Plant Molecular Biology42: 45–75, 2000.© 2000Kluwer Academic Publishers. Printed in the Netherlands.

45

Contributions of plant molecular systematics to studies of molecularevolution

E. Douglas Soltis and Pamela S. SoltisDepartment of Botany, Washington State University, Pullman, WA 99164-4238, USA

Key words:angiosperms, land plants, model organisms, phylogenetics, polyploidy, seed plants

Abstract

Dobzhansky stated that nothing in biology makes sense except in the light of evolution. A close corollary, andthe central theme of this paper, is that everything makes a lot more sense in the light of phylogeny. Systematicsis in the midst of a renaissance, heralded by the widespread application of new analytical approaches and theintroduction of molecular techniques. Molecular phylogenetic analyses are now commonplace, and they haveprovided unparalleled insights into relationships at all levels of plant phylogeny. At deep levels, molecular studieshave revealed that charophyte green algae are the closest relatives of the land plants and suggested that liverwortsare sister to all other extant land plants. Other studies have suggested that lycopods are sister to all other vascularplants and clarified relationships among the ferns. The impact of molecular phylogenetics on the angiospermshas been particularly dramatic – some of the largest phylogenetic analyses yet conducted have involved theangiosperms. Inferences from three genes (rbcL, atpB, 18S rDNA) agree in the major features of angiospermphylogeny and have resulted in a reclassification of the angiosperms. This ordinal-level reclassification is perhapsthe most dramatic and important change in higher-level angiosperm taxonomy in the past 200 years. At lowertaxonomic levels, phylogenetic analyses have revealed the closest relatives of many crops and ‘model organisms’for studies of molecular genetics, concomitantly pointing to possible relatives for use in comparative studies andplant breeding. Furthermore, phylogenetic information has contributed to new perspectives on the evolution ofpolyploid genomes. The phylogenetic trees now available at all levels of the taxonomic hierarchy for angiospermsand other green plants should play a pivotal role in comparative studies in diverse fields from ecology to molecularevolution and comparative genetics.

Introduction

During the past 10–15 years, molecular phylogeneticshas dramatically reshaped our views of organismal re-lationships and evolution. This impact has been man-ifested at all taxonomic levels of the hierarchy of life,from the species level (and below) to kingdoms (andabove). New inferences include relationships amongstrains of viruses, such as HIV, the potential sisterrelationship between fungi and animals, and the recog-nition of two divergent groups of prokaryotes, with theArchaea being the sister group of the eukaryotes [260,261]. Phylogenetic analyses have also revealed theclosest relatives of many crops and ‘model organisms’for genetic and molecular genetic studies and point to

possible relatives for use in comparative studies andplant breeding.

At the upper end of the taxonomic hierarchy, mole-cular phylogenies have not only provided insights intothe origin of land plants [12, 107, 127], but havealso clarified relationships among: (1) all green plants[114]; (2) all major groups of extant land plants [126,127, 146, 184, 225]; and (3) major groups of an-giosperms [27, 204, 219, 222]. At the highest level,molecular phylogenies indicate the presence of threemajor groups of life: eubacteria, archaebacteria, andeukaryotes [72, 260, 261]. Revised concepts of re-lationship based on phylogenetic analyses are alsoresulting in revised classifications in many groupsof plants. For example, our concept of the major

46

groups of angiosperms and their interrelationships haschanged dramatically from traditional views [37, 39–41, 242, 243, 245]. Traditional classifications of theangiosperms inaccurately portray some relationshipsand therefore should not serve as the basis for com-parisons in molecular evolutionary studies; instead,recent phylogenetic trees should be consulted [27,164, 204, 219, 222], as should the first attempt at anew classification that is based on these (and other)phylogenetic trees [5].

Although molecular systematists, molecular evo-lutionists, and developmental geneticists potentiallyhave much in common in a broad sense, they oftenare not aware of the ongoing developments and con-troversies in each other’s fields. One general purposeof this paper, therefore, is to review some of the recentdevelopments in the field of molecular systematics.A second purpose is to illustrate the major impactof molecular phylogenetics on plant systematics. Wehope both to direct the nonsystematist to the appro-priate literature so that his/her study organism can bebetter placed in the appropriate phylogenetic contextand to promote the value of phylogenetic thinking inother disciplines.

Why phylogeny matters

Most systematists and evolutionary biologists agreethat phylogenies should be the central underpinningof research in much of biology. For example, it iscritical to place model organisms in the appropriatephylogenetic context to obtain a better understandingof both patterns and processes of evolution. Com-parisons of taxa in a phylogenetic context providethe most meaningful insights into biology [6, 95].Dobzhansky [48] stated that nothing in biology makessense except in the light of evolution. A close corol-lary is that everything makes a lot more sense in thelight of phylogeny [6]. For example, the fact thattomato,Lycopersicon esculentum, is actually embed-ded within a well-marked clade ofSolanumspeciesand is therefore really a species ofSolanumrather thana distinct genus, is a powerful evolutionary statement.This discovery is of great importance to geneticists andmolecular biologists, as well as plant breeders, in thatit reveals the closest relatives ofLycopersicon esculen-tum(now known asSolanum lycopersicon; see below)and points to these relatives as the focal points for de-tailed future studies. Similarly, the phylogenetic infer-ences that the model organismArabidopsis thalianais

actually part of a broadly circumscribed Brassicaceaethat also includes Capparaceae, and that Brassicaceaeare, in turn, part of a well-supported ‘glucosinolateclade’ (a group of plants traditionally considered dis-tantly related) are important to those hoping to testthe applicability of the ABC model of floral develop-ment [16, 17, 34]. If this model holds throughout abroadly defined Brassicaceae, then the closest relativesof the family [199] become the focal point for futurestudies. Robust phylogenetic hypotheses for Asteridae[2, 169] and related studies demonstrating the poly-phyly of Scrophulariaceae [170, 172] similarly laythe foundation for studies of floral development andevolution based on the developmental genetics of thesnapdragon,Antirrhinum majus[52, 188].

Current themes in molecular systematics

Multiple data sets and phylogenetic incongruence

Systematists have become increasingly aware that re-liance on a single data set may result in insufficientresolution or an erroneous picture of phylogenetic re-lationships. As a result, it is now common practice touse multiple data sets (preferably both molecular andnonmolecular) for phylogenetic inference. This aspectof systematics has been facilitated by automated DNAsequencing, which has made the rapid acquisition ofmultiple molecular data sets relatively straightforward.In fact, numerous DNA regions representing the nu-clear and chloroplast genomes are now routinely usedfor phylogenetic inference for plants, and mitochon-drial DNA sequence data are being used more fre-quently [42, 68, 145, 217].

In addition to DNA sequence data and nonmolec-ular data sets, molecular systematists have recognizedfor nearly a decade the phylogenetic potential of struc-tural rearrangements of the chloroplast genome, aswell as deletions and losses of genes and introns [67,217]. There is currently great interest in these typesof characters in that they may be particularly valuablein marking major evolutionary splits and in clarify-ing deep phylogenetic branches where gene sequencedata alone have, to date, provided inadequate resolu-tion. For example, a 30 kb inversion in the chloroplastgenome is shared by bryophytes, but is absent fromfrom all vascular plants except the lycopsids [187].The fact that the lycopsids have the cpDNA structureof the bryophytes suggests that the former representthe sister group to all other vascular plants. Simi-larly, three introns in the mitochrondrial genesnad1

47

andcox2are absent from liverworts, green algae, andother eukaryotes, but are present in mosses, hornworts,and all vascular plant lineages [182–184]. These datatherefore suggest that liverworts are the sister to allother extant land plants. A unique pattern oftrans-splicing for the loss of an intron in the mitochondrialgenenad1 may help elucidate relationships amongsome of the first-branching angiosperms [183]. Spe-cific genes have also been occasionally lost from thechloroplast genome, with the sequence transferred tothe nucleus; such gene transfer events can mark ma-jor clades. For example, the genetufA resides in thechloroplast genome of algae, but occurs in the nucleargenome of all land plants [45].

Although multiple data sets are needed for estimat-ing phylogenetic relationships reliably, different genesmay possess different histories. Consequently, incor-porating multiple data sets into phylogenetic studiesis not a casual undertaking. Essential tasks in theanalysis of multiple data sets include assessing con-gruence between different phylogenetic trees and datasets and ascertaining whether multiple data sets shouldor should not be combined into a single data matrixprior to phylogeny reconstruction.

Three alternatives have been proposed for handlingmultiple data sets in phylogenetic analyses: thecom-binedapproach, theconsensusapproach, and thecon-ditional combinationapproach [106]. Kluge [121] andKluge and Wolf [122], among others, advocate the firstalternative, suggesting that all available data shouldbe combined into a single matrixbeforephylogeneticanalyses. Because the phylogenetic information fromall characters is considered simultaneously in suchanalyses, and conflict between individual characterscan be assessed [133], this method has also been re-ferred to as thetotal evidenceor character congruenceapproach. In contrast, Miyamoto and Fitch [160] arguethat multiple data sets should be analyzed separatelyand the different phylogenetic estimates compared.This method seeks similarities between independentanalyses for phylogenetic corroboration and assessestaxonomic congruence. Conditional combinationin-volves combining data except in those instances inwhich significant heterogeneity exists between datasets and that heterogeneity appears to be attributableto different branching histories [19, 46, 47, 143].

There has been considerable debate regarding theadvantages and limitations of the combined, consen-sus, and conditional combination approaches [19, 47,70, 78, 105, 106, 111, 112, 121, 132, 143, 157,160]. From our standpoint, and that of many others,

the conditional combination method is a very reason-able approach. In some cases, however, it may beappropriate to combine data immediately. For exam-ple, because the chloroplast genome is uniparentallyinherited as a unit and not subject to recombination,multiple cpDNA sequences and restriction sites canbe readily combined [217]. Empirical examples of as-sessing congruence of data sets in plants are providedin several recent papers [111, 148, 211].

Several authors have stressed, and we emphasizethis point again here, that statistical tests for congru-ence may not provide a definitive answer as to whetherit is appropriate to combine data sets [111, 238]. Thatis, even if congruence tests reveal some low levelof heterogeneity between data sets, the investigatormay be justified in combining data sets. This shouldnot be viewed pessimistically with a ‘why bother’ at-titude toward congruence tests. Instead, congruencetests represent a means for exploring data, providingthe information needed to make informed, justifiabledecisions regarding how best to handle multiple datamatrices. Furthermore, while incongruence amongdata sets does present problems for phylogenetic in-ference, incongruence can also provide importantinsights into evolutionary (including molecular evolu-tionary) processes [254]. For example, incongruencebetween a chloroplast and nuclear-gene phylogenymay be indicative of past introgression. One importanttype of introgression involves the chloroplast genomeand a process referred to as chloroplast capture [193].As a result of hybridization and subsequent backcross-ing to one parent, the chloroplast genome of species Acan be moved to species B, which still has the nucleargenome of B; such chloroplast capture events have oc-curred frequently in plants [193]. Lineage sorting canalso lead to discordance between data sets.

Genetic processes, including intragenic recombi-nation, interlocus interactions, concerted evolution,rate heterogeneity among sites, base composition bi-ases, and RNA editing [254], may also lead to dis-cordance between trees. Gene duplication presentsthe problem of identifyingorthologousgenes (genesrelated directly by descent) vs.paralogousgenes(genes resulting from duplications). Unintentionallysampling a mixture of orthologous and paralogousgenes, as well as erroneous assessments of orthologyand paralogy, can lead to incongruence and erroneousphylogenetic interpretations [201, 254].

48

Large groups of organisms and large data sets

The rapid acquisition of multiple molecular data setshas also had important implications for the analy-sis of large data sets. Phylogenetic relationships inmany large groups of organisms have remained enig-matic despite intensive study. Classic examples inplants include relationships among major groups ofgreen plants, and relationships within green algae,land plants, ferns, and angiosperms. A current themein systematics is the analysis of large data sets, in-volving many taxa and many characters (additionalgene sequences as well as nonmolecular characters),to elucidate phylogenetic relationships. As examples,Källersjö et al. [114] analyzed a data set of 2538rbcL sequences representing all major groups of greenplants, and Soltiset al. [219] analyzed a data set ofsequences of three genes (rbcL, 18S rDNA, atpB),providing an aligned matrix of over 4700 bp of se-quence, for each of 560 angiosperm species. Theseare among the largest data sets analyzed phylogenet-ically. Combining multiple data sets has also beenemployed to elucidate relationships in other large,enigmatic groups of green plants, including the ferns[263] and green algae [23]. One noteworthy aspectof most of these analyses of major groups of landplants is that they have often involved numerous col-laborators who have willingly contributed previouslyunpublished data to accomplish broad phylogeneticgoals not easily achieved by one or a few investigators[23, 27, 204, 219, 222]. In many cases, the first authorsof these papers have sought permission (without suc-cess) from publishers and journals for special methodsof citation to give credit to the first three or fourauthors, rather than have these important individualcontributions buried under ‘et al.’. This was the stimu-lus, for example, for the multiauthored reclassificationof angiosperms (discussed below) published under theauthorship of Angiosperm Phylogeny Group (APG)[5]. These large collaborative ventures are a tribute tothe spirit of cooperation that exists among plant sys-tematists and often contrast sharply with systematicefforts in other groups of organisms.

Although the phylogenetic analysis of large datasets is becoming increasingly common, the feasibilityof analysis of large data sets has been much debated[104, 179]. Large data sets pose problems for parsi-mony analyses because of the large number of treesthat must be examined – the number of potential solu-tions increases logarithmically as taxa are added [79].Because of the large number of possible trees for even

a modest number of taxa (e.g. 282 million rooted treesfor 10 taxa [79]), the analysis of large data sets wasconsidered a serious problem. Based on simulationstudies, Hillis et al. [104] suggested that in someinstances correct phylogeny estimation for only fourtaxa would require over 10 000 bp of DNA sequence.This degree of complexity implied much greater dif-ficulty with much larger data sets and stimulated thesuggestion that large phylogenetic problems be brokeninto a series of smaller problems [92, 158, 189].

Despite the dire predictions noted above, recentsimulation studies as well as phylogenetic analysesinvolving hundreds of species of angiosperms haveimportant implications for the analysis of large datasets [103, 221]. Simulation studies [93, 103] indi-cate the importance of adding taxa; additional taxacan break up long branches that would otherwise at-tract each other and can actually make the problemeasier to solve. Empirical studies involving parsimonyand bootstrap [80] analyses of separate and combineddata sets using three genes (a total of 4733 bp) and190 angiosperms support the findings of the simu-lation studies and also demonstrate the importanceof adding more characters. Furthermore, analyses ofthe combined data sets showed great improvementsin computer run times compared to the separate datasets; the combined data sets also have higher inter-nal support for clades [221]. The general conclusionfrom these studies is that more is better: more taxaand more characters. The approach of employing mul-tiple data sets and numerous taxa to infer phylogeneticrelationships in large clades of plants has been ap-plied to angiosperms [24, 25, 27, 204, 214, 219,222], ferns [263], green algae [23], and all greenplants (Green Plant Phylogeny Research CoordinationGroup, ongoing research).

Other solutions to the problems posed by largedata sets involve improvements in computer software.These should not be considered mutually exclusivefrom the addition of characters and taxa, but insteadshould be used in concert. For example, parsimonyanalysis (see the paper by Doyle and Gaut in thisissue for more information on general methods ofdata analysis) of the 560-taxon angiosperm data setfor three genes was greatly facilitated by the com-puter program NONA [87] and the ratchet method(Nixon, unpublished data). These programs quicklyfound trees shorter than those obtained using PAUP∗4.0 [241]. Large data sets can also be analyzed withparsimony jackknifing [76, 77] and the ‘fast’ boot-strap and jackknife options of PAUP∗. A consensus

49

of the trees produced via the bootstrap or jackknifeapproach depictsonly well-supported clades; thesemethods provide a rapid assessment of branch support.

Revised views of plant phylogeny

In this section we present an overview of the con-tributions of molecular systematics to inferences ofphylogeny on a broad scale: land plants, seedplants, and angiosperms. Results of research ongreen plant phylogeny are also available through theGreen Plant Phylogeny Research Coordination Group(http://ucjeps.berkeley.edu/bryolab/greenplantpage.ht-ml). In addition, the Tree of Life provides a summaryof phylogenies (http://phylogeny.arizona.edu/tree/phy-logeny.html), and TreeBASE (http://herbaria.harvard.edu/treebase/) archives both phylogenetic trees andreferences. The latter two web sites do not containall available plant phylogenies but are valuable start-ing points for relevant references and for the names ofsystematists working on various groups of plants.

Land plants: origin and relationships

For many broadly based studies of molecular evolu-tion, phylogenetic hypotheses for all major lineagesof land plants, as well as comparison with the clos-est relatives of land plants, are essential. As examplesof such molecular evolutionary studies, investigatorshave attempted to examine the diversification of genefamilies, including genes encoding heat shock proteins[251], phytochrome [124, 208], and actin [11, 156,161] across land plants.

A wealth of data indicates that the closest rela-tives of land plants are from the charophycean lineageof green algae [23, 90, 127, 184]. These consist ofthe orders Coleochaetales, Charales, Zygnematales,and Klebsormidiales. The charophycean algae andland plants together are monophyletic [11, 12, 23,62, 107, 129, 145, 146, 184]. Which of the charo-phycean algae is the immediate sister group of the landplants is less certain, however. For example,rbcL se-quence data indicate that two orders of charophyceanalgae, Coleochaetales and Charales, are each mono-phyletic groups closely related to the land plants, butthe branching order among Coleochaetales, Charales,and land plants is not well resolved; the remain-ing orders of charophycean algae, Zygnematales andKlebsormidiales, are then the sister to the above [23].In contrast, 18S rDNA sequence data show a grade

of charophyte algae as sister to the land plants. Theimmediate sister group to the land plants in 18S rDNAtrees [12, 23, 107] is a clade of Coleochaetales, Zygne-matales, and Klebsormidiales, with Charales sister toall of the above (Figure 1).

Several studies using gene sequence data (18SrDNA sequences, or 18S rDNA plus partial 26SrDNA), as well as cpDNA rearrangements and geneor intron loss data, have focused on the placement ofspecific groups in the phylogeny of land plants as awhole [68, 98, 107, 126, 136, 159, 181, 184, 187].Recent reviews of attempts to reconstruct phylogenyamong land plants are provided by Kenrick and Crane[119, 120], Doyle [53], and Qiu and Palmer [184].A summary tree that attempts to synthesize these dataand depict relationships among land plants is providedin Figure 2.

Among the land plants (embryophytes), the threegroups of bryophytes (liverworts, hornworts, andmosses) are paraphyletic [53, 119, 120, 159, 184].Both molecular and morphological analyses indicatethat vascular plants (tracheophytes) are a well-definedsubclade among the land plants as a whole. How-ever, the relationships among the bryophyte lineagesare unclear, as is the immediate sister to the vascularplants (tracheophytes). In fact, the studies conductedto date often portray very different patterns of re-lationships among major clades of land plants andtherefore cannot accurately provide a general frame-work of land-plant phylogeny. For example, recentstudies emphasizing bryophytes [20, 21, 98] producedtrees that fundamentally disagree on the relationshipsamong the major lineages of bryophytes. However, arecent survey of the distribution of introns in two mito-chondrial genes suggests that liverworts are the sisterto all other extant land plants [181, 184] (Figure 2), aresult supported by several analyses of DNA sequencedata [136, 159].

Although progress has been made in resolving re-lationships among the clades of vascular plants, manyquestions remain. cpDNA structural data suggest thatthe lycopods represent the sister group to all othervascular plants [187], a result supported by mtDNAsequence data [68]. Considerable progress has alsobeen made in reconstructing phylogeny across theferns and their allies. There is some evidence, bothmolecular and nonmolecular, that suggests thatEquis-teummay be closely related to the ferns [53, 68, 119,120].

Researchers of fern phylogeny have combined datafrom morphology,rbcL, andatpB gene sequences to

50

Figure 1. Phylogeny of green plants based on 18S rDNA sequences. The tree depicted is the shortest tree resulting from a weighted parsimonyanalysis.Cyanophora paradoxaserved as outgroup. Bootstrap values are indicated by branch thickness. The tree depicts a paraphyleticCharophyceae as sister to the land plants (Bryophyta, Pteridophyta, Gymnospermae, Dicotyledonae, and Monocotyledonae). Charophyceaeand land plants= Streptophyta (from Bhattacharya and Medlin [12]; see also Huss and Kranz [168]).

provide a well-supported phylogenetic tree for theseplants. Significantly, several studies have placed theenigmatic vascular plant family Psilotaceae with thefern group Ophioglossales [97, 147, 176, 262, 263].This is noteworthy in that Psilotaceae (consisting onlyof Psilotum and Tmesipteris) have sometimes beenconsidered the most ancient extant land plants due to asuite of morphological characters that resemble thoseof some of the first vascular plants.

Resolution and support of the basal nodes for landplants in general, and the vascular plants in particular,are often weak in trees derived from separate analysesof 18S rDNA andrbcL sequences. Problems in phy-logenetic inference using 18S rDNA at deep levels ofplant phylogeny may result from constraints imposedby the secondary structure of the rRNA [224, 225].Other difficulties in reconstructing phylogeny acrossall land plants, or even vascular plants, may be due toinsufficient characters and taxa. Therefore, addition oftaxa and characters [93, 103, 221] should help to re-solve differences between land plant trees and identifythe immediate sister group of the land plants among

the charophycean algae (although a data set of 100 18SrDNA sequences was not sufficient to do this [225]).

Still another difficulty in reconstructing phylogenyacross all land plants involves the extinction of manyancient lineages. Thus, simply adding living taxa andadditional characters will not necessarily resolve thisfundamental problem. The importance of adding fos-sils to these efforts to reconstruct phylogeny at suchdeep levels has been noted [53–57]. Ultimately, analy-sis of a combined DNA sequence data set, other DNAcharacters (e.g., cpDNA structural changes, gene andgene intron loss), nonmolecular data sets (e.g. mor-phology and anatomy), and fossils (when available)would likely be most informative regarding both thesister group of the land plants and relationships amongmajor groups of land plants. This approach (encour-aged and facilitated by the Green Plant PhylogenyResearch Coordination Group) should provide the bestestimate of plant phylogeny.

51

Figure 2. A composite phylogeny of extant land plants constructedbased on molecular and morphological phylogenetic studies so farconducted (see text). The tree depictsChara and Coleochaeteaspossible sisters to the land plants.

Seed plants

Knowledge of the relationships among the extantlineages of seed plants (cycads,Ginkgo, conifers,Gnetales (Gnetum, Ephedra, Welwitschia), and an-giosperms), is relevant to many investigations. Partic-ularly critical to many studies of molecular evolutionand developmental genetics is an understanding ofthe closest relatives of the angiosperms. However,phylogenetic analyses based on morphological andmolecular data disagree on the relationships amongthe extant lineages of seed plants; a sample of thealternative relationships is given in Figure 3 [53, 54].

We focus briefly on the uncertain position of Gne-tales [54]. Most analyses of morphological data sug-gest that angiosperms and Gnetales are sister groups(Figure 3A-C) (e.g. [36, 53–57, 140]). In contrast, themorphological cladistic analyses of Nixonet al. [166]and Hickey and Taylor [102] nest the angiospermswithin a paraphyletic Gnetales. For example, in Nixonet al. [166], GnetumandWelwitschiaappeared as theclosest relatives of the angiosperms, withEphedrathesister of theGnetum-Welwitschia-angiosperm clade(Figure 3C). These results from analyses of morphol-ogy have given rise to the widely accepted view that at

Figure 3. Phylogenetic relationships among extant seed plants illus-trating some of the morphologically based and DNA-based topolo-gies that have been published to date (see also Doyle [53]). A.Morphology [36, 55, 57]. B. Morphology [55, 57]. C. Morphology[140]. D. rbcL [96]. E. 18S and 26S rRNA [94]. F. cpITS [88]; 18SrDNA [30]; cox3mRNA [145].

least some Gnetales are the closest living relatives offlowering plants.

Most molecular phylogenetic studies indicate, incontrast, that the angiosperms and Gnetales arenotsister groups (Figure 3D-F). The phylogenetic analysisof 500 rbcL sequences of seed plants [27] suggestedto some that Gnetales are sister to the angiosperms,but this result was achieved only by rooting the seedplants with cycads. Other analyses ofrbcL sequences[96] indicate thatGinkgo, conifers, cycads, and Gne-tales are monophyletic and sister to the angiosperms,with Gnetales variously placed as sister to cycadsor as sister to all other living gymnosperms (Fig-ure 3D). Additional molecular phylogenetic analysesinvolving sequences from the nuclear (18S rDNA [30,225]), chloroplast (cpITS [88]), and mitochondrial(cox3 [145]) genomes all suggest that (1) the gym-nosperms are monophyletic, (2) cycads andGinkgoare sister taxa, and (3) conifers and Gnetales are sisters(Figure 3F). At this point, therefore, Gnetales shouldnot be considered asthe extant sister group of theangiosperms; their position is still uncertain.

Clearly, additional research is needed to resolve therelationships among the extant lineages of seed plants.

52

Reconstructing phylogeny among extant seed plants ismade difficult by several factors, including the pos-sibility that the seed plants radiated rapidly in thePaleozoic, the presence of extinct lineages known onlyfrom the fossil record, and long branches to many ex-tant lineages. Some of these same problems also existin reconstructing phylogeny across all land plants (seeabove). Again, we stress the importance of adding fos-sils to the morphological data matrices and combiningmorphological and multiple DNA sequence data sets[53–57]. As noted above for land plants, this approachis currently facilitated by the Green Plant PhylogenyResearch Coordination Group.

Angiosperms

The impact of molecular phylogenetics on the an-giosperms has been particularly dramatic. The mostcomprehensive phylogenetic analyses ever conductedfor any group of organisms have involved the an-giosperms. Källersjöet al. [114] recently analyzed2538 taxa representing all photosynthetic life usingrbcL sequences. Soltiset al. [219] compiled a genesequence data set for 560 angiosperms based on threegenes (ca. 4700 bp per species; the complete dataset and shortest trees obtained are available at theGreen Plant Phylogeny Research Coordination Groupwebsite). Phylogenetic analysis of the latter data sethas provided for the first time a well-resolved andstrongly supported topology for the angiosperms. Asa result, the big picture of angiosperm relationshipshas solidified during just the past few years.

From a historical standpoint, the phylogeneticanalysis of 500rbcL sequences [27] provided thefirst molecular-based hypothesis for all major groupsof angiosperms. Controversy surrounded this analy-sis, however; because of the large number of taxainvolved, no optimal (shortest) tree (or set of trees)was found. Subsequent reanalyses of that data sethave yielded slightly shorter trees [189] and suggestedsome relationships that differ from those portrayedin Chaseet al. [27]. However, these differences areslight, and none are strongly supported [24]. Hence,the overall picture of angiosperm relationships re-mained unchanged following the analysis of Riceet al.[189].

Soltis et al. [222] provided a nuclear-based phy-logenetic hypothesis for angiosperms for comparisonwith the rbcL tree [27]. Using 18S rDNA sequencesand 228 species, their analyses yielded topologies thatlargely agreed with therbcL trees. Although the place-

ment of some taxa differed between the 18S rDNA andrbcL trees, none of these differences was strongly sup-ported, and the overall picture is one of concordance[24, 25, 214].

Phylogenetic trees derived from a second chloro-plast gene (atpB) agreed closely with those obtainedfrom rbcL and 18S rDNA. These multiple, concordantmolecular data sets for angiosperms prompted com-bined analyses of theserbcL, atpB, and 18S rDNAdata sets [204, 214, 221]. As noted above, these analy-ses suggest that one solution to the computationaland analytical dilemmas posed by large data sets isthe addition of nucleotides (and other nonmolecularcharacters) as well as taxa [93, 221].

Nandi et al. [164] constructed a data set for 161species based on morphology, chemistry, and othernon-DNA characters and combined this data set withan rbcL data set for a comparable suite of taxa. Al-though the trees recovered in the analysis of non-DNAcharacters clearly differ in some respects from those ofthe above-noted molecular studies, there are also manypatterns congruent with the trees based on gene se-quence data. This analysis, as well as similar ongoingefforts in ferns [263], green algae [23], and bryophytes[159], illustrate well the current trend of integratingnonmolecular and DNA characters in phylogeneticanalyses.

Phylogenetic analyses of the large data sets of 18SrDNA, rbcL, andatpBsequences have also facilitatedstudies of the molecular evolution of these three genes[24, 25, 117, 173, 224]. Researchers should con-sult these papers as a starting point for informationregarding the molecular evolution of these three genes.

These phylogenetic analyses of angiospems havealso resulted in a reclassification of the angiosperms[5]. This ordinal-level reclassification is perhaps themost dramatic and important change in higher-levelangiosperm taxonomy in the past 200 years. Further-more, this system represents one of the few times todate that a major group of organisms has been re-classified based on explicit phylogenetic trees (anotherexample is provided by molecular phylogenetic stud-ies that indicate the presence of three major groupsof life: eubacteria, archaebacteria, and eukaryotes [72,260, 261]).

The picture of angiosperm relationships that hasemerged (Figure 4) agrees with some facets of tra-ditional classifications [37, 39–41, 242, 243, 245]and disagrees with others. Because of significantdifferences between relationships inferred in recentphylogenetic analyses and those implied by earlier

53

Figure 4. Summary of angiosperm relationships depicting major clades based on phylogenetic analysis of a combined 18S rDNA,rbcL, atpBdata set for 560 angiosperms. Seven gymnosperms served as outgroups. This is a summary tree based on parsimony jackknife analysis [27].Jackknife values≥50% are provided above branches; branches not receiving jackknife support≥50% are reduced to a polytomy (from Soltiset al. [219]).

classifications, long-standing taxonomies should notbe assumed to reflect relationships among these largegroups; caution is also needed in relying on traditionalfamily circumscriptions (see below). Rather, the re-cent, improved views of angiosperm phylogeny andthe Angiosperm Phylogeny Group [5] classificationshould be consulted when questions of broad evolu-

tionary patterns are addressed. This reclassificationof the angiosperms is viewed as an ongoing process;improved or refined classifications will be publishedby the Angiosperm Phylogeny Group in the future asmore data become available.

All major classifications of angiosperms [37, 39–41, 242, 243, 245] recognize a primary divergence

54

between monocots (Class LiliopsidasensuCronguist[37]) and dicots (Class MagnoliopsidasensuCron-quist [37]). However, phylogenetic analyses basedon both molecular and morphological data fail tosupport this division. Instead, most molecular phy-logenetic trees reveal a grade of successive branchesat the base of the angiosperms (Figure 4). Althoughnot monophyletic, these basal angiosperms have, forconvenience, been referred to as the ‘monosulcates’,reflecting the fact that many of these plants possessuniaperturate pollen (pollen with one groove or lineof weakness), the same condition observed in non-flowering seed plants. Most basal lineages are dicots,but the monocots are embedded within this grade. Thebasal grade is then followed by a large, well-supportedclade, referred to as the ‘eudicots’ or true dicots [58];these plants share triaperturate or triaperturate-derivedpollen. Hence, as Figure 4 clearly illustrates,there isno monocot-dicot split in the angiosperms; to makemonocot-dicot comparisons or to discuss molecular orgenome evolution in terms of the monocot-dicot splitis inappropriate. Monocot remains a useful term in thatit refers to a monophyletic group, but the dicots areclearly not monophyletic. More useful comparisonswould involve comparisons of monocots with eudi-cots, or of early-branching angiosperms with lowereudicots and core eudicots (see below).

Among the basal branches of the angiosperm treeare several well-supported clades, including the mono-cots and several groups formerly classified as ‘prim-itive’ dicots, such as Magnoliales, Laurales, andPiperales (Figure 4). The first few branches of theangiosperm tree, based on analyses of both DNA se-quences [219] and phytochrome gene duplication data[150], appear to be Amborellaceae, Nymphaeaceae,Schisandraceae, Illiciaceae, and Austrobaileyaceae, aresult first suggested by analysis of 18S rDNA se-quences [222]. Although other data are needed to sub-stantiate whether these actually arethefirst-branchingextant angiosperms, the possible placement of thesetaxa on the early branches of the angiosperm treenonetheless has major implications. Those investiga-tors comparing developmental genetics or molecularevolution across seed plants or angiosperms should in-clude at least some of these early-branching taxa, suchasAmborella, Illicium, Austrobaileya, andNymphaea.Investigators interested in the molecular evolutionof genes controlling petal and stamen development(APETALA 3and PISTILLATA) have started to ana-lyze basal angiosperms, includingNymphaea, and arecomparing the results to findings for eudicots [125].

Within the large eudicot clade, several groups(e.g., Ranunculales, Trochodendraceae, Platanaceae,Sabiaceae, Proteaceae) form a grade, followed bya well-supported clade of core eudicots. Just as theearly-branching angiosperms are critical for interpre-tations of evolutionary patterns in the angiospermsas a whole, these early-branching eudicots are criti-cal for investigations of morphological and molecu-lar evolution in the core eudicots. The compositionof Proteales (Proteaceae, Platanaceae, and Nelum-bonaceae) illustrates well in microcosm the value ofthese broad phylogenetic analyses in that this has tobe one of the major surprises of molecular phylogenet-ics; these three families were not considered closelyrelated in any classification scheme [37, 39–41, 242,243, 245]. Yet, the combined three-gene analysis pro-vides very strong support for this clade; furthermore,the monophyly of this clade is supported by severalmorphological and anatomical features [163, 204].

Within the core eudicots are several well-supportedmajor clades (Figure 4). These include (1) Saxifra-gales, which are sister to (2) a large rosid clade,and two other large clades: (3) the asterids and(4) the caryophyllids (or CaryophyllalessensuAn-giosperm Phylogeny Group [5]). These rosid, as-terid, and Carophyllales clades are expanded versionsof their counterparts from traditional classifications(e.g. [37]). The DNA-based rosid clade contains,in addition to traditional Rosidae (sensuCronquist[37]), members of subclasses Hamamelidae and Dil-leniidae. Molecular phylogenies also provide strongsupport for an expanded asterid clade that, in addi-tion to traditional Asteridae, includes some traditionalrosids (e.g. Apiaceae, Araliaceae, Pittosporaceae,Cornaceae, Hydrangeaceae), dilleniids (e.g., Eri-caceae, Actinidiaceae, Theaceae, Sarraceniaceae) anda hamamelid (Eucommia). Phylogenetic analyses ofDNA sequences also recover an expanded Caryophyll-idae (Caryophyllales of Angiosperm Phylogeny Group[5]) that includes the traditional Caryophyllales, aswell as Plumbaginaceae and Polygonaceae, plus twofamilies of carnivorous plants (Droseraceae and Ne-penthaceae), as well as several other families of tradi-tional Dilleniidae (e.g. Tamaricaceae, Frankeniaceae).The traditionally recognized subclasses Hamamelidaeand Dilleniidae are polyphyletic and comprise severalphylogenetically distant lineages [24, 25, 27, 214,222]; both terms have been abandoned [5].

Several major subclades within the angiosperms,including the asterids and the monocots, have alsobeen the focus of additional detailed phylogenetic

55

analysis using several different molecular markers.The most detailed analysis of the asterid clade yetconducted employed four genes (atpB, rbcL, ndhF,18S rDNA) and 158 species [2]; this analysis providesthe greatest resolution and support for clades withinthe asterids. All of these studies support the mono-phyly of a broadly defined asterid clade (Asteridaes.l. [169]). Ericales and Cornales are early-branchingclades within the asterids; the monophyly of each ofthese clades is strongly supported. The remaining as-terids also form a well-supported clade that, in turn,consists of two large subclades, euasterid I and eu-asterid II. The morphological and chemical featuresshared by members of this asterid clade, as well asnon-DNA characters supporting euasterid I and II,have been discussed elsewhere [2, 164, 169, 204].

The monocots have also been the focus of a num-ber of phylogenetic analyes [26, 28, 29, 42, 69].Readers are urged to consult Chaseet al. [28], whoanalyzed a combined data set of three genes for 126monocots, for the most thorough phylogenetic analy-sis of monocots so far conducted. The studies con-ducted to date have not identified a clear sister groupto the monocots. However, both Soltiset al. [219] andChaseet al. [28] place the monocots in a large cladein which Chloranthaceae are sister to a grade of mono-cots, Piperales, Laurales, Winteraceae/Canellaceae,and Magnoliales. These two analyses are the first toprovide any measure of internal support for this largeclade, which includes the monocots. The monocots,includingAcorus, form a well-supported clade in thesestudies.

In Chaseet al. [28] and Soltiset al. [219] themonophyly of the remaining monocots, excludingAcorus, is also strongly supported. FollowingAcorus,a strongly supported Alismatales clade appears asthe sister to the remaining monocots. Alismatales,including Tofieldiaceae, Araceae, and a strongly sup-ported subclade of Zosteraceae, Hydrocharitaceae,and Aponogetonaceae are sister to the remainingmonocots. These remaining monocots also form awell-supported clade that, in turn, comprises sev-eral subclades that correspond to the circumscriptionsgiven in Angiosperm Phylogeny Group [5]: commeli-noids, Dioscoreales, Japonoliriaceae/Petrosaviaceae,Pandanales, Liliales, and Asparagales. Relationshipsamong these major subclades of monocots are unclear,however.

The importance of consulting recent phylogeneticanalyses of the angiosperms, as well as the An-giosperm Phylogeny Group [5] classification, is also

evident at the family level. Many well-known fam-ilies are now known to be nonmonophyletic; theyare either polyphyletic or require merger with otherfamilies to form a monophyletic group. Hence, anumber of families require a circumscription that dif-fers fundamentally from traditional views, illustratingagain the problems in relying solely on traditionaltreatments (e.g. [87]). An excellent example of awell-known angiosperm family that is now known tobe polyphyletic is Scrophulariaceae, which containthe model organismAntirrhinum (see below). Sim-ilarly, Brassicaceae in the traditional sense are nowknown to be nested within a paraphyletic Cappa-raceae [197–199]. They have been combined into asingle large Brassicaceae [5, 113], a result of inter-est to those studying the model organismsBrassicaand Arabidopsis(see below). As another example,the distinction of Malvaceae (which containsGossyp-ium, cotton) from the related families Bombaceae,Tiliaceae, and Sterculiaceae has not been supportedby recent phylogenetic studies [4, 10]. These analy-ses suggest, instead, a single broadly defined familyMalvaceae that includes Bombacaceae, Tiliaceae, andSterculiaceae. Other angiosperm families also knownto be nonmonophyletic, requiring revised circumscrip-tions (either splitting into several separate families ormerger into one larger family), include Portulacaceae[101], Santalaceae [165], Sapindaceae [83], Lami-aceae and Verbenaceae [249], Euphorbiaceae, Fla-courtiaceae [219], Escalloniaceae, Chenopodiaceae,Amaranthaceae, Grossulariaceae, and Saxifragaceae[27, 204, 219, 222]. Again, readers are encouraged toconsult the Angiosperm Phylogeny Group [5] reclassi-fication and subsequent updates for these family levelchanges.

Although plant molecular systematists have thegreatest respect for authors of traditional classifica-tion schemes [37, 39–41, 242, 243, 245] and theirenormous contributions to angiosperm systematics,the results summarized above illustrate clearly whyit is no longer appropriate to use these classifica-tion schemes to infer higher-level relationships in theangiosperms. Readers interested in angiosperm rela-tionships should refer to recent phylogenetic trees [27,204, 219, 222] and consult the recent ordinal classi-fication of angiosperms that is phylogenetically based[5].

56

Where do model organisms fit in?

In this section, we point the reader to phylogeneticstudies that encompass many ‘model’ plant species.This is not a comprehensive overview, but it dealswith the best-known model plants that have been thefocus of considerable genetic research; many of theplants discussed are also major crops. Interested read-ers are encouraged to consult the authors of the paperscited below for more recent (perhaps still unpublished)phylogenetic trees.

Arabidopsis, Brassica(Brassicaceae), and themustard oil clade

Several members of Brassicaceae serve as model sys-tems, Arabidopsis for developmental genetics andsome members ofBrassica for genome and poly-ploid evolution (see below and the paper by Wendelin this volume). Phylogenetic inferences for Brassi-caceae and relatives will allow theArabidopsismodelof floral development [16, 17, 34] to be tested fur-ther. A well-supported clade of 15 families, includingBrassicaceae in the broad sense, has been identified(Figure 5), although most of these families are mor-phologically diverse and were not considered to beclose relatives of Brassicaceae in traditional classifi-cations [195–199] (see above). However, the closestrelatives of Brassicaceae are still unresolved.

Phylogenetic analyses [199] also demonstrate thatCapparaceae, long consideredthe closest relative ofBrassicaceae, are paraphyletic relative to Brassicaceaein the traditional sense (Figure 5). That is,CleomeandCapparis are successive sisters to traditional mem-bers of Brassicaceae (Figure 5);Setchellanthus(alsoof Capparaceae) is more distantly related. As a result,Brassicaceae and Capparaceae have been combinedto form a more broadly defined Brassicaceae, andSetchellanthushas been placed in its own family,Setchellanthaceae [5]. Further study is needed to clar-ify the closest relatives of this more broadly definedBrassicaceae.

ArabidopsisGiven the importance ofArabidopsis thalianaas amodel organism, it is surprising that its phylogeneticrelationships have not been thoroughly investigated;only recently has this species been included in a phy-logenetic analysis that encompassed most of its likelyclose relatives [123]. The first phylogenetic analy-sis involving Arabidopsisemployed cpDNA restric-tion sites [181], but sampling of close relatives was

low. Arabidopsishas often been placed in the tribeSisymbrieae; however, phylogenetic trees based oncpDNA restriction site data suggest instead thatAra-bidopsisis part of an expanded tribe Arabideae. ThiscpDNA-based tree further suggests thatCardaminop-sis arenosaand Arabis lyrata are more closely re-lated toA. thalianathan are some other species nowassigned toArabidopsis. O’Kane et al. [168] re-constructed a phylogeny forA. thalianaand severalpresumed close relatives inCardaminopsiswhile in-vestigating the origin of the allotetraploidArabidopsissuecica.

The most detailed and well-supported topology todate forA. thalianaand relatives [123] involved 33taxa and employed sequences of the internal tran-scribed spacers of the nuclear 18S–26S rDNA (ITS-1 and ITS-2). A major result of this study is thatsome species traditionally classified as members of thegenusArabidopsisare not actually closely related toA. thaliana; conversely, several species traditionallyplaced in genera other thanArabidospsisnow appearto be closely related toA. thaliana. Molecular biolo-gists need to be aware of these new insights (see also[168]).

To summarize the work of Kochet al.[123] briefly,an Arabidopsisclade was identified (Figure 6). Theclosest relatives ofA. thaliana (with a chromosomenumber ofn = 5) are several self-compatible specieswith n = 8, recently reclassified intoArabidopsis(andincluding species ofCardaminopsis), such asA. lyrataandA. halleri, followed byA. griffithianaandA. kor-shinskyi. Sister to thisArabidopsisclade is a clade thatincludes some species ofArabis, which is polyphyleticin this analysis,Halimolobus, and Arabidopsis hi-malaica. Future research will benefit greatly from thepublication of phylogenetic trees that not only suggestthe closest relatives ofArabidopsis, but also providebroader coverage of the entire Brassicaceae in thebroad sense.

BrassicaSeveral species ofBrassicaare of considerable eco-nomic importance:B. rapa (turnip), B. junceaandB. nigra(source of the condiment mustard), andB. ol-eracea(collards, kale, cauliflower, broccoli, kohlrabi,Brussels sprouts). All of these species are part ofthe well-knownBrassicatriangle or ‘Triangle of U’[246], which provides an excellent example of thecomplexity of polyploid genomes (Figure 7). Themembers of this triangle have also been the focus ofdetailed mapping of the nuclear genome and studies

57

Figure 5. Phylogenetic relationships among mustard-oil taxa (names in bold) and putative relatives based on analysis of a combinedrbcLand 18S rDNA data set. The tree depicted is one of four shortest trees resulting from parsimony analysis withEuphorbiadesignated as root.Bootstrap values are provided above the branches; branch lengths followed by decay values are provided below the branches. Arrows mark themajor mustard-oil clade and the second, unrelated lineage ofDrypetes(from Rodmanet al. [199]). Note that Brassicaceae and Capparaceaewere still considered distinct families in Rodmanet al. [199], whereas now they are combined into a single, broadly defined family Brassicaceae(see text and Angiosperm Phylogeny Group [5]).

58

Figure 6. Relationships ofArabidopsis thalianabased on ITS sequence data. The tree depicted is one of 12 shortest trees resulting from aparsimony search usingBrassica napusas outgroup (not shown). Bootstrap support≥50% is provided below branches and branch lengthsabove branches. Weakly supported clades are highlighted by a broad line (from Kochet al. [123]).

Figure 7. Triangle of U [246], depicting relationhips among cul-tivated diploid and allotetraploid species ofBrassica. The trianglehas been updated to show the maternal parents of each allotetraploid(indicated by arrows). The origin ofB. napusis complex, involvingmultiple origins with reciprocal maternal parentage.

of genome organization and evolution [85, 130, 131,185, 186]. Phylogenetic information has contributed tonew perspectives on polyploid evolution (see below).This model system is based on three diploid species:B. rapa (A genome),B. oleracea(C genome), andB. nigra (B genome), and their allopolyploid deriva-tivesB. juncea(AB), B. napus(AC), andB. carinata(BC). Numerous lines of evidence have confirmed theTriangle of U [75, 178]; restriction site analyses havealso provided information on the maternal parent ofeach allotetraploid [227, 228] (Figure 7).

Warwick and Black [250] have helped to place theparental diploids of the Triangle of U in the appro-priate phylogenetic framework. Two of the diploids,B. oleraceaandB. rapa, are sister taxa within a larger‘Rapa/Oleracea’ lineage. However, the third diploid,B. nigra, is distantly related to the Rapa/Oleracea

59

lineage and appears as part of a second clade of bras-sicas (the ‘Nigra’ lineage [250]). This informationis useful in interpreting observations on cytoplasmic-nuclear interactions in the allopolyploids, discussedbelow. The phylogenetic tree of Warwick and Black[250] also places other genera, such asSinapisandRaphanus(which includes radish), withinBrassica.As in the case of the tomato (Lycopersicon), whichis derived from withinSolanum(see below), membersof Sinapisand Raphanusappear to be derived fromwithin a more speciose genus.

Mustard-oil glucosides, or glucosinolates, areoxime-derived sulfur-containing compounds; thebreakdown products of these compounds are respon-sible for the pungent principles of mustard, radish,and capers [81]. The hydrolysis of glucosinolates tomustard oils is catalyzed by myrosinases, a familyof thioglucoside glucohydrolases concentrated in spe-cial myrosin cells. The biosynthesis of glucosinolatesand subsequent catalysis by myrosinases has been ofgreat interest to molecular biologists (see paper byRasket al., this volume). The feeding of herbivoresis thought to bring enzyme and substrate into contact,releasing the toxic mustard oils [71, 198].

While this ‘mustard oil bomb’ [142] is best knownin Brassicaceae in the broad sense (which includesCapparaceae), including the model organismAra-bidopsis, it occurs as well in 15 other angiospermfamilies (Akaniaceae, Bataceae, Bretschneideraceae,Setchellanthaceae, Caricaceae, Putranjivaceae (for-merly part of Euphorbiaceae), Gyrostemonaceae,Koeberliniaceae, Limnanthaceae, Moringaceae, Pen-tadiplandraceae, Resedaceae, Salvadoraceae, Tovari-aceae, and Tropaeolaceae). Traditional classifications[37, 242, 243, 245; but see 39–41] place these 16 fami-lies in several widely separate orders, implying that theglucosinolate-myrosinase defense system arose sev-eral times in the angiosperms. This view, in turn, hasinfluenced the study of host fidelity and evolution byherbivores and pathogens adapted to glucosinolate-producing plants [32, 198, 199].

Rodman and coworkers have addressed the rela-tionships of mustard oil plants [194–198]. Phyloge-netic analysis of sequences of the chloroplast generbcL [198] and the nuclear 18S rRNA gene [199] con-cur in demonstrating the presence of a single majorclade of mustard-oil plants comprising 15 of the 16families noted above, with one phylogenetic outlier,Drypetes(Putranjivaceae, formerly part of Euphor-biaceae). Phylogenetic analysis of a combinedrbcLand 18S rDNA data set yields strong support (Fig-

ure 5) for the dual origins of glucosinolate biosynthe-sis, once inDrypetesand once in the common ancestorof the other 15 families. In short, the mustard oil bombwas invented twice [199].

Antirrhinum (snapdragon)

The snapdragon,Antirrhinum majus (Scrophulari-aceae), represents one of the best model systems forthe study of floral ontogeny and developmental ge-netics [18, 22, 209, 210]. Much of this research hasfocused on the MADS box gene family, members ofwhich specify organ identity during floral morphogen-esis [9, 16, 17, 34].Antirrhinum majushas also been amodel for the study of genes involved in the develop-ment of zygomorphic flowers [3, 141]. Several papershave discussed the importance of understanding thephylogenetic relationships ofAntirrhinum to clarifythe process of floral diversification [52, 73, 74, 188].

Although Antirrhinum has long been placed inthe family Scrophulariaceae, molecular phylogeneticstudies have revealed that Scrophulariaceae are poly-phyletic (Figure 8), comprising at least three distinctlineages [170, 172]. One lineage contains the parasiticmembers of the family (Pedicularis, Castilleja), aswell as the family Orobanchaceae, which is also para-sitic. A second lineage containsVerbascumandScro-phularia (and hence must ultimately retain the nameScrophulariaceae). A third lineage contains manywell-known ‘scrophs’, includingDigitalis, Veronica,and genera that have typically been placed in their ownfamilies (PlantagoandCallitriche); this last clade alsocontainsAntirrhinum.

Recent research in the genetics of floral develop-ment has focused on distantly related angiosperms,most notablyArabidopsisandAntirrhinum. One of thenext challenges in evaluating current models of floraldevelopment is the intensive study of a clade whosemembers exhibit extensive floral variation. Donoghueet al.[52] and Reeves and Olmstead [188] have extrap-olated from what is known about the genetics of floraldevelopment inAntirrhinumto other asterids, doing soin a phylogenetic context. These studies thus provideimportant examples of the value of molecular system-atics in framing hypotheses of floral development andevolution.

Antirrhinum majus is part of a well-supportedclade that contains other species with highly divergentfloral morphologies [188]; in fact, the members ofthis clade were not previously thought to be closelyrelated. In addition toAntirrhinum, with showy, zy-

60

Figure 8. Phylogenetic relationships among Scrophulariaceae based on analysis of a combinedrbcL, ndhF, rps2data set. The tree depicted isone of 80 shortest trees resulting from parsimony analyses withBorago, Nicotiana, andGentianaas outgroups. Branch lengths are providedabove branches, bootstrap values below branches. A star indicates that a node collapses in the strict consensus tree. Numbers correspond toproposed names provided. From Olmsteadet al. [170].

61

gomorphic flowers, the clade also containsPlantago,which has actinomorphic, wind-pollinated flowers, aswell as several enigmatic aquatic plants,HippurisandCallitriche. Hippuris is missing one perianth whorl;three whorls are missing inCallitriche, and the plantsare monoecious.

Using their hypothesis of phylogenetic relation-ships (Figure 8), Reeves and Olmstead [188] note thevalue of the topology for inferring the changes in floralmorphology that have evolved from anAntirrhinum-like ancestor. They also use these derived floral condi-tions observed inPlantago, Hippuris, andCallitricheto pose hypotheses of floral evolution that can betested using the ABC model of floral organogenesis.For example, inHippurisa three-whorled floral goundplan has evolved; the floral organs associated with oneperianth whorl have been lost. It is unclear, however,whether the perianth organs that remain are sepals orpetals: either whorl could have been lost in the courseof evolution. These hypotheses can be tested [188]: be-cause B class MADS box genes are expressed in petalsand not sepals, potential insights could be garneredvia in situ hybridization studies to detectHippuris or-thologues of B class genes fromAntirrhinum(such asDEFICIENSandGLOBOSA).

Hypotheses regarding the evolution of zygomor-phy within asterids can also be addressed [52]. Phy-logenetic trees for asterids imply that zygomorphic(bilaterally symmetrical) flowers have evolved sev-eral times from actinomorphic (radially symmetrical)flowers and that there have been several reversals toactinomorphy. Mutations in the genesCYCLOIDEAandDICHOTOMAresult in the development of radi-ally symmetrical flowers inAntirrhinum[22, 52, 188].Although it is tempting to speculate that mutations inthese genes have resulted in many of the actinomor-phic flowers present in Asteridae, Donoghueet al.[52] caution that there are likely several mechanismsby which reversals to actinomorphy have occurred. Itwill be interesting to determine whether orthologuesof CYCLOIDEAand DICHOTOMA are involved inthe production of at least some of the actinomorphicflowers of Asteridae s.l.

Gossypium (cotton)

Gossypium(Malvaceae) encompasses about 50 species,including four of considerable economic importance(the cultivated cottons). To investigate cotton fiberevolution and disease resistance, genomic maps havebeen constructed to determine the chromosomal lo-

cation of QTLs [110, 264]. Seelananet al. [211]explored phylogenetic relationships within the entiretribe to which Gossypiumbelongs using both nu-clear and chloroplast gene sequences (see also [129]).The closest relatives ofGossypiumidentified by theseanalyses (Kokia and Gossypioides) were unexpectedbased on previous morphological studies.

Molecular data have provided critical insights intorelationships withinGossypiumand the origin of theallotetraploid members of the genus [253, 255–257](Figure 9). Most species ofGossypiumare diploid,and previous studies recognized eight distinct diploidgenome groups (A through G, and K). Phylogeneticanalyses suggest that the diploid genomes are dis-tinct lineages (Figure 9). The origins of the NewWorld allotetraploid species have long been of in-terest. Molecular analyses also indicate that all al-lotetraploids inGossypiumshare a common ancestryand combine an Old World A genome and a NewWorld D genome, suggesting that polyploid forma-tion may have occurred only once. These inferencesare now supported by over 10 kb of sequence perspecies [257]. Furthermore, all allopolyploids in thegenus contain Old World (A-genome) chloroplast andmitochondrial genomes, indicating that the maternalparent (cytoplasmic donor) for the allopolyploids (ADgenome) was an A-genome diploid [255, 257]. Mole-cular data also suggest that the allopolyploids origi-nated relatively recently, perhaps within the last 1–2million years. Comparisons of nuclear and chloroplastDNA phylogenies demonstrate frequent episodes ofhybridization between species and also indicate thatthe genetic architecture of the allotetraploidG. hirsu-tum, the most important species of cotton commer-cially, has also been affected by hybridization andintrogession [257].

Grasses (e.g. maize, wheat)

The grass family, Poaceae, includes a number of plantsof major economic importance, as well as species thatare important genetic and evolutionary models. Re-cently, for example, the family has served as one focalpoint for comparative genetic studies involving the or-ganization of genes within and among genomes [84,85].

At a higher taxonomic level, the closest rela-tives of Poaceae have been controversial. Severalrecent molecular phylogenetic studies [26, 28, 29,61, 69, 118, 229] point to Joinvilleaceae, followedby Restionaceae, as perhaps the closest relatives of

62

Figure 9. Evolutionary history of diploid and allopolyploidGossypium. Phylogenetic relationships of the major groups of diploid cotton areshown, as inferred from multiple molecular data sets from the chloroplast and nuclear genomes, using other genera in the tribe Gossypieaeas phylogenetic outgroups. Each diploid clade (alln = 13) is congruent with classically recognized ‘genome groups’ that have been assigneddesignations (A through G and K) based on chromosome size variation and meiotic pairing behavior in interspecific hybrids. 2C genome sizesare shown in circles for each genome group. Outgroup analysis indicates that Australian diploids (C-, G-, K-genomes) are basal within thegenus. A-genome species are African-Asian whereas D-genome species are endemic to the New World subtropics, primarily Mexico. Based onmolecular dataGossypiumis suggested to have originated ca. 10–20 million years ago. After trans-oceanic dispersal of an A-genome taxon to theNew World, hybridization between the immigrant A-genome taxon and a D-genome taxon led to the evolution of the New World allopolyploids(AD-genome). This reunion among two diploid genomes (in an A-genome cytoplasm) permitted the evolution of novel and agronomicallyimportant fiber traits. Arrows are shown as originating from the taxa most often implicated in allopolyploidization, i.e.,G. herbaceumandG. raimondii. Subsequent to allopolyploid formation, the allopolyploids radiated into the three lineages shown, which include the commerciallymost important species,G. hirsutum(upland cotton) andG. barbadense(pima cotton). Sequence divergence data and the palynological recordsuggest that the A-genome and D-genome groups diverged from a common ancestor 5 to 10 million years ago, and that the two diverged diploidgenomes became reunited in a common nucleus, via allopolyploidization, in the mid-Pleistocene, or 1 to 2 million years ago [257].

Poaceae. The recent phylogenetic analyses of 126monocot species for three genes (rbcL, atpB, 18SrDNA) [28] reveal a clade that includes the grassesand Anarthriaceae, Ecdeiocoleaceae, Restionaceae,and Flagellariaceae (Joinvilleaceae were not includedin the analysis). These families, together with Xyri-daceae, Typhaceae, Cyperaceae, Juncaceae, Thur-niaceae, Mayacaceae, and Sparganiaceae, form awell-supported Poales. Readers should consult Chaseet al. [28] and Angiosperm Phylogeny Group [5] for arecent reclassification of the monocots.

Several phylogenetic studies within the Poaceaehave also been published [8, 33, 38, 50, 94, 115, 162,229]. Perhaps the best resolved and supported phy-logenies to date are those of Clarket al. [33] basedon ndhF sequence data and Soreng and Davis [229]based on both cpDNA restriction sites and non-DNAcharacters. As a result of these studies, the phylogenyof the family is becoming well understood. Not only

do phylogenetic analyses [33, 229] reveal two pri-mary subdivisions within Poaceae (the ‘BOP’ clade ofbambusoids, orzyzoids, and pooids versus the ‘PACC’clade of panicoids, arundinoids, chloridoids, and cen-tothecoids), but the topologies also place in a verygeneral sense the economically important genera suchasOryza (rice), Zea (maize),Hordeum(barley), andAvena(oats).

Phylogenetic trees for Poaceae have served thedual purpose of resolving relationships and clarify-ing the evolution of various characters, thus illus-trating one of the themes of our review. A particu-larly useful study is that of Soreng and Davis [229]who examined the evolution of a number of features,including morphological and anatomical characters,self-incompatibility, and chromosome numbers. In ad-dition, phylogenetic hypotheses for Poaceae have beenused to investigate the evolution of C4 photosynthesisin the family. Although C4 photosynthesis is thought

63

to involve complex anatomical and metabolic changes,mapping the evolution of this trait onto existing phy-logenies [8, 33, 229] suggests that the pathway hasarisen at least four times in the grass family alone[212]. In addition, Poaceae have been the focus ofconsiderable effort for genome mapping and compar-ative genetic studies. Maps of the nuclear genomes ofwheat, barley, rye, oats, maize, sugar cane, sorghum,rice, pearl millet, and other grasses are now avail-able, with others in progress [84, 85]. Phylogenetichypotheses afford the opportunity to place studies ofgenome evolution in the proper context. Kellogg [115]has attempted to do just this and suggests possibleevolutionary patterns of genome rearrangement in thefamily.

One tribe of Poaceae, the Triticeae, includes wheat,barley, rye, and several hundred other related speciesof grasses. Wheat and barley have been of partic-ular interest in terms of their origins and genomeevolution; other members of the tribe are importantsources of germplasm. A clear picture of phyloge-netic relationships within the Triticeae would point towild taxa that may serve as sources of novel genesfor cultivated members [115, 116]. Comparing genetrees among diploid Triticeae based on sequencesrepresenting three nuclear genes, as well as cpDNArestriction sites, Kellogget al. [116] found four dif-ferent phylogenetic histories for diploid members oftribe Triticeae. Several possible explanations for theincongruence were noted, one of which is repeatedhybridization and introgression.

Molecular systematic studies have clarified phy-logenetic relationships withinZea [49, 51]. Theseanalyses support the division ofZeainto two sections,ZeaandLuxuriantes, and agree with the theory thatmaize,Zea maysssp. mays, is a domesticated formof teosinte; these data further suggest thatZ. maysssp. parviglumis is the ancestor of maize. Other re-cent studies have provided compelling evidence thatmaize is an allopolyploid, having undergone extensivechromosomal rearrangement. The pattern of sequencedivergence among duplicated genes suggests that themaize genome is the product of a segmental allote-traploid that formed about 11.4 million years ago [86,258].

Helianthus (sunflower)

Helianthus, a North American genus of approximately49 species, contains the sunflower,H. annuus, aspecies of economic importance as an oil crop.He-

lianthus annuusand other annual species have alsoserved as models for genetic and evolutionary studiesof hybridization and introgression [99, 190–192, 194](see paper by Rieseberget al., this volume). Phylo-genetic trees of the entire Asteraceae, the family towhich Helianthusbelongs, have been estimated mostrecently by Jansen and Kim [108]. Phylogenetic analy-ses ofHelianthusand putative close relatives, basedon cpDNA restriction sites and ITS sequences [205–207], suggest a close relationship betweenHelianthusandPhoebanthus. Based on the geographic distribu-tions of the closest relatives ofHelianthus, Schillinget al. [206] suggest that the genus likely originated inMexico, with subsequent migration throughout NorthAmerica. Although the ITS and cpDNA topologiesagree on some relationships, they differ in other in-stances. The annual species of sectionHelianthusform a clade that in turn comprises several weaklysupported subclades; one of these subclades containsH. annuus, as well asH. argophyllusand H. bolan-deri. Several species ofHelianthusare known to be ofhybrid origin [190–192, 194]. In some instances in theITS analyses, these hybrid species exhibit diagnosticbases of one or the other parent or are polymorphic(these sequences were termed ‘intermediate’ types bySchilling et al. [206]); these hybrid species weresometimes placed in an unresolved phylogenetic po-sition, whereas in other cases they were placed withone or the other parent [206]. These results furtherillustrate the need for extreme caution in the phylo-genetic analysis of groups that include hybrids andhybrid species, as stressed by McDade [152–155] andRieseberg [191].

Legumes (e.g. soybean,Phaseolus, Pea) andsymbiotic nitrogen fixation in angiosperms

Numerous legumes are of economic importance, andsome have been studied intensively in terms of geneticmapping and genome evolution. These include mem-bers ofArachis, Glycine, Lens, Phaseolus, Pisum, andVigna [15, 212, 252]. Phylogenetic analyses of theFabaceae have been conducted at several different lev-els although, due to the enormous size of the family(the third largest angiosperm family), much work re-mains to be done. At higher levels, the relationshipsof the family have long been debated. However, broadphylogenetic analyses of angiosperms point to Poly-galaceae+ Stylobasiaceae as the sister group to theFabaceae [27, 204, 219, 222] (Figure 10).

64

Figure 10. Phylogenetic relationships among taxa that engage innodular nitrogen-fixing symbioses (the nitrogen-fixing clade) basedon analysis of a combined 18S rDNA+ rbcL+ atpBsequence dataset for 567 taxa [219]. The tree depicted shows only the well sup-ported nitrogen-fixing clade and relationships within that clade. Allfamilies having representatives engaged in nodular nitrogen-fixingsymbioses are indicated by an asterisk. In Celtidacae, nitrogen fix-ation is known to occur only inParasponia(a member of what wasonce subfamily Celtoideae of Ulmaceae, now recognized as a sepa-rate family, Celtidaceae). Celtidaceae are represented by the relatedgeneraCeltis and Trema. Gunnera, which hosts nitrogen-fixingcyanobacterial symbionts in leaf glands rather than root nodules,is not closely related to the nitrogen-fixing clade. Values above (orsometimes below with arrows) branches are jackknife values.

Phylogenetic analyses of the Fabaceae suggest thatof the three subfamilies, Mimosoideae and Papil-ionoideae are probably monophyletic whereas Cae-salpinioideae are paraphyletic [59, 63]. These phylo-genetic analyses also place, in a general sense, severalimportant crop genera, includingArachis, Medicago,Glycine, Vigna, and Pisum [63]. Other noteworthyphylogenetic studies of genera or tribes of legumesinclude those of Sanderson and Wojciechowski [202]

and Lavinet al. [134]. Phylogenetic estimates for thefamily have also been used to analyze the evolution ofnitrogen-fixing symbioses; these studies suggest thatroot nodulation has arisen several times independentlywithin the family [60, 63].

At lower levels, several phylogenetic studies havefocused on the evolution of genera containing cul-tivated legumes, includingGlycine [64–66], Vigna[247], Phaseolus[43, 44, 62],Lens[151], andPisum[177]. The results of these phylogenetic analyses aregenerally consistent with patterns of morphologicalvariation. The clades retrieved and relationships de-picted in these studies should be of value in breed-ing programs; the trees are also of potential valueto those pursuing analyses of molecular evolutionin these plants and their relatives. For example, acpDNA-based phylogenetic tree forPisum supportsthe idea that the cultivated pea,Pisum sativum, wasdomesticated primarily from northern populations ofP. humile[177]. Evidence was also obtained for sec-ondary hybridization involving cultivatedP. sativumand eitherP. elatiusor southern populations ofP. hu-mile; this hybridization may have occurred during thedomestication process [177].

The legumes are the best known of the fami-lies that form symbiotic associations with nitrogen-fixing bacteria, but nitrogen-fixing symbioses involv-ing root nodules also occur in some (but not all)members of Celtidaceae (formerly considered part ofUlmaceae), Betulaceae, Casuarinaceae, Elaeagnaceae,Myricaceae, Rhamnaceae, Rosaceae, Datiscaceae,and Coriariaceae [7, 162]. Nodules are induced andinhabited by either of two very different groups of bac-teria. Species of Rhizobiaceae nodulate the legumes[60] andParasponia(Celtidaceae), whereas species ofFrankianodulate hosts in the remaining eight families,plants referred to collectively as actinorhizal [1].

Traditional classifications suggest that many of the10 ‘nitrogen-fixing’ families are distantly related [37,39–41, 242, 243, 245], implying that nitrogen-fixingsymbioses evolved independently several, if not many,times [7]. This view has in turn influenced attitudesregarding the likelihood of transferring genes respon-sible for symbiotic nitrogen fixation to crop speciesthat do not possess this ability [233]. In other words,the great phylogenetic distance these traditional clas-sifications suggested among host plants suggested thatthe bacterial symbionts can adapt to a wide range ofgenetic backgrounds.

Understanding the relationships among these 10angiosperm families is not only of interest from an

65

evolutionary standpoint, but is also of critical impor-tance to those interested in elucidating the complexprocess of nodulation (see paper by Gualtieri and Bis-seling, this volume). Phylogenetic analyses ofrbcLsequences first suggested that representatives of all10 families with nitrogen-fixing symbioses actuallygroup together, with several families lacking this asso-ciation (e.g. Begoniaceae, Cucurbitaceae), in a single‘nitrogen-fixing clade’ [220]. A more recent studyinvolving 560 angiosperms and sequences for threegenes also reveals the same ‘nitrogen-fixingclade’, butwith high internal support [219]. Hence, phylogeneticstudies indicate that only one lineage of closely relatedtaxa achieved the underlying genetic architecture nec-essary for symbiotic nitrogen fixation in root nodules(Figure 10). Increased sampling of taxa from withinthe nitrogen-fixing clade and additional phylogeneticanalyses suggest that within this single well-supportedclade, nitrogen-fixing symbioses may have evolvedmore than once [239]. As noted above, phylogenetichypotheses for Fabaceae suggest, in fact, that mul-tiple origins of this symbiotic relationship may haveoccurred just within the legume family alone [60, 63].

Solanaceae (tomato, potato, and tobacco)

The family Solanaceae contains a number of well-studied plants, as well as species of considerableeconomic importance, inlcudingPetunia, Nicotiana(tobacco),Lycopersicon esculentum(Solanum lyco-persicon; tomato),Solanum tuberosum(potato),Cap-sicum (peppers), andPhysalis. The family has alsoserved as a model for the study of incompatibility (seepaper by Richman and Kahn, this volume) and for nu-clear genome organization and evolution [244]. Thereare several new genetic maps for Solanaceae: one forCapsicum[139] and one forSolanumsectionEtubero-sum[180], the immediate sister to the clade of tomatoand potato.

Initial attempts have been made to reconstructphylogeny across this large family using 17 repre-sentatives and multiple genes [174]. These analysessuggest, for example, thatSolanum(which includesLycopersicon; see below),Capsicum, Datura, andPhysalisform a well-supported clade, whereasPetu-nia is an early-branching member of the family.

At lower taxonomic levels within Solanaceae, a se-ries of molecular phylogenetic analyses has examinedthe relationships of potatoes and tomatoes [13, 171,230–232]. Tomato, formerly classified asLycopersi-con esculentum, is clearly embedded within the large

genusSolanumand therefore does not represent a dis-tinct genus (Figure 11).Lycopersicon esculentumhasbeen formally transferred toSolanumasSolanum ly-copersicon[230]. Molecular systematic studies furthersuggest that species formerly placed inLycopersiconare members ofSolanumsectionPotatoe; in fact, asister-group relationship of potatoes and tomatoes issuggested [13, 171, 230]. These phylogenetic resultsalso help place other data in a new light. For example,earlier genetic linkage studies of tomato and potatohad suggested a close relationship [14]. The phyloge-netic trees demonstrate that potato and tomato sharevery similar linkage maps because they share a recentcommon ancestor. This phylogenetic information alsosets the stage for further studies of genome evolutionin Solanum.

Pinus (pines)

Most model plant species are angiosperms, but someconifers are notable exceptions, having attracted greatinterest because of their considerable economic impor-tance. The genusPinus (the pines) includes a largenumber of economically important species. Phylo-genetic analyses of restriction sites (cpDNA [237]and nuclear [89]) support the morphologically baseddivision of the genus into two well-separated sub-groups, subgenusPinus(the hard pines) and subgenusStrobus(the soft pines). Strauss and Doerkson [237]maintained that the high molecular similarity of taxawithin each of these two subclades suggested a rela-tively recent radiation. More recently, a phylogeneticanalysis of restriction sites and rearrangements withinthe chloroplast genome has helped to elucidate re-lationships within subgenusPinus [128]. This treereveals a major division within subgenusPinus be-tween the North American and the Eurasian species.Many of the clades detected correspond to specific ge-ographic areas, suggesting a biogeographic hypothesisfor cladogenesis [128]. This study also reveals a num-ber of well-supported clades and helps place severaleconomically important species, includingP. taeda,P. contorta, andP. ponderosa. The strongly supportedDNA phylogeny calls into question portions of the tax-onomic scheme of Little and Critchfield [137]; thisis another example of how traditional classificationsmust be used with caution.

Other phylogenetic analyses have considered therelationships ofPinus on a broader scale, providinghypotheses of phylogenetic relationships of Pinaceae,as well as other gymnosperm lineages [30, 234].

66

Figure 11. Relationships of tomato (Lycopersicon esculentum= Solanum lycopersicon) and potato (S. tuberosum) based on analysis of cpDNArestriction site data; depicted here is the strict consensus tree for species ofSolanum(outgroups not depicted). The three major clades identified(Clades I, II, II) are indicated by brackets; ? indicate sections unassigned to subgenus. Bootstrap values are indicated above branches (fromOlmstead and Palmer [171]).

For example, partial 26S rDNA sequences showeda monophyletic conifer clade with Pinaceae the firstbranching family, being the sister to the remainingconifers; Taxaceae are nested within the conifer clade,closely related to Cephalotaxaceae [234].

Polyploidy

Polyploidy is a major force in plant evolution. Perhaps70% of all angiosperms have experienced one or more

episodes of polyploidy [91, 149, 235, 236]; the fre-quency of polyploidy in pteridophytes (ferns and theirallies) could be as high as 95% [91]. Many crop plantsare of polyploid origin, including wheat, maize, cot-ton, and apples. Some of these same crops also serveas models for developmental and genetic studies, asdiscussed above. However, the fact that some studyorganisms are polyploid is sometimes overlooked bynonsystematists and systematists alike. Nonetheless,polyploid genomes offer special challenges and op-portunities to those studying molecular evolution and

67

developmental genetics. Molecular systematics hasplayed a pivotal role in elucidating the parents ofpolyploids, but perhaps more importantly, molecu-lar approaches have dramatically reshaped views ofpolyploid evolution. The topic of polyploidy is con-sidered in more detail in the paper by Wendel; we willtherefore only cover several pertinent issues here.

Perhaps one of the more significant recent dis-coveries involving polyploid species is that many arepolyphyletic, having reformed recurrently from differ-ent populations of their progenitors [215, 216, 218].In addition, these multiple origins can involve, in part,reciprocal origins with different progenitor speciesserving as the maternal and paternal parent, respec-tively. Reciprocal parentage can result in differingmorphologies as well as differing nuclear-cytoplasmicgene interactions. For example, the allotetraploidTragopogon miscellus(Compositae) has formed mul-tiple times from the diploid parentsT. dubiusandT. pratensis[35, 223]. WhenT. dubiusis the ma-ternal parent, the allotetraploid produces floweringheads with long-ray flowers; whenT. pratensisis thematernal parent, the tetraploid bears flowering headswith short-ray flowers [175, 223]. These allopoly-ploid plants of reciprocal derivation are very differentmorphologially and could be mistaken for differentspecies. Studies ofBrassicasuggest that the reciprocalparentage of allopolyploids can also affect nuclear-cytoplasmic interactions [227, 228]. Thus, not allpopulations of a given allopolyploid are ‘equal’; thepopulation or accession chosen for analysis can have amajor impact in molecular evolutionary studies.

Recent studies of a few polyploid species com-plexes are changing our views of the evolution ofpolyploid genomes. Instead of the static polyploidgenomes envisioned by Stebbins [235, 236] and Wag-ner [248], polyploid genomes may change very rapidly(for reviews see [218]; and paper by Wendel, this vol-ume), as evidenced by polyploid species ofBrassicaandGossypium. Polyploid evolution in these genera isreviewed below.

Brassica

The nuclear genomes of the allopolyploidsB. juncea(AB), B. napus(AC), andB. carinata(BC) have ap-parently experienced considerable reshuffling sincetheir origins [131, 185, 186]. Songet al. [227, 228]suggest that the cytoplasmic genomes of the allopoly-ploids have played important roles in the subsequentevolution of the nuclear genomes of these polyploids.

That is, they observed that the nuclear DNA compo-sition of each allotetraploid is more closely relatedto the diploid that contributed the cytoplasm than tothat tetraploid. Thus, for example, the nuclear RFLPpattern ofB. juncea(AB) was very similar to thatof the maternal parent,B. campestris(A genome),and less similar to that of the paternal parent,B. ni-gra (B genome). When the cytoplasmic genomes ofthe parental diploids are highly differentiated and dis-tantly related (as measured, for example, by cpDNArestriction site data [250]), the nuclear contributionof the male donor has been altered much more thanthe nuclear contribution of the female parent. Thus,because the B cytoplasm is very distinct from the Aand C cytoplasms, the nuclear genomes ofB. carinata(BC) andB. juncea(AB) are more similar to that of thediploid that contributed the cytoplasm. That is,B. cari-nata(BC) andB. juncea(AB) exhibit highly modifiedB genomes that have been altered in the direction ofB. oleracea(C genome) andB. rapa (A genome),respectively. The A and C cytoplasms, donated byB. rapaandB. oleracea, in contrast, are very similarand closely related phylogenetically [250], and thusthe A and C genomes inB. napus(AC) have evolvedat similar rates.

Songet al. [226] also examined genome change intwo pairs of synthetic allopolyploidBrassicaspecieswith reciprocal parentage (AB and BA; AC and CA).Extensive changes in the nuclear genome of eachsynthetic polyploid occurred during just five genera-tions. In contrast, there was no evidence of change inthe chloroplast and mitochondrial genomes of thesesynthetic polyploids. Furthermore, the frequency ofgenome change differed between the two pairs of syn-thetic polyploids – that is, twice as many genomechanges were detected in the AB and BA polyploids asin the AC and CA polyploids. When the phylogeneticrelationships of the diploid parents are considered, thedata suggest that the higher the degree of divergencebetween the parental diploid genomes, the greaterthe frequency of nuclear genomic change in the re-sulting polyploid. Lastly, based on preliminary data,chromosome rearrangements involving intergenomic(homoeologous) recombination could be a major fac-tor contributing to the extensive genome change seenin these syntheticBrassicapolyploids.

Comparative genetic data indicate that ‘diploid’Brassicaspecies are themselves ancient polyploids,with genomes so extensively reorganized that theirpolyploid nature was not obvious. Ancient polyploidywas determined by cross-mapping of the genes of the

68

model organismArabidopsisonto the diploidBras-sica genomes. These studies revealed that ‘diploid’Brassica genomes consist of three complete, butrearranged, copies of an ancestralArabidopsis-likegenome [130]. Based on their data, Quiros andcoworkers [185, 186] suggest that the ‘diploid’Bras-sica species may be ancient tetraploids (rather thanhexaploids). The data not only suggest that ‘diploid’brassicas are ‘diploidized’ polyploids, perhaps rep-resenting ancient hexaploids, but also indicate thatBrassicagenomes have evolved through chromosomalfusions and frequent rearrangements.

Gossypium

The comprehensive studies of Wendel and coworkerson polyploid cottons have also provided important in-sights into polyploidization and molecular evolution.Wendelet al. [256] analyzed five allopolyploid (ADgenome) species of cotton and their A- and D-genomediploid progenitors for sequence variation in the nu-clear ITS regions and the 5.8S gene of the rDNAcistron. These arrays occur at four chromosomal lociin allopolyploid cotton, two in each of the A and Dparental genomes. Initially, rDNA sequences repre-senting both the A and D subgenomes must have beenpresent in allotetraploid cotton. However, these rDNAarrays have been homogenized in the allotetraploidsby interlocus concerted evolution: only one repeat typeis present in each allotetraploid. This homogenizationhas been in different directions in different species;phylogenetic analyses indicate that the rDNA fromfour polyploid cottons has been homogenized to a D-genome repeat type whereas that of the fifth speciesstudied is an A-genome type. If not recognized, suchbidirectional interlocus concerted evolution followinghybridization or polyploidy could lead to erroneous as-sessments of relationship and inferences of molecularevolution. However, concerted evolution has not ho-mogenized the rDNA of other polyploid species. Bothparental ITS sequences are still present, for example,in polyploid species ofPaeoniathat are approximatelyone million years old [203].

Other investigations have also revealed the dy-namic nature of polyploid genomes following theirformation. For example, GISH [31, 109, 135] andcomparative genome mapping [100, 130, 138, 185,186, 259] have demonstrated extensive reorganizationof some polyploid genomes (for reviews see [135,218]; paper by Wendel, this volume). Multiple originsof polyploids and genomic reshuffling are important

sources of genetic diversity in polyploid species; theyare also features of polyploids that must be taken intoconsideration by those studying molecular evolutionand conducting genetic studies of polyploids.

Summary

Phylogenetic trees provide a critical underpinning, notonly for systematic studies, but also for investiga-tions of molecular evolution and comparative genetics.Hypotheses of relationships, whether based on mole-cular data (including DNA sequence data, structuralchanges, and gene/intron loss), morphology and othernon-DNA characters or a combination of characters,provide the framework for the most meaningful evo-lutionary comparisons. Considerable progress in re-constructing phylogeny has been made at all levelsin the green plant hierarchy of life, including amonggreen algal lineages, ascertaining the closest relativesof land plants, across all land plants, among ferns,and particularly among the angiosperms. From a his-torical standpoint, this current period during whichphylogenetics has become a central discipline willlikely be remembered as one of the golden ages ofsystematics. Phylogenetic analyses conducted at alltaxonomic levels and in all major groups of plantshave provided some of the most dramatic changes inclassification and concepts of relationship to have oc-curred in the past 100–200 years. The importance ofconsulting these and other recent phylogenetic analy-ses is exemplified well by the angiosperms, whereit is no longer appropriate to use traditional classi-fication schemes to infer higher-level relationships.Furthermore, many well-known angiosperm familiesare now known to be nonmonophyletic; they are eitherpolyphyletic or require merger with other families toform a truly monophyletic group. Phylogenetic hy-potheses present excellent opportunities for broadlybased investigations of molecular evolution, includ-ing studies of the diversification of gene families. Inaddition, a number of so-called model organisms aswell as a number of crops have now been placed inthe appropriate phylogenetic context; sister taxa and aclade of closest relatives have been identified. Phylo-genetic hypotheses provide the opportunity to extendthe knowledge garnered from the study of such modelorganisms to the closest relatives of these plants innatural populations.

69

Acknowledgements

This research was supported in part by NSF GrantDEB-9707868. The authors thank two anonymousreviewers for helpful comments.

References

1. Akkermans ADL, van Dijk C: Non-leguminous root-nodulesymbioses with actinomycetes andRhizobium. In: BroughtonWJ (ed), Nitrogen Fixation, Vol. 1, pp. 57–103. Clarendon,Oxford (1981).

2. Albach DC, Soltis PS, Soltis DE, Olmstead RG: Phyloge-netic analysis of the Asteridae s.l. using sequences of fourgenes. Ann Miss Bot Gard (submitted).

3. Almeida J, Rocheta M, Galego L: Genetic control of flowershape inAntirrhinum majus. Development 124: 1387–1392(1997).

4. Alverson WS, Karol KG, Baum DA, Chase MW, SwensenSM, McCourt R, Sytsma J: Circumscription of the Mal-vales and relationships to other Rosidae: evidence fromrbcLsequence data. Am J Bot 85: 876–877 (1998).

5. Angiosperm Phylogeny Group (APG): An ordinal classifica-tion for the families of flowering plants. Ann Miss Bot Gard85: 531–553 (1998).

6. Avise JC: Molecular Markers, Natural History, and Evolu-tion. Chapman & Hall, New York (1994).

7. Baker DD, Mullin BC: Actinorhizal symbioses. In: StaceyG, Burris RH, Evans HJ (eds), Biological Nitrogen Fixation,pp. 259–291. Chapman & Hall, New York (1992).

8. Barker NP, Linder HP, Harley EH: Polyphyly of Arundi-noideae (Poaceae): evidence fromrbcL sequence data. SystBot 20: 423–435 (1995).

9. Baum DA: The evolution of plant development. Curr OpinPlant Biol 1: 79–86 (1998).

10. Bayer CM, Fay MF, de Bruijn AY, Savolainen V, Morton CM,Kubitzki K, Chase MW: Support for an expanded concept ofMalvaceae within a recircumscribed order Malvales: a com-bined analysis of plastidatpBandrbcL DNA sequences. BotJ Linn Soc 129: 267–303 (1999).

11. Bhattacharya D, Ehlting L: Actin coding regions: genefamily evolution and use as a phylogenetic marker. ArchProtistenkd 145: 155–164 (1995).

12. Bhattacharya D, Medlin L: Algal phylogeny and the origin ofland plants. Plant Physiol 116: 9–15 (1998).

13. Bohs L, Olmstead RG: Phylogenetic relationships inSolanum(Solanaceae) based onndhF sequences. Syst Bot22: 5–17 (1997).

14. Bonierbale MW, Plaisted RL, Tanksley SD: RFLP mapsbased on a common set of clones reveal modes of chro-mosomal evolution in potato and tomato. Genetics 120:1095–1103 (1988).

15. Boutin SR, Young ND, Olson TC, Hu Z-H, ShoemakerRC, Vallejos CE: Genome conservation among three legumegenera detected with DNA markers. Genome 38: 928–937(1995).

16. Bowman JL, Smyth DR, Meyerowitz EM: Genes direct-ing flower development inArabidopsis. Plant Cell 1: 37–52(1989).

17. Bowman JL, Smyth DR, Meyerowitz EM: Genetic interac-tions among floral homeotic genes ofArabidopsis. Develop-ment 112: 1–20 (1991).

18. Bradley D, Carpenter R, Sommer H, Hartley N, Coen E:Complementary floral homeotic phenotypes result from op-posite orientations of a transposon at thePLENA locus ofAntirrhinum. Cell 72: 85–95 (1993).

19. Bull JJ, Huelsenbeck JP, Cunningham CW, Swofford, DL,Waddell, PJ: Partitioning and combining data in phylogeneticanalysis. Syst Biol 42: 384–397 (1993).

20. Capesius I: A molecular phylogeny of bryophytes based onthe nuclear encoded 18S rRNA genes. J Plant Physiol 146:59–63 (1995).

21. Capesius I, Bopp M: New classification of liverworts basedon molecular and morphological data. Plant Syst Evol 207:87–97 (1997).

22. Carpenter R, Coen ES: Floral and homeotic mutations pro-duced by transposon mutagenesis inAntirrhinum majus.Genes Dev 4: 1483–1493 (1990).

23. Chapman RL, Buchheim MA, Delwiche CF, Friedl T, HussVAR, Karol KG, Lewis LA, Manhart, J, McCourt RM, OlsenJL, Waters DA: Molecular systematics of the green algae.In: Soltis DE, Soltis PS, Doyle JJ (eds), Molecular System-atics of Plants II: DNA Sequencing, pp. 508–540. KluwerAcademic Publishers, Boston (1998).

24. Chase MW, Albert VA: A perspective on the contribution ofplastid rbcL DNA sequences to angiosperm phylogenetics.In: Soltis DE, Soltis PS, Doyle JJ (eds), Molecular System-atics of Plants II: DNA Sequencing, pp. 248–507. KluwerAcademic Publishers, Boston (1998).

25. Chase MW, Cox AV: Gene sequences, collaboration andanalysis of large data sets. Aust Syst Bot 11: 215–229 (1998).

26. Chase MW, Duvall MR, Hills HG, Conran JG, Cox AV,Eguiarte LE, Hartwell J, Fay MF, Chaddick LR, CameronKM, Hoot S: Molecular phylogenetics of Lilianae. In: RudallPJ, Cribb, PJ, Cutler DF, Humphries CJ (eds), Mono-cotyledons: Systematics and Evolution, pp. 109–137. RoyalBotanic Gardens, Kew (1995).

27. Chase MW, Soltis DE, Olmstead RG, Morgan D, Les DH,Mishler BD, Duval MR, Price RA, Hills HG, Qiu Y-L, KronKA, Rettig JH, Conti E, Palmer JD, Manhart JR, SytsmaKJ, Michael HJ, Kress WJ, Karol KG, Clark WD, HedrenM, Gaut BS, Jansen RK, Kim K-J, Wimpee CF, Smith JF,Furnier GR, Strauss SH, Xiang Q-Y, Plunkett GM, Soltis PS,Swensen SM, Williams SE, Gadek PA, Quinn CJ, EguiarteLE, Golenberg E, Learn GH Jr., Graham SW, Barrett SCH,Dayanandan S, Albert VA: Phylogenetics of seed plants: ananalysis of nucleotide sequences from the plastid generbcL.Ann Miss Bot Gard 80: 528–580 (1993).

28. Chase MW, Soltis DE, Soltis PS, Rudall PJ, Fay MF, HahnWH, Sullivan S, Joseph J, Givnish T, Sytsma KJ, Pires C:Higher-level systematics of the monocotyledons: an assess-ment of current knowledge and a new classification. SydneyMonocot Symposium (in press).

29. Chase MW, Stevenson DW, Wilkin P, Rudall PJ: Monocotsystematics: a combined analysis. In: Rudall PJ, Cribb, PJ,Cutler DF, Humphries CJ (eds), Monocotyledons: Systemat-ics and Evolution, pp. 685–730. Royal Botanic Gardens, Kew(1995).

30. Chaw S-M, Zharkikh A, Sung H-M, Lau T-C, Li W-H:Molecular phylogeny of extant gymnosperms and seed plantevolution: analysis of nuclear 18S rRNA sequences. Mol BiolEvol 14: 56–68 (1997).

31. Chen Q, Armstrong K: Genomic in situ hybridization inAvena sativa. Genome 37: 607–612 (1994).

32. Chew FS: Are crucivores controlled by mustard oils? No.In: Cutler HG (ed), Biologically Active Natural Products,

70

pp. 155–181. American Chemical Society, Washington, DC(1988).

33. Clark LG, Zhang W, Wendel JF: A phylogeny of the grassfamily (Poaceae) based onndhFsequence data. Syst Bot 20:436–460 (1995).

34. Coen ES, Meyerowitz EM: The war of the whorls: geneticinteractions controlling flower development. Nature 353: 31–37 (1991).

35. Cook LM, Soltis PS, Brunsfeld SJ, Soltis DE: Multiple inde-pendent formations ofTragopogontetraploids (Asteraceae):evidence from RAPD markers. Mol Ecol 7: 1293–1302(1998).

36. Crane, PR: Phylogenetic analysis of seed plants and theorigin of angiosperms. Ann Miss Bot Gard 72: 716–793(1985).

37. Cronquist A: An Integrated System of Classification of Flow-ering Plants. Columbia University Press, New York (1981).

38. Cummings MP, King LM, Kellogg EA: Slipped-strand mis-pairing in a plastid gene:rpoC2 in grasses (Poaceae). MolBiol Evol 11: 1–8 (1994).

39. Dahlgren RT: A system of classification of the angiospermsto be used to demonstrate the distribution of characters. BotNot 128: 119–147 (1975).

40. Dahlgren RT: A commentary on a diagrammatic presentationof the angiosperms in relation to the distribution of characterstates. Plant Syst Evol (Suppl. 1): 253–283 (1977).

41. Dahlgren RT: A revised system of classification of the an-giosperms. Bot J Linn Soc 80: 91–124 (1980).

42. Davis JI, Simmons MP, Stevenson DW, Wendel JW: Datadecisiveness, data quality, and incongruence in phylogeneticanalysis: an example from the monocotyledons using mito-chondrialatpAsequences. Syst Biol 47: 282–310 (1998).

43. Delgado-Salinas A, Bruneau A, Doyle JJ: Chloroplast DNAphylogenetic studies in New World Phaseolinae (Legumi-nosae: Papilionoideae: (Phaseoleae)). Syst Bot 18: 6–17(1993).

44. Delgado-Salinas A, Turley T, Richman A, Lavin M: Phyloge-netic analysis of the cultivated and wild species ofPhaseolus(Fabaceae). Syst Bot, in press.

45. Delwiche, CF, Kuhsel M, Palmer, JD: Phylogenetic analy-sis of tufA sequences indicates a cyanobacterial origin of allplastids. Mol Phyl Evol 4: 110–128 (1995).

46. de Queiroz A: For consensus (sometimes). Syst Biol 42: 368–372 (1993).

47. de Queiroz A, Donoghue MJ, Kim J: Separate versus com-bined analysis of phylogenetic evidence. Annu Rev Ecol Syst26: 657–681 (1995).

48. Dobzhansky T: Nothing in biology makes sense except in thelight of evolution. Am Biol Teacher 35:125–129 (1973).

49. Doebley J: Molecular systematics ofZea(Gramineae). May-dica 35: 143–150 (1990).

50. Doebley J, Durbin M, Golenberg EM, Clegg MT, MaDP: Evolutionary analysis of the large subunit of carboxy-lase (rbcL) nucleotide sequence data among the grasses(Poaceae). Evolution 44: 1097–1108 (1990).

51. Doebley J, Renfroe W, Blanton A: Restriction site varia-tion in theZeachloroplast genome. Genetics 117: 139–147(1987).

52. Donoghue MJ, Ree RH, Baum DA: Phylogeny and the evo-lution of flower symmetry in the Asteridae. Trends Plant Sci3: 311–317 (1998).

53. Doyle JA: Phylogeny of vascular plants. Annu Rev Ecol Syst29: 567–599 (1998).

54. Doyle JA: Molecules, morphology, fossils and the relation-ship of angiosperms and Gnetales. Mol Phyl Evol 9: 448–462(1998).

55. Doyle JA, Donoghue MJ: Seed plant phylogeny and the ori-gin of angiosperms: an experimental cladistic approach. BotRev 52: 321–431 (1986).

56. Doyle JA, Donoghue MJ: The importance of fossils inelucidating seed plant phylogeny and macroevolution. RevPalaeobot Palynol 50: 63–95 (1987).

57. Doyle JA, Donoghue MJ: Fossils and seed plant phylogenyreanalyzed. Brittonia 44: 89–106 (1992).

58. Doyle JA, Hotton CL: Diversification of early angiospermpollen in a cladistic context. In: Blackmore S, Barnes SH(eds), Pollen and Spores, pp. 169–195. Clarendon Press,Oxford.

59. Doyle JJ: Phylogeny of the legume family: an approach tounderstanding the origins of nodulation. Annu Rev Ecol Syst25: 325–49 (1994).

60. Doyle JJ: Phylogenetic perspectives on nodulation: evolvingviews of plants and symbiotic bacteria. Trends Plant Sci 3:473–478 (1998).

61. Doyle JJ, Davis JI, Soreng RJ, Garvin D, Anderson MJ:Chloroplast DNA inversions and the origin of the grass fam-ily (Poaceae). Proc Natl Acad Sci USA 89: 7722–7726(1992).

62. Doyle JJ, Doyle JL: Chloroplast DNA phylogeny of the pa-pilionoid legume tribe Phaseoleae. Syst Bot 18: 309–327(1993).

63. Doyle JJ, Doyle JL, Ballenger JA, Dickson EE, Kajita T,Ohashi H: A phylogeny of the chloroplast generbcL in theLeguminosae: taxonomic correlations and insights into theevolution of nodulation. Am J Bot 84: 541–554 (1997).

64. Doyle JJ, Doyle JL, Brown AHD: A chloroplast DNA phy-logeny of the wild perennial relatives of soybean (GlycinesubgenusGlycine): congruence with morphological andcrossing groups. Evolution 44: 371–389 (1990).

65. Doyle JJ, Doyle JL, Brown AHD. Incongruence in the diploidB-genome species complex ofGlycine(Leguminosae) revis-ited: histone H3-D alleles vs. chloroplast haplotypes. MolBiol Evol 16: 354–362 (1999).

66. Doyle JJ, Kanazin V, Shoemaker RC: Phylogenetic utility ofhistone H3 intron sequences in the perennial relatives of soy-bean (Glycine: Leguminosae). Mol Phyl Evol 64: 438–447(1996).

67. Downie SR, Palmer JD: Use of chloroplast DNA rearrange-ments in reconstructing plant phylogeny. In: Soltis PS, SoltisDE, Doyle JJ (eds.), Molecular Systematics of Plants, pp. 14–35. Chapman & Hall, New York (1992).

68. Duff RJ, Nickrent DL: Phylogenetic relationships of landplants using mitochondrial small-subunit rDNA sequences.Am J Bot 86: 372–386 (1999).

69. Duvall MR, Clegg MT, Chase MW, Clark WD, Kress WJ,Hills HG, Eguiarte LE, Smith JF, Gaut BS, Zimmer EA,Learn GH Jr: Phylogenetic hypotheses for the monocotyle-dons constructed formrbcL sequences. Ann Miss Bot Gard80: 607–619 (1993).

70. Eernisse DJ, Kluge AG: Taxonomic congruence versus to-tal evidence, and the phylogeny of anmiotes inferred fromfossils, molecules and morphology. Mol Biol Evol 10: 1170–1195 (1993).

71. Ehrlich PR, Raven PH: Butterflies and plants: a study incoevolution. Evolution 18: 586–608 (1965).

71

72. Embley MT, Hirt RP, Williams DM: Biodiversity at the mole-cular level: the domains, kingdoms and phyla of life. PhilTrans R Soc Lond 345: 21–31 (1994).

73. Endress PK: Evolution and floral diversity: the phylogeneticsurroundings ofArabidopsisandAntirrhinum. Int J Plant Sci153: S106–S122 (1992).

74. Endress PK:Antirrhinum and Asteridae: evolutionarychanges of floral symmetry. Soc Exp Biol Symp Series 51:133–140 (1997).

75. Erickson LR, Strauss NA, Beversdorf WB: Restriction pat-terns reveal origins of chloroplast genomes inBrassicaamphidiploids. Theor Appl Genet 65: 201–206 (1983).

76. Farris JS: Jac, Version 4.4. Swedish Museum of NaturalHistory, Stockholm (1996).

77. Farris JS, Albert VA, Källersjö M, Lipscomb D, KlugeAG: Parsimony jackknifing outperforms neighbor-joining.Cladistics 12: 99–124 (1996).

78. Farris JS, Källersjö M, Kluge AG, Bult C: Testing signifi-cance of incongruence. Cladistics 10: 315–319 (1995).

79. Felsenstein J: The number of evolutionary trees. Syst Zool27: 27–33 (1978).

80. Felsenstein, J: Confidence limits on phylogenies: an ap-proach using the bootstrap. Evolution 39: 783–791 (1985).

81. Fenwick GR, Heaney RK, Mullin WJ: Glucosinolates andtheir breakdown products in food and food plants. Crit RevFood Sci Nutr 18: 123–201 (1983).

82. Friedl T: The evolution of the green algae. Plant Syst Evol(1997). [Suppl.] II: 87–101.

83. Gadek PA, Fernando ES, Quinn CJ, Hoot SB, Terrazas T,Sheahan MC, Chase MW: Sapindales: molecular delim-itation and infraordinal groups. Am J Bot 83: 802–811(1996).

84. Gale MD, Devos, KM: Comparative genetics in the grasses.Proc Natl Acad Sci USA 95: 1971–1974 (1998).

85. Gale MD, Devos, KM: Plant comparative genetics after 10years. Science 282: 656–658 (1998).

86. Gaut BS, Doebley JF: DNA sequence evidence for the seg-mental allopolyploid origin of maize. Proc Natl Acad SciUSA 94: 6809–6814 (1997).

87. Goloboff P: Pee-Wee and NONA. Computer programs anddocumentation. New York (1993).

88. Goremykin V, Bobrova V, Pahnke J, Troitsky A, Antonov A,Martin W: Noncoding sequences from the slowly evolvingchloroplast inverted repeat in addition torbcL data do notsupport gnetalean affinities of angiosperms. Mol Biol Evol13: 383–396 (1996).

89. Govindaraju DR, Lewis P, Cullis C: Phylogenetic analysisof pines using ribosomal DNA restriction fragment lengthpolymorphisms. Plant Syst Evol 179: 141–153 (1992).

90. Graham LE: Origin of Land Plants. Wiley, New York (1993).91. Grant, V: Plant Speciation, 2nd ed. Columbia University

Press, New York (1981).92. Graur D, Duret L, Gouy M: Phylogenetic position of the

order Lagomorpha (rabbits, hares and allies). Nature 379:333–335 (1996).

93. Graybeal A: Is it better to add taxa or characters to a difficultphylogenetic problem? Syst Biol 47: 9–17 (1998).

94. Hamby RK, Zimmer EA: Ribosomal RNA as a phylogenetictool in plant systematics. In: Soltis PS, Soltis DE, Doyle JJ(eds), Molecular Systematics of Plants, pp. 50–91. Chapman& Hall, New York (1992).

95. Harvey PH, Pagel MD: The Comparative Method in Evolu-tionary Biology. Oxford University Press, New York (1991).

96. Hasebe M, Kofuji R, Ito M, Kato M, Iwatsuki K, UedaK: Phylogeny of gymnosperms inferred fromrbcL genesequences. Bot Mag 105: 673–679 (1992).

97. Hasebe M, Wolf PG, Pryer KM, Ueda K, Ito M, Sano R,Gastony GJ, Yokoyama J, Manhart JR, Murakami N, CraneEH, Haufler CH, Hauk WD: Fern phylogeny based onrbcLnucleotide sequences. Am Fern J 85: 134–181 (1995).

98. Hedderson TA., Chapman RL, Rootes WL: Phylogeneticrelationships of bryophytes inferred from nuclear-encodedrRNA gene sequences. Plant Syst Evol 200: 213–224 (1996).

99. Heiser CB: Study in the evolution of the sunflower speciesHelianthus annuusand H. bolanderi. Univ Calif Publ Bot23: 157–196 (1949).

100. Helentjaris T, Weber D, Wright S: Identification of the ge-nomic locations of duplicate nucleotide sequences in maizeby analysis of restriction fragment length polymorphisms.Genetics 118: 353 (1988).

101. Hershkovitz MA, Zimmer EA: On the evolutionary originsof the cacti. Taxon 46: 217–232 (1997).

102. Hickey LJ, Taylor DW: Origin of the angiosperm flower. In:Taylor DW, Hickey LJ (eds), Flowering Plant Origin, Evo-lution and Phylogeny, pp. 176–231. Chapman & Hall, NewYork (1996).

103. Hillis DM: Inferring complex phylogenies. Nature 383: 130(1996).

104. Hillis DM, Huelsenbeck JP, Swofford DL: Hobgoblin ofphylogenetics? Nature 369: 363–364 (1994).

105. Huelsenbeck JP, Bull JJ: A likelihood ratio test for detec-tion of conflicting phylogenetic signal. Syst Biol 45: 92–98(1996).

106. Huelsenbeck JP, Bull JJ, Cunningham CW: Combining datain phylogenetic analysis. Trends Ecol Evol 11: 152–158(1996).

107. Huss VAR, Kranz HD: Charophyte evolution and the originof land plants. Plant Syst Evol (Suppl.) 11: 103–114 (1997).

108. Jansen RK, Kim K-J: Implications of chloroplast DNAdata for the classification and phylogeny of the Asteraceae.In: Hind DJN, Beentje H (eds), Compositae: Systematics,Vol. 1., pp. 317–339. Proceedings of the International Com-positae Conference, Kew, 1994. Royal Botanic Gardens,Kew (1996).

109. Jellen EN, Gill BS, Cox TS: Genomic in situ hybridizationdifferentiates between A/D- and C-genome chromatin anddetects intergenomic translocations in polyploid oat species(genusAvena). Genome 37: 607–612 (1994).

110. Jiang, Q-X, Wright RJ, El-Zik KM, Paterson AH: Polyploidformation created unique avenues for response to selection inGossypium(cotton). Proc Natl Acad Sci USA 95: 4419–4424(1998).

111. Johnson LA, Soltis DE: Assessing congruence: empirical ex-amples from molecular data. In: Soltis DE, Soltis PS, DoyleJJ (eds), Molecular Systematics of Plants II: DNA Sequenc-ing, pp. 297–348. Kluwer Academic Publishers, Boston(1998).

112. Jones TR, Kluge AG, Wolf AJ: When theories and method-ologies clash: a phylogenetic reanalysis of the North Ameri-can ambystomatid salamanders (Caudata: Ambystomatidae).Syst Biol 42: 92–102 (1993).

113. Judd R, Sanders RW, Donoghue MJ: Angiosperm familypairs: preliminary phylogenetic analyses. Harvard Pap Bot5: 1–51 (1994).

114. Källersjö M, Farris JS, Chase MW, Bremer B, Fay MF,Humphries CJ, Petersen G, Seberg OL, Bremer K: Simul-taneous parsimony jackknife analysis of 2538rbcL DNA

72

sequences reveals support for major clades of green plants,land plants, seed plants and flowering plants. Plant Syst Evol213: 259–287 (1998).

115. Kellogg EA: Relationships of cereal crops and other grasses.Proc Natl Acad Sci USA 95: 2005–2010 (1998).

116. Kellogg EA, Appels R, Mason-Gamer RJ: When genes telldifferent stories: the diploid genera of Triticeae (Gramineae).Syst Bot 21: 321–347 (1996).

117. Kellogg EA, Juliana ND: The structure and function of Ru-BisCO and their implications for systematic studies. Am JBot 84: 413–428 (1997).

118. Kellogg EA, Linder HP: Phylogeny of Poales. In: Rudall PJ,Cribb PJ, Cutler DF, Humphries CJ (eds), Monocotyledons:Systematics and Evolution, pp. 511–542. Royal BotanicGardens, Kew (1995).

119. Kenrick P, Crane PR: The Origin and Early Diversification ofLand Plants. Smithsonian Institution Press, Washington, DC(1997).

120. Kenrick P, Crane PR: The origin and early evolution of plantson land. Nature 389: 33–39 (1997).

121. Kluge AG: A concern for evidence and a phylogenetichypothesis of relationships amongEpicrates(Boidae, Ser-pentes). Syst Zool 38: 7–25 (1989).

122. Kluge AG, Wolf AJ: Cladistics: what’s in a word? Cladistics9: 183–199 (1993).

123. Koch M, Gurganus M, Bishop J, Mitchell-Olds T: Phylogenyand microsatellites forArabidopsisandArabis: resources forfunctional and evolutionary genetics. Plant Biol (in press).

124. Kolukisaoglu HM, Marx MS, Wiegmann C, Hanelt S,Schneider-Portsch AW: Divergence of the phytochrome genefamily predates angiosperm evolution and suggests thatSe-laginella andEquisetumrose prior toPsilotum. J Mol Evol41: 329–337 (1995).

125. Kramer EM, Dorit RL, Irish VF: Molecular evolution ofgenes controlling petal and stamen development: duplica-tion and divergence within theAPETALA3andPISTILLATAMADS-box gene lineages. Genetics 149: 765–783 (1998).

126. Kranz HD, Huss VAR: Molecular evolution of ferns andallies, and their relationship to seed plants: evidence fromcomplete 18S rRNA gene sequences. Plant Syst Evol 202:1–11 (1996).

127. Kranz HD, Miks D, Siegler M-L, Capesius I, Sensen CHW,Huss VA: The origin of land plants: phylogenetic relation-ships between Charophytes, Bryophytes, and vascular plantsinferred from complete small subunit ribosomal RNA genesequences. J Mol Evol 41: 74–84 (1995).

128. Krupkin AB, Liston A, Strauss, SH: Phylogenetic analysis ofthe hard pines (PinussubgenusPinus, Pinaceae) from chloro-plast DNA restriction site analysis. Am J Bot 83: 489–498(1996).

129. La Duke JC, Doebley J: A chloroplast based phylogeny ofthe Malvaceae. Syst Bot 20: 259–271 (1995).

130. Lagercrantz U: Comparative mapping betweenArabidopsisthalianaandBrassica nigraindicates thatBrassicagenomeshave evolved through extensive genome replication accom-panied by chromosome fusions and frequent rearrangments.Genetics 150: 1217–1228 (1998).

131. Lagercrantz U, Lydiate D: Comparative genome mapping inBrassica. Genetics 144: 1903–1910 (1996).

132. Lanyon SM: Phylogenetic frameworks: towards a firmerfoundation for the comparative approach. Biol J Linn Soc 49:45–61 (1993).

133. Larson A: The comparison of morphological and moleculardata in phylogenetic systematics. In: Schierwater B, Streit B,

Wagner GP, DeSalle R (eds), Molecular Ecology and Evolu-tion: Approaches and Applications, pp. 371–390. BirkhaüserVerlag, Basel, Switzerland (1994).

134. Lavin M, Eshbaugh E, Hu J-M, Matthes S, Sharrock RA:Monophyletic subgroups of the tribe Millettieae (Legumi-nosae) as revealed by phytochrome nucleotide sequence data.Am J Bot 85: 412–433 (1998).

135. Leitch IJ, Bennett MD: Polyploidy in angiosperms. TrendsPlant Sci 2: 470–476 (1997).

136. Lewis LA, Mishler BD, Vilgalys R: Phylogenetic relation-ships of the liverworts (Hepaticae), a basal embryophytelineage, inferred from nucleotide sequence data of the chloro-plast generbcL. Mol Phyl Evol 7: 377–393 (1997).

137. Little EL Jr, Critchfield WB: Subdivisions of the genusPinus(pines). USDA, Forest Service, Washington, DC (Miscella-neous Publications 1144) (1969).

138. Liu B, Vega J M, Feldman M: Rapid genomic changes innewly synthesized amphiploids ofTriticum andAegilops. II.Changes in low-copy coding DNA sequences. Genome 41:535–542 (1998).

139. Livingstone KD, Lackney VK, Blauth JR, Van Wijk R, JahnMK: Genome mapping inCapsicumand the evolution ofgenome structure in the Solanaceae. Theor Appl Genet (inpress).

140. Loconte H, Stevenson DW: Cladistics of the Spermatophyta.Brittonia 42: 197–211 (1990).

141. Luo D, Carpenter R, Vincent C, Copsey L, Coen E: Originof floral asymmetry inAntirrhinum. Nature 383: 794–799(1996).

142. Luthy B, Matile PH: The mustard oil bomb: rectified analy-sis of the subcellular organisation of the myrosinase system.Biochem Physiol Pflanzen 179: 5–12 (1984).

143. Lutzoni F, Vilgalys R: Integration of morphological andmolecular data sets in estimating fungal phylogenies. Can JBot (Suppl.) 73: S649–S659 (1995).

144. Maddison DR: The discovery and importance of multipleislands of most-parsimonious trees. Syst Zool 40: 315–328(1991).

145. Malek O, Lattig K, Hiesel R, Brennicke A, Knoop V: RNAediting in bryophytes and a molecular phylogeny of landplants. EMBO J 15: 1403–1411 (1996).

146. Manhart JR: Phylogenetic analysis of green plantrbcL se-quences. Mol Phyl Evol 3: 114–127 (1994).

147. Manhart JR: Chloroplast 16SrDNA sequences and phyloge-netic relationships of fern allies and ferns. Am Fern J 85:182–192 (1995).

148. Mason-Gamer RJ, Kellogg EA: Testing for phylogeneticconflict among molecular data sets in the tribe Triticeae(Gramineae). Syst Biol 45: 522–543 (1996).

149. Masterson J: Stomatal size in fossil plants: evidence for poly-ploidy in majority of angiosperms. Science 264: 421–423(1994).

150. Mathews S, Donoghue MJ: Duplicate gene rooting and dif-ficult rooting problems in plant phylogeny. Am J Bot 85(abstract): 143 (1998).

151. Mayer MS, Soltis PS: Chloroplast DNA variation inLens(Leguminosae): phylogenetic relationships and the origin ofthe cultivated lentil. Theor Appl Genet 87: 773–781 (1994).

152. McDade LA: Hybrids and phylogenetic systematics. I. Pat-terns of character expression in hybrids and their implicationsfor cladistic analysis. Evolution 44: 1685–1700 (1990).

153. McDade LA: Hybrids and phylogenetic systematics. In: Theimpact of hybrids on cladistic analysis. Evolution 46: 1329–1346 (1992).

73

154. McDade LA: Hybridization and phylogenetics. In: HochPC, Stephenson AG (eds.) Experimental and Molecular Ap-proaches to Plant Biosytematics, pp. 305–331. MissouriBotanical Garden, St. Louis, MO (1995).

155. McDade LA: Hybrids and phylogenetic systematics. III.Comparison with distance methods. Syst Bot 22: 669–683(1997).

156. McDowell JM, Huang S, McKinney EC, An Y-Q, MeacherRB: Structure and evolution of the actin gene family inArabidopsis thaliana. Genetics 142: 587–602 (1996).

157. Mickevich MF, Farris JS: The implications of congruence inMenida. Syst Zool 30: 351–370 (1981).

158. Mishler BD: Cladistic analysis of molecular and morpholog-ical data. Am J Phys Anthropol 94: 143–156 (1994).

159. Mishler BD, Lewis LA, Buchheim MA, Renzaglia KS, Gar-bary DJ, Delwiche CF, Zechman FW, Kantz TS, ChapmanRL: Phylogenetic relationships of the ‘green algae’ and‘bryophytes’. Ann Miss Bot Gard 81: 451–483 (1994).

160. Miyamoto MM, Fitch WM: Testing species phylogenies andphylogenetic methods with congruence. Syst Biol 44: 64–76(1995).

161. Moniz de Sa M, Drouin G: Phylogeny and substitution ratesof angiosperm actin genes. Mol Biol Evol 13: 1198–1212(1996).

162. Mullin BC, Swensen SM, Goetting-Minesky P: Hypothesesfor the evolution of actinorhizal symbioses. In: GresshoffPM, Roth LE, Stacey G, Newton WE (eds), Nitrogen Fix-ation Achievements and Objectives, pp. 781–787. Chapman& Hall, New York (1990).

163. Nadot S, Bajon R, Lejeune B: The chloroplast generps4 as atool for the study of Poaceae phylogeny. Plant Syst Evol 191:27–38 (1994).

164. Nandi WI, Chase MW, Endress PK: A combined cladisticanalysis of angiosperms usingrbcL and non-molecular datasets. Ann Miss Bot Gard 85: 137–212 (1998).

165. Nickrent DL, Duff JR, Colwell AE, Wolfe AD, Young ND,Steiner KE, dePamphilis CW: Molecular phylogenetic andevolutionary studies of parasitic plants. In: Soltis DE, SoltisPS, Doyle JJ (eds), Molecular Systematics of Plants II: DNASequencing, pp. 211–241. Kluwer Academic Publishers,Boston (1998).

166. Nixon KC, Crepet WL, Stevenson D, Friis EM: A reeval-uation of seed plant phylogeny. Ann Miss Bot Gard 81:484–533 (1994).

167. O’Kane SL, Al-Shebaz IA: A synopsis ofArabidopsis(Bras-sicaceae). Novon 7: 323–327 (1997).

168. O’Kane SL, Schaal BA, Al-Shebaz IA: The origins ofArabidopsis suecica(Brassicaceae) as indicated by nuclearrDNA sequences. Syst Bot 21: 559–566 (1996).

169. Olmstead RG, Bremer B, Scott K, Palmer JD: A parsimonyanalysis of the Asteridae sensu lato based onrbcL sequences.Ann Miss Bot Gard 80: 700–722 (1993).

170. Olmstead RG, dePamphilis CW, Wolfe AD, Young ND,Reeves PA: Disintegration of the Scrophulariaceae (submit-ted).

171. Olmstead RG, Palmer JD: Implications for the phylogeny,classification, and biogeography ofSolanumfrom cpDNArestriction site variation. Syst Bot 22: 19–30 (1994).

172. Olmstead RG, Reeves PA: Evidence for the polyphyly ofthe Scrophulariaceae based on chloroplastrbcL and ndhFsequences. Ann Miss Bot Gard 82: 176–193 (1995).

173. Olmstead RG, Reeves PA, Yen AC: Patterns of sequence evo-lution and implications for parsimony analysis of chloroplastDNA. In: Soltis DE, Soltis PS, Doyle JJ (eds), Molecular

Systematics of Plants II: DNA Sequencing, pp. 164–187.Kluwer Academic Publishers, Boston (1998).

174. Olmstead RG, Sweere JA: Combining data in phylogeneticsystematics: an empirical approach using three moleculardata sets in the Solanaceae. Syst Biol 43: 467–481 (1994).

175. Ownbey M, McCollum GD: Cytoplasmic inheritance andreciprocal amphiploidy inTragopogon. Am J Bot 40: 788–796.

176. Pahnke J, Goremykin V., Bobrova V, Troitsky A, AntonovA, Martin W: Utility of rDNA internal transcribed spacersequences from the inverted repeat of chloroplast DNA inpteridophyte molecular phylogenetics. In: Camus JM, GibbyM, Johns RJ (eds), Pteridology in Perspective, pp. 217–230.Royal Botanic Gardens, Kew, UK.

177. Palmer JD, Jorgensen RA, Thompson WF: Chloroplast DNAvariation and evolution inPisum: patterns of change andphylogenetic analysis. Genetics 109: 195–213 (1985).

178. Palmer JD, Shields CR, Cohen DB, Orton TJ: ChloroplastDNA evolution and the origin of the amphidiploidBrassicaspecies. Theor Appl Genet 65: 181–189 (1983).

179. Patterson C, Williams DM, Humphries CJ: Congruence be-tween molecular and morphological phylogenies. Annu RevEcol Syst 24: 153–188 (1993).

180. Perez F, Menendez A, Dehal P, Quiros CF: Genomic struc-tural differentiation inSolanum: comparative mapping of theA and E genomes. Theor Appl Genet 98: 1183–1193 (1999).

181. Price RA, Palmer JD, Al-Shehbaz IA: Systematic relation-ships ofArabidopsis: a molecular and morphological per-spective. In: Somerville C, Meyerowitz E (eds),Arabidopsis,pp. 7–19. Cold Spring Harbor Laboratory Press, Cold SpringHarbor, NY (1994).

182. Qiu Y-Q, Cho Y., Cox JC, Palmer JD: The gain of three mi-tochondrial introns identifies liverworts as the earliest landplants. Nature 394: 671–674 (1998).

183. Qiu Y-Q, Palmer, JD: Mitochondrial genome evolution andland phylogeny. Am J Bot 84 (abstract): 113–114 (1996).

184. Qiu Y-Q, Palmer, JD: Phylogeny of early land plants: insightsfrom genes and genomes. Trends Plant Sci 4: 26–30 (1998).

185. Quiros, CF: Molecular markers and their applications to ge-netics, breeding and the evolution ofBrassica. J Jpn Soc HortSci 67: 1180–1185 (1998).

186. Quiros CF, Hu J, Truco MJ: DNA based markerBrassicamaps. In: Phillips RL, Vasil IK (eds), DNA-Based Mark-ers in Plants, pp. 199–222. Kluwer Academic Publishers,Dordrecht, Netherlands (1993).

187. Raubeson, LA Jansen RK: Chloroplast DNA evidence onthe ancient evolutionary split in vascular land plants. Science255: 1697–1699 (1992).

188. Reeves PA, Olmstead RG: Evolution of novel morphologi-cal, ecological, and reproductive traits in a clade containingAntirrhinum. Am J Bot 85: 1047–1056 (1998).

189. Rice KA, Donoghue MJ, Olmstead RG: Analyzing large datasets:rbcL 500 revisited. Syst Biol 46: 554–563 (1997).

190. Rieseberg LH: Homoploid reticulate evolution inHelianthus(Asteraceae): evidence from ribosomal genes. Am J Bot 75:753–766 (1991).

191. Rieseberg LH: The role of hybridization in evolution: oldwine in new skins. Am J Bot 82: 944–953 (1995).

192. Rieseberg LH, Sinervo B, Linder CR, Ungerer MC, AriasDM: Role of gene interactions in hybrid speciation: evidencefrom ancient and experimental hybrids. Science 272: 741–745 (1996).

74

193. Rieseberg LH, Soltis DE: Phylogenetic consequences of cy-toplasmic gene flow in plants. Evol Trends Plants 5: 65–84(1991).

194. Rieseberg LH, Van Fossen C, Desroches A: Genomic reorga-nization accompanies hybrid speciation in wild sunflowers.Nature 375: 313–316 (1995).

195. Rodman JE: A taxonomic analysis of glucosinolate-producing plants, part 1: Phenetics. Syst Bot 16: 598–618(1991).

196. Rodman JE: A taxonomic analysis of glucosinolate-producing plants, part 2: Cladistics. Syst Bot 16: 619–629(1991).

197. Rodman JE, Karol KG, Price RA, Sytsma KJ: Molecules,morphology, and Dahlgren’s expanded order Capparales.Syst Bot 21: 289–307 (1996).

198. Rodman JE, Price RA, Karol K, Conti E, Sytsma KJ, PalmerJD: Nucleotide sequences of therbcL gene indicate mono-phyly of mustard oil plants. Ann Miss Bot Gard 80: 686–699(1993).

199. Rodman JE, Soltis PS, Soltis DE, Sytsma KJ, Karol KG:Parallel evolution of glucosinolate biosynthesis inferred fromcongruent nuclear and plastid gene phylogenies. Am J Bot85: 997–1006 (1998).

200. Rodrigo AG, Kelly-Borges M, Bergquist PR, Bergquist PL:A randomisation test of the null hypothesis that two clado-grams are sample estimates of a parametric phylogenetic tree.NZ J Bot 31: 257–268 (1993).

201. Sanderson MJ, Doyle JJ: Reconstruction of organismal andgene phylogenies from data on multigene families: concertedevolution, homoplasy, and confidence. Syst Biol 41: 4–17(1992).

202. Sanderson MJ, Wojciechowski MF: Diversification rates ina temperate legume clade: are there ‘so many species’ ofAstragalus(Fabaceae)? Am J Bot 83: 1488–1502.

203. Sang T, Crawford DJ, Stuessy TF: Documentation of reticu-late evolution in peonies (Paeonia) using internal transcribedspacer sequences of nuclear ribosomal DNA: implications forbiogeography and concerted evolution. Proc Natl Acad SciUSA 92: 6813–6817 (1995).

204. Savolainen V, Chase MW, Morton CM, Hoot SB, Soltis DE,Bayer C, Fay MF, de Bruijn A, Sullivan S, Qiu Y-L: Phylo-genetics of flowering plants based upon a combined analysisof plastidatpBandrbcL gene sequences. Syst Biol (in press).

205. Schilling EE: Phylogenetic analysis ofHelianthus (Aster-aceae) based on chloroplast DNA restriction site data. TheorAppl Genet 94: 925–933 (1997).

206. Schilling EE, Linder CR, Noyes RD, Rieseberg LH: Phy-logenetic relationships inHelianthus(Asteraceae) based onnuclear ribosomal DNA internal transcribed spacer regionsequence data. Syst Bot 23: 177–189 (1998).

207. Schilling EE, Panero JL: Phylogenetic reticulation in subtribeHelianthinae. Am J Bot 83: 939–948 (1996).

208. Schneider-Poetsch HA, Marx S, Kolukisaoglu H, Hanelt S,Braun B: Phytochrome evolution: phytochrome genes inferns and mosses. Physiol Plant 91: 241–250 (1994).

209. Schwarz-Sommer Z, Hue I, Huijser P, Flor PJ, Hansen R,Tetens F, Lonnig W-E, Saedler H, Sommer H: Characteri-zation of theAntirrhinum floral homeotic MADS-box genedeficiens: evidence for DNA binding and autoregulation of itspersistent expression throughout flower development. EMBOJ 11: 251–263 (1992).

210. Schwarz-Sommer Z, Huijser P, Nacken W, Saedler H, Som-mer H: Genetic control of flower development by homeoticgenes inAntirrhinum majus. Science 250: 931–936 (1990).

211. Seelanan T, Schnabel H, Wendel JF: Congruence and con-sensus in the cotton tribe. Syst Bot 22: 259–290 (1997).

212. Shoemaker R, Olson T, Kanazin V: Soybean genome or-ganization: evolution of a legume genome. In: GustafsonJP, Flavell RB (eds), Genomes of Plants and Animals. 21stStadler Genetics Symposium, pp. 139–150. Plenum Press,New York (1996).

213. Sinha NR, Kellogg EA: Parallelism and diversity in multipleorigins of C4 photosynthesis in the grass family. Am J Bot83: 1458–1470 (1996).

214. Soltis DE, Hibsch-Jetter C, Soltis PS, Chase MW, Farris JS:Molecular phylogenetic relationships among angiosperms:an overview based onrbcL and 18S rDNA sequences. In:Iwatsuki K, Raven PH (eds.), Evolution and Diversificationof Land Plants, pp. 157–178. Springer-Verlag, New York(1997).

215. Soltis DE, Soltis PS: Molecular data and the dynamic natureof polyploidy. Crit Rev Plant Sci 12: 243–273 (1993).

216. Soltis DE, Soltis PS: The dynamic nature of polyploidgenomes. Proc Natl Acad Sci USA 92: 8089–8091 (1995).

217. Soltis DE, Soltis PS: Choosing an approach and an appro-priate gene for phylogenetic analysis. In: Soltis DE, SoltisPS, Doyle JJ (eds), Molecular Systematics of Plants II: DNASequencing, pp. 1–42. Kluwer Academic Publishers, Boston(1998).

218. Soltis DE, Soltis PS: Polyploidy: recurrent formation andgenome evolution. Trends Ecol Evol 14: 348–392 (1999).

219. Soltis DE, Soltis PS, Chase ME, Mort ME, Albach DC, ZanisM, Savolainen V, Hahn WH, Hoot SB, Fay MF, Axtell, M,Swensen SM, Nixon KC, Farris JS: Angiosperm phylogenyinferred from a combined data set of 18S rDNA,rbcL, andatpβ sequences. Bot J Linn Soc (submitted).

220. Soltis DE, Soltis PS, Morgan DR, Swensen SM, MullinBC, Dowd JM, Martin PG: Chloroplast gene sequence datasuggest a single origin of the predisposition for symbioticnitrogen fixation in angiosperms. Proc Natl Acad Sci USA92: 2647–2651 (1995).

221. Soltis DE, Soltis PS, Mort ME, Chase MW, Savolainen V,Hoot SB, Morton CM: Inferring complex phylogenies usingparsimony: an empirical approach using three large DNAdata sets for angiosperms. Syst Biol 47: 32–42 (1998).

222. Soltis DE, Soltis PS, Nickrent DL, Johnson LA, Hahn WJ,Hoot SB, Sweere JA, Kuzoff RK, Kron KA, Chase MW,Swensen SM, Zimmer EA, Chaw S-M, Gillespie LY, KressWJ, Sytsma KJ: Angiosperm phylogeny inferred from 18Sribosomal DNA sequences. Ann Miss Bot Gard 84: 1–49(1997).

223. Soltis PS, Plunkett GM, Novak SJ, Soltis DE: Genetic vari-ation inTragopogonspecies: additional origins of the allote-traploidsT. mirusandT. miscellus(Compositae). Am J Bot82: 1329–1341 (1995).

224. Soltis PS, Soltis DS: Molecular evolution of 18S rDNAin angiosperms: implications for character weighting inphylogenetic analysis. In: Soltis DE, Soltis PS, Doyle JJ(eds), Molecular Systematics of Plants II: DNA Sequencing,pp. 188–211. Kluwer Academic Publishers, Boston (1998).

225. Soltis PS, Soltis DE, Wolf PG, Nickrent DL, Chaw S-M,Chapman RL: The phylogeny of land plants inferred from18S rDNA sequences: pushing the limits of rDNA signal?Mol Biol Evol (in press).

226. Song KM, Lu P, Tang K, Osborn TC: Rapid genome changein synthetic polyploids ofBrassicaand its implications forpolyploid evolution. Proc Natl Acad Sci USA 92: 7719–7723(1995).

75

227. Song KM, Osborn, TC: Polyphyletic origins ofBrassica na-pus: new evidence based on organelle and and nuclear RFLPanalyses. Genome 35: 992 (1993).

228. Song KM, Osborn TC, Williams PH:Brassica taxonomybased on nuclear restriction fragment length polymorphisms(RFLPs). 1. Genome evolution of diploid and amphidiploidspecies. Theor Appl Genet 75: 784–794 (1988).

229. Soreng RJ, Davis JI: Phylogenetics and character evolutionin the grass family (Poaceae): simultaneous analysis of mor-phological and chloroplast DNA restriction site data sets. BotRev 64: 1–85 (1998).

230. Spooner DM, Sytsma KJ: Reexamination of series rela-tionships of Mexican and Central American wild potatoes(Solanumsect. Petota): evidence from chloroplast DNArestriction site variation. Syst Bot 17: 432–448 (1992).

231. Spooner DM, Sytsma KJ, Conti E: Chloroplast DNA evi-dence for genome differentiation in wild potatoes (Solanumsect.Petota: Solanaceae). Am J Bot 78: 1354–1366 (1991).

232. Spooner DM, Anderson GJ, Jansen RK: Chloroplast DNAevidence for the interrelationships of tomatoes, potatoes, andpepinos (Solanaceae). Am J Bot 80: 676–688 (1993).

233. Sprent JI, Raven JR: Evolution of nitrogen-fixing symbioses.In: Stacy G, Burris RH, Evans HJ (eds), Biological NitrogenFixation, pp. 461–496. Chapman & Hall, New York (1992).

234. Stefanovic S, Jager M, Deutsch J, Broutin J, Masselot M:Phylogenetic relationships of conifers inferred from partial28S rRNA gene sequences. Am J Bot 85: 688–697 (1998).

235. Stebbins GL: Variation and Evolution in Plants. ColumbiaUniversity Press, New York (1950).

236. Stebbins GL: Chromosomal Evolution in Higher Plants.Edward Arnold, London (1971).

237. Strauss SH, Doerksen AH: Restriction fragment analysis ofpine phylogeny. Evolution 44: 1081–1096 (1990).

238. Sullivan, J: Combining data with different distributions ofamong-site rate variation. Syst Biol 45: 375–380 (1996).

239. Swensen SM: The evolution of actinorhizal symbioses: evi-dence for multiple origins of the symbiotic association. Am JBot 83: 1503–1512 (1996).

240. Swofford DL: When are phylogeny estimates from molecu-lar and morphological data incongruent? In: Miyamoto MM,Cracraft J (eds), Phylogenetic Analysis of DNA Sequences,pp. 295–333. Oxford University Press, New York (1991).

241. Swofford DL: Phylogenetic Analysis Using Parsimony∗(PAUP∗), version 4.0. Sinauer Associates, Sunderland, MA(1998).

242. Takhtajan A: System of Magnoliophyta. Academy of Sci-ences U.S.S.R., St. Petersburg (1987).

243. Takhtajan A: Diversity and Classification of FloweringPlants. Columbia University Press, New York (1997).

244. Tanksley SD, Bernatzky R, Lapitan NL, Prince JP: Conser-vation of gene repertoire but not gene order in pepper andtomato. Proc Natl Acad Sci USA 84: 6419–6423 (1988).

245. Thorne RF: Classification and geography of the floweringplants. Bot Rev 58: 225–348 (1992).

246. U N: Genome analysis inBrassicawith special reference tothe experimental formation ofB. napusand peculiar mode offertilization. Jpn J Bot 7: 389 (1935).

247. Vaillancourt RE, Weeden NF, Bruneau A, Doyle JJ: Chloro-plast DNA phylogeny of Old WorldVigna (Leguminosae).Syst Bot 18: 642–651 (1993).

248. Wagner WH: Biosystematics and evolutionary noise. Taxon19: 146–151 (1970).

249. Wagstaff SJ, Olmstead RG: Phylogeny of Labiatae andVerbenaceae inferred fromrbcL sequences. Syst Bot 22:165–177 (1997).

250. Warwick SI, Black LD: Molecular systematics ofBras-sica and allied genera (subtribe Brassicinae, Brassicaceae):chloroplast genome and cytodeme congruence. Theor ApplGenet 82: 81–92 (1991).

251. Waters ER, Vierling E: The diversification of plant cytosolicsmall heat shock proteins preceded the divergence of mosses.Mol Biol Evol 16: 127–139 (1999).

252. Weeden NF, Muehlbauer FJ, Ladizinsky G: Extensive con-servation of linkage relationships between pea and lentilgenetic maps. J Hered 83: 123–129 (1992).

253. Wendel JF: New World tetraploid cottons contain Old Worldcytoplasm. Proc Natl Acad Sci USA 86: 4132–4136 (1989).

254. Wendel JF, Doyle JJ: Phylogenetic incongruence: windowinto genome history and molecular evolution. In: SoltisDE, Soltis PS, Doyle JJ (eds), Molecular Systematics ofPlants II: DNA Sequencing, pp. 265–296. Kluwer AcademicPublishers, Boston (1998).

255. Wendel JF, Olson PD, Steward JMcD: Genetic diversity, in-trogression, and independent domestication of Old Worldcultivated cottons. Am J Bot 76: 1795–1806 (1989).

256. Wendel JF, Schnabel A, Seelanan T: Bidirectional interlocusconcerted evolution following allopolyploid speciation incotton (Gossypium). Proc Natl Acad Sci USA 92: 280–284(1995).

257. Wendel JF, Small RL, Cronn, RC, Brubaker, CL: Genes,jeans, and genomes: reconstructing the history of cotton.In: Proceedings of the International Organization of PlantBiosystematists, Amsterdam (in press).

258. White S, Doebley J: Genes and genomes and the origins ofmaize. Trends Genet 14: 327–332 (1998).

259. Whitkus R, Doebley J, Lee M: Comparative genome map-ping of sorghum and maize. Genetics 132: 1119–1130.(1993).

260. Woese CR: Bacterial evolution. Microbiol Rev 51: 221–271(1987).

261. Woese CR., Kandler O, Wheelis ML: Towards a naturalsystem of organisms: proposal for the domains Archaea, Bac-teria, and Eucarya. Proc Natl Acad Sci USA 87: 4576–4579(1990).

262. Wolf PG: Evaluation ofatpB nucleotide sequences for phy-logenetic studies of ferns and other pteridophytes. Am J Bot84: 1429–1440 (1997).

263. Wolf PG, Pryer KM, Smith AR, Hasebe M: Phylogeneticstudies of extant pteridophytes. In: Soltis DE, Soltis PS,Doyle JJ (eds), Molecular Systematics of Plants II: DNASequencing, pp. 541–556. Kluwer Academic Publishers,Boston (1998).

264. Wright RJ, Thaxton PM, El-Zik KM, Paterson AH: D-subgenome bias ofXcm resistance genes in tetraploidGossypium(cotton) suggests that polyploid formation hascreated novel avenues for evolution. Genetics 149: 1987–1996 (1998).