952Heleoglochin Dumort. acusickii Mack. ex Piper & Beattie MT (Flathead) Schuyler 4989 (WIS) DQ461135 DQ460964

prairea Dewey ex A.W. Wood WI (Rock) Hipp & Zimmerman 602 (WIS) AY779144 DQ461020Holarrhenae (Doll) Pax sartwellii Dewey WI (Dane) Hipp 515 (WIS) AY779154 DQ461030

acuraica Kunth Siberia, West Sayan, Russia Krasnoborov et al. 7/21/1976 (MO) AY779092 DQ460963Macrocephalae Kuk. macrocephala Willd. ex Spreng. OR (Tillamook) Voss 12990 (WIS) DQ461141 DQ460994Multiflorae Kunth avulpinoidea Michx. WI (Rock) Cochrane 13345 (WIS) AY779180 DQ461052Phaestoglochin Dumort. cephaloidea (Dewey) Dewey WI (Trempealeau) Hipp & Rothrock 1220 (WIS) AY779080 DQ460953

cephalophora Muhl. ex Willd. WI (Dane) Hipp 528 (WIS) AY779081 DQ460954hoodii Boott CA (Alpine) Hipp & Clifton 705 (WIS) . . . DQ460983muehlenbergii Schkuhr ex Willd. WI (Dane) Hipp 545 (WIS) AY779124 DQ461001aradiata (Wahlenb.) Small WI (Dane) Hipp 162 (WIS) DQ461147 DQ461025rosea Schkuhr ex Willd. WI (Dane) Hipp 514 (WIS) AY779153 DQ461029occidentalis L.H. Bailey NM (Sandoval) Hipp et al. 2067 (WIS) AY779128 DQ461005

Physoglochin Dumort. gynocrates Wormsk. ex Drejer MI (Alger) Henson 1504 (WIS) DQ461140 DQ460979Potosinae Mack. potosina Hemsl. Mexico (Zacatecas) Villegas & Garcia s.n. (WIS) AY779142 DQ461018Remotae (Aschers.) C.B. Clarke aplanata Franch. & Sav. Japan (Honshu) Tsugaru & Takahashi 26567 (MO) AY779141 DQ461017

remota L. Russia Novikov 5863 (MO) AY779150 DQ461026Stellulatae Kunth exilis Dewey ME (Hancock) Little s.n. (WIS) DQ461139 DQ460972

interior L.H. Bailey WI (Trempealeau) Thompson 399 (WIS) AY779112 DQ460988echinata Murray WI (Marquette) Cochrane 13551 (WIS) DQ461137 DQ460969

Vulpinae Kunth stipata Muhl. ex Willd. var. stipata WI (Dane) Hipp 506 (WIS) AY779162 DQ461037Section Ovales Kunth:‘‘Alatae’’ balata Torr. GA (Calhoun) Rothrock 3922 (MICH) AY779066 DQ460936

albolutescens Schwein. OH (Lawrence) McCormac et al. 6807 (MICH) AY779067 DQ460937acumulata (L.H. Bailey) Mack. ME (Washington) Reznicek 10924 (WIS) AY779091 DQ460962blongii Mack. Mexico (Michoacan) Zamudio et al. 11237 (MICH) AY779115 DQ460991ozarkana P.E. Rothrock & Reznicek AR (Scott) Hyatt 9357 (MICH) AY779135 DQ461011vexans F.J. Herm. FL (Pasco) Rothrock 2379 (MICH) AY779179 DQ461051

‘‘Athrostachyae’’ bathrostachya Olney CA (Nevada) Hipp et al. 794 (WIS) AY779070 DQ460941unilateralis Mack. OR (Benton) Wilson 5882 (MICH) AY779177 DQ461049

1044

Table A1

(Continued )

‘‘Cyperoideae’’ abohemica Schreb. Austria (Niederosterreich-Zwettl) Wallnofer 13755 (WIS) AY779073 DQ460944a,bsychnocephala J. Carey WI (Waushara) Hipp 2578 (WIS) AY779169 DQ461043

‘‘Festivae’’ abrupta Mack. CA (Nevada) Hipp 799 (WIS) AY779064 DQ460934bonplandii Kunth Bolivia (La Paz: Nor Yungas

Province)Solomon et al. 18926 (MICH) AY779074 DQ460946

cf. microptera Mack. Mexico (Durango) Gonzalez & Reznicek 10303(MICH)

AY779085 DQ460957

ebenea Rydb. CO (Ouray) Hipp 1683 (WIS) AY779095 DQ460968gracilior Mack. CA (Mendocino) Hipp 363 (WIS) AY779102 DQ460977harfordii Mack. CA (Marin) Hipp 309 (WIS) AY779104 DQ460980bhaydeniana Olney UT (Summit) Hipp 140.2 (WIS) AY779105 DQ460981aillota L.H. Bailey CA (Alpine) Hipp & Clifton 700 (WIS) AY779110 DQ460986integra Mack. CA (Nevada) Hipp et al. 774 (WIS) AY779111 DQ460987amacloviana D’Urv. WY (Albany) Hipp 1893 (WIS) AY779117 DQ460993mariposana L.H. Bailey CA (Mariposa) Hipp & Clifton 644 (WIS) AY779118 DQ460995amicroptera Mack. CO (Gunnison) Hipp 1681 (WIS) DQ461142 DQ460997orizabae Liebm. Mexico (Ixtapaluca) Rzedowski 36822 (WIS) AY779130 DQ461007pachystachya Cham. ex Steud. UT (Duchesne) Goodrich 21180 (RM) AY779136 DQ461012roraimensis Steyerm. Venezuela (Roraima) Reznicek 11054 (MICH) AY779152 DQ461028stenoptila F.J. Herm. UT (Summit) Hipp 1848 (WIS) AY779161 DQ461036subbracteata Mack. CA (Humboldt) Hipp 448 (WIS) AY779165 DQ461040teneraeformis Mack. CA (El Dorado) Hipp & Clifton 716 (WIS) AY779172 DQ461046

‘‘Festivae’’ or ‘‘Festucaceae’’ multicostata Mack. CA (El Dorado) Hipp & Clifton 714 (WIS) AY779125 DQ461002preslii Steud. MT (Flathead) Lesica 7874 (MICH) AY779146 DQ461022a,bpeucophila T. Holm Mexico (Morelos) Gonzalez & Reznicek 10552 (WIS) DQ461146 DQ461015subfusca W. Boott NV (Washoe) Hipp 833 (WIS) AY779167 DQ461042

‘‘Festucaceae’’ bbebbii (L.H. Bailey) Fernald WI (Dane) Hipp 516 (WIS) AY779071 DQ460942abicknellii Britton bicknellii WI (Dane) Hipp 549 (WIS) AY779072 DQ460943abrevior (Dewey) Mack. ex Lunell WI (Dane) Reznicek 10345b (MICH) AY779075 DQ460947cf. brevior (Dewey) Mack. ex Lunell Mexico (Chiapas) Gonzalez & Reznicek 10497

(MICH)AY779082 DQ460955

‘‘Buffalo River’’ AR (Marion) Hyatt 10461 (MICH) AY779077 DQ460951acrawfordii Fernald ME (Hancock) Reznicek & Reznicek 10918 (WIS) AY779089 DQ460960egglestonii Mack. CO (Grand) Hipp 1594 (WIS) AY779097 DQ460971bfestucacea Schkuhr ex Willdenow WI (Juneau) Hipp et al. 561 (WIS) AY779098 DQ460973hyalina Boott MS (Tunica) Rothrock 2947 (MICH) AY779109 DQ460985amerritt-fernaldii Mack. NH (Strafford) Rothrock 3475 (MICH) AY779119 DQ460996amissouriensis P.E. Rothrock & Reznicek MO (Macon) Rothrock 3567b (MICH) AY779121 DQ460998amolesta Mack. ex Bright MS (Bolivar) Bryson 12209 (MICH) DQ461143 DQ460999amolestiformis Reznicek & P.E. Rothrock TN (Jackson) Rothrock 3729c (MICH) DQ461144 DQ461000a,bnormalis Mack. WI (Dane) Hipp 159 (WIS) DQ461145 DQ461004opaca (F.J. Herm.)

P.E. Rothrock & ReznicekIL (Washington) Reznicek 10856 (MICH) AY779129 DQ461006

oronensis Fernald ME (Penobscot) Reznicek et al. 10931 (WIS) AY779131 DQ461008reniformis (L.H. Bailey) Small AR (Dallas) Hyatt, P.E. 6996 (WIS) AY779151 DQ461027scoparia Schkuhr ex Willd. IN (Newton) Rothrock 3633b (MICH) AY779155 DQ461031scoparia Schkuhr ex Willd. var.

tessellata Fernald & WiegandME (Washington) Reznicek 10923 (WIS) AY779156 DQ461032

ashinnersii P.E. Rothrock & Reznicek TX (Delta) Reznicek 10367 (MICH) AY779157 DQ461033bstraminiformis L.H. Bailey NV (Washoe) Hipp 847 (WIS) AY779164 DQ461039atenera Dewey var. echinodes (Fernald)

WiegandWI (Ozaukee) Hipp & Rothrock 1188 (WIS) DQ461138 DQ460970

atenera Dewey var. tenera OH (Lucas) Hipp 1191 (MICH) DQ461149 DQ461045tincta (Fernald) Fernald ME (Hancock) Rothrock 3734 (MICH) AY779174 DQ461047

‘‘Fetae’’ a,bfeta L.H. Bailey CA (Humboldt) Hipp 457 (WIS) AY779099 DQ460974bhormathodes Fernald ME (Hancock) Reznicek 10929 (WIS) AY779108 DQ460984bstraminea Willd. ex Schkuhr WI (Juneau) Hipp et al. 560 (WIS) AY779163 DQ461038suberecta (Olney) Britton WI (Rock) Hipp & Zimmerman 598 (WIS) AY779166 DQ461041

1045

Table A1

(Continued )

Section/Species group Species Location Collector ITS ETS

‘‘Foeneae’’ badusta W. Boott ME (Washington) Reznicek 10922 (WIS) AY779065 DQ460935arapahoensis Clokey CO (Gunnison) Hipp 1659 (WIS) AY779068 DQ460938bargyrantha Tuck. ex Dewey ME (Washington) Reznicek 10921 (WIS) AY779069 DQ460940foenea Willd. ME (Washington) Reznicek 10928 (WIS) AY779100 DQ460975xerantica L.H. Bailey Ontario (Thunder Bay) Oldham & Bakowsky 17732

(MICH)AY779182 DQ461054

‘‘Fractae’’ a,bfracta Mack. CA (Mariposa) Hipp 635 (WIS) AY779101 DQ460976‘‘Leporinae’’ leporinella Mack. CA (El Dorado) Tallent 815 (MICH) AY779114 DQ460990

bovalis Gooden. New Zealand Ford 30/98 (MICH) AY779132 DQ461009bovalis Gooden. OR (Benton) Wilson & Kuykendall 7027 (WIS) AY779134 DQ461010bphaeocephala Piper UT (Summit) Hipp 135 (WIS) AY779139 DQ461016praticola Rydb. Ontario (Rainy River) Oldham & Bakowsky 21854

(MICH)AY779145 DQ461021

tahoensis Smiley CA (El Dorado) Hipp 879 (WIS) AY779170 DQ461044‘‘Specificae’’ constanceana Stacey CA (Nevada) Hipp et al. 800 (WIS) AY779088 DQ460959

davyi Mack. CA (Alpine Hipp 901a (WIS) AY779093 DQ460965petasata Dewey MT (Gallatin) Morse & Jordan 2082 (MICH) AY779138 DQ461014aspecifica L.H. Bailey CA (El Dorado) Hipp 861 (WIS) AY779160 DQ461035awootonii Mack. NM (Ruidoso) Hyatt 8294 (MICH) AY779181 DQ461053

‘‘Tribuloideae’’ cristatella Britton WI (Rock) Hipp & Zimmerman 606 (WIS) AY779090 DQ460961a,bmuskingumensis Schwein. WI (Iowa) Hipp & Biggs 2009 (WIS) AY779126 DQ461003aprojecta Mack. WI (Adams) Hipp et al. 1206 (WIS) AY779147 DQ461023tribuloides Wahlenb. var. tribuloides WI (Iowa) Hipp 185 (WIS) AY779176 DQ461048

Ovales, unallied cf. lagunensis M.E. Jones Mexico (Durango) Gonzalez et al. 4482 (MICH) AY779084 DQ460956interjecta Reznicek Mexico (Morelos) Zika 15398 (MICH) AY779113 DQ460989maackii Maxim. Japan (Honshu) Kan 8031 (RSA) AY779116 DQ460992

a Sequence has been verified by a second acquisition (data not reported in this study).b Taxon was used to evaluate discontinuity in morphological characters analyzed in this article.

Literature Cited

Ball PW, AA Reznicek 2002 Carex Linnaeus. Pages 254–572 in Flora

of North America Editorial Committee, eds. Flora of North

America north of Mexico. Vol 23. Oxford University Press, New

York.Barker FK, FM Lutzoni 2002 The utility of the incongruence length

difference test. Syst Biol 51:625–637.

Baum DA, R Small, JF Wendel 1998 Biogeography and floral

evolution of baobabs (Adansonia, Bombacaceae) as inferred from

multiple data sets. Syst Biol 47:181–207.

Buckler ES, A Ippolito, TP Holtsford 1997 The evolution of ribo-

somal DNA: divergent paralogues and phylogenetic implications.

Genetics 145:821–832.Catalan P, EA Kellogg, RG Olmstead 1997 Phylogeny of Poaceae

subfamily Pooideae based on chloroplast ndhF gene sequences. Mol

Phylogenet Evol 8:150–166.

Cayouette J, P Morisset 1985 Chromosome studies on natural

hybrids between maritime species of Carex (sections Phacocystis

and Cryptocarpae) in northeastern North America, and their

taxonomic implications. Can J Bot 63:1957–1982.

——— 1992 Hybridization in the genus Carex with special reference

to North America. Bot Rev 58:351–438.

Cooperrider TS, JH Morrison 1967 Lactic-acetic-orcein as chromo-

some stain. Mich Bot 6:176–178.

Darlu P, G Lecointre 2002 When does the incongruence length

difference test fail? Mol Biol Evol 19:432–437.

Dowton M, AD Austin 2002 Increased congruence does not neces-

sarily indicate increased phylogenetic accuracy: the behavior of the

incongruence length difference test in mixed-model analyses. Syst

Biol 51:19–31.

Doyle JJ, JL Doyle 1987 A rapid DNA isolation procedure for small

quantities of fresh leaf tissues. Phytochem Bull 19:11–15.

Dragon JA, DS Barrington Forthcoming East vs. west: monophyletic

clades within the paraphyletic Carex acuta complex, section

Phacocystis (Cyperaceae). In RFC Naczi, BA Ford, eds. Sedges:

uses, diversity, and systematics of the Cyperaceae. Proceedings from

the First International Conference on Sedges, Delaware State

University, June 6–8, 2002. Monogr Syst Bot.Eaton RJ 1959 Additional note on vegetative reproduction in Carex

tribuloides and C. projecta. Rhodora 61:294–295.

Egorova TV 1999 The sedges (Carex L.) of Russia and adjacent

states. Missouri Botanical Garden, St. Louis.

Farris JD, M Kallersjo, AG Kluge, C Bult 1994 Testing significance of

incongruence. Cladistics 10:315–319.——— 1995 Constructing a significance test for incongruence. Syst

Biol 44:570–572.

Ford BA, M Iranpour, RFC Naczi, JR Starr, CA Jerome 2006 Phy-

logeny of Carex subgenus Vignea (Cyperaceae) based on non-

coding nrDNA sequence data. Syst Bot 31:70–82.

Goetghebeur P 1998 Cyperaceae. Pages 141–190 in K Kubitzki, ed.

The families and genera of vascular plants. Vol 4. Springer, Berlin.Halanych KM, JR Demboski, BJ van Vuuren, DR Klein, JA

Cook 1999 Cytochrome b phylogeny of North American hares

and jackrabbits (Lepus, Lagomorpha) and the effects of saturation

in outgroup taxa. Mol Phylogenet Evol 11:213–221.

Hall TA 1999 BioEdit: a user-friendly biological sequence alignment

editor and analysis program for Windows 95/98/NT. Nucleic Acids

Symp Ser 41:95–98.

Hendrichs M, S Michalski, D Begerow, F Oberwinkler, FH Hell-

wig 2004a Phylogenetic relationships in Carex, subgenus Vignea

(Cyperaceae), based on ITS sequences. Plant Syst Evol 246:109–125.

Hendrichs M, F Oberwinkler, D Begerow, R Bauer 2004b Carex,

subgenus Carex (Cyperaceae): a phylogenetic approach using ITS

sequences. Pl Syst Evol 246:89–107.Hipp AL 2004 Molecular systematics of Carex section Ovales

(Cyperaceae). PhD diss. University of Wisconsin–Madison.

——— Forthcoming Phylogeny and patterns of convergence in Carex

section Ovales (Cyperaceae): evidence from ITS and 5.8S sequences.

In RFC Naczi, BA Ford, eds. Sedges: uses, diversity, and systematics

of the Cyperaceae. Proceedings from the First International Con-

ference on Sedges, Delaware State University, June 6–8, 2002.

Monogr Syst Bot.

Hipp AL, JC Hall, KJ Sytsma 2004 Congruence versus phylogenetic

accuracy: revisiting the incongruence length difference (ILD) test.

Syst Biol 53:81–89.Hipp AL, PE Rothrock, AA Reznicek, PE Berry 2006 Chromosome

number changes associated with speciation in sedges: a phylogenetic

study in Carex section Ovales (Cyperaceae) using AFLP data. In JT

Columbus, EA Friar, JM Porter, LM Prince, MG Simpson, eds.

Monocots: comparative biology and evolution—Poales. Rancho

Santa Ana Botanic Garden, Claremont, CA.Huelsenbeck JP, F Ronquist 2001 MrBayes: Bayesian inference of

phylogeny. Bioinformatics 17:754–755.

Kishino H, T Miyata, M Hasegawa 1990 Maximum likelihood

inference of protein phylogeny and the origin of chloroplasts.

J Mol Evol 31:151–160.Kukenthal G 1909 Cyperaceae-Caricoideae. Pages 1–824 in A

Engler, eds. Das Pflanzenreich IV. 20, Heft 38. Engelmann, Leipzig.

Lee CE 2000 Global phylogeography of a cryptic copepod species

complex and reproductive isolation between genetically proximate

‘‘populations.’’ Evolution 54:2014–2027.Mabberley DJ 1997 The plant-book: a portable dictionary of the

vascular plants. Cambridge University Press, New York.

Mackenzie KK 1931–1935 Carex: Cariceae. N Am Flora 18:

169–478.——— 1940 North American Cariceae. Vol 1. New York Botanical

Garden, Bronx, NY.

Maddison DR 1991 The discovery and importance of multiple

islands of most parsimonious trees. Syst Zool 40:315–328.

Maddison WP, DR Maddison 2005 Mesquite: a modular system for

evolutionary analysis. Version 1.06. http://mesquiteproject.org.

Mastrogiuseppe J 2002 Carex Linnaeus section Cyperoideae G. Don.

Pages 331–332 in Flora of North America Editorial Committee, eds.

Flora of North America north of Mexico. Vol 23. Oxford University

Press, New York.Mastrogiuseppe J, PE Rothrock, AC Dibble, AA Reznicek 2002

Carex L. section Ovales Kunth. Pages 332–378 in Flora of North

America Editorial Committee, eds. Flora of North America north of

Mexico. Vol 23. Oxford University Press, New York.

Okuyama Y, N Fujii, M Wakabayashi, A Kawakita, M Ito, M

Watanabe, N Murakami, M Kato 2005 Nonuniform concerted

evolution and chloroplast capture: heterogeneity of observed

introgression patterns in three molecular data partition phylogenies

of Asian Mitella (Saxifragaceae). Mol Biol Evol 22:285–296.

Posada D, KA Crandall 1998 MODELTEST: testing the model of

DNA substitution. Bioinformatics 14:817–818.Reznicek AA 1990 Evolution in sedges (Carex, Cyperaceae). Can

J Bot 68:1409–1432.

——— 1993 A revision of the Mexican members of genus

Carex, section Ovales (Cyperaceae). Contrib Univ Mich Herb 19:

97–136.——— 1996 Carex roraimensis (Cyperaceae) revisited. Novon 6:

419–422.

1047HIPP ET AL.—PHYLOGENY OF CAREX SECTION OVALES

Reznicek AA, PM Catling 1986 Vegetative shoots in the taxonomy ofsedges (Carex, Cyperaceae). Taxon 35:495–501.

Roalson EH, JT Columbus, EA Friar 2001 Phylogenetic relationships

in Cariceae (Cyperaceae) based on ITS (nrDNA) and trnT-L-F(cpDNA) region sequences: assessment of subgeneric and sectional

relationships in Carex with emphasis on section Acrocystis. Syst Bot26:318–341.

Roalson EH, EA Friar 2004 Phylogenetic relationships and biogeo-graphic patterns in North American members of Carex section

Acrocystis (Cyperaceae) using nrDNA ITS and ETS sequence data.

Plant Syst Evol 243:175–187.

Rothrock PE, AA Reznicek 1996 Chromosome numbers in Carexsection Ovales (Cyperaceae) from eastern North America. Sida 17:

251–258.

Rothrock PE, AA Reznicek, LR Ganion 1997 Taxonomy of theCarex straminea complex (Cyperaceae). Can J Bot 75:2177–2195.

Savile DBO, JA Calder 1953 Phylogeny of Carex in the light of

parasitism by the smut fungi. Can J Bot 31:164–174.

Simmons MP, H Ochoterena 2000 Gaps as characters in sequence-based phylogenetic analyses. Syst Biol 49:369–381.

Slowinski JB, CGuyer 1993 Testingwhether certain traits have caused

amplified diversification: an improved method based on a model of

random speciation and extinction. Am Nat 142:1019–1024.Stacey JW 1938 Notes on Carex. XIV. Leaflets of Western Botany 2:

121–124.

Starr JR, RJ Bayer, BA Ford 1999 The phylogenetic position of Carexsection Phyllostachys and its implications for phylogeny and

subgeneric circumscription in Carex (Cyperaceae). Am J Bot 86:

563–577.

Starr JR, SA Harris, DA Simpson 2003 Potential of the 59 and 39 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. Int J Plant Sci 164:213–227.

——— 2004 Phylogeny of the unispicate taxa in Cyperaceae tribe

Cariceae. I. Generic relationships and evolutionary scenarios. Syst

Bot 29:528–544.

Swofford DL 2002 PAUP*: phylogenetic analysis using parsimony

(*and other methods). Version 4.0b10. Sinauer, Sunderland, MA.

Urbatsch LE, BG Baldwin, MJ Donoghue 2000 Phylogeny of the

coneflowers and relatives (Heliantheae: Asteraceae) based on

nuclear rDNA internal transcribed spacer (ITS) sequences and

chloroplast DNA restriction site data. Syst Bot 25:539–565.White TJ, T Bruns, S Lee, JW Taylor 1990 Amplification and direct

sequencing of fungal ribosomal RNA genes for phylogenetics. Pages

315–322 in MA Innis, DH Gelfand, JJ Sninsky, TJ White, eds. PCR

protocols: a guide to methods and applications. Academic Press,

New York.

Whitkus R 1988 Systematics and evolution of the Carex pachy-

stachya complex (Cyperaceae). PhD diss. Ohio State University,

Columbus.Whitkus R, JG Packer 1984 A contribution to the taxonomy of the

Carex macloviana aggregate (Cyperaceae) in western Canada and

Alaska. Can J Bot 61:1592–1607.Yen AC, RG Olmstead 2000 Molecular systematics of Cyperaceae

tribe Cariceae based on two chloroplast DNA regions: ndhF and

trnL intron-intergenic spacer. Syst Bot 25:479–494.

1048 INTERNATIONAL JOURNAL OF PLANT SCIENCES