SystematicBotany (2004),29(2): pp. 251– 259 q Copyright 2004by the AmericanSociety of PlantTaxonomists Phylogeny ofHorsetails ( Equisetum)based on the Chloroplast rps4 Gene and Adjacent Noncoding Sequences JEAN-MICHEL GUILLON LaboratoireEcologie ,Syste´matique et Evolution. UMR 8079CNRS. Ba ˆtiment 360,U niversite´Paris-Sud,91405 Orsay-cedex,France (email: [email protected]) CommunicatingEditor: P aulS .Manos ABSTRACT. Equisetum isa genus of15 extant species thatare the sole surviving representatives ofthe class Sphenopsida. ChloroplastD NAsequence data were used toexamine the monophylyof the twoaccepted subgenera ( Equisetum and Hippochaete )and the putative basal position of Equisetumgiganteum .Aplastid DNAregionthat includes rps4 was sequenced forall species of Equisetum.Phylogenetic analyses usingparsimony ,likelihoodor posterior probability criteria, all support the followinginferences: (1) Equisetumbogotense isbasal withinthe genus; (2)all otherspecies group intw omajorsister clades: (i)the restof subgenus Equisetum (7 species) and (ii)subgenus Hippochaete (7 species). Onthe basis ofthe present phylogeny,bisexuality would notbe the ancestral state inthe Equisetum genus, but characters shared amongspecies ofthe subgenus Equisetum,suchas supercial stomata and protrudingantheridia, could be ancestral inthe genus. Horsetails ( Equisetum L.)are the only survivors of gametophytes, alsoshow aclear-cutdistin ction be- the formerly more diverse Sphenopsida,free-sporing tween the twosubgenera (Table1; Duck ett 1973, plants characterized by articulatestems bearing 1979a). whorls oflea ves ateach node .According tothe fossil Most living species ofhorsetails have very broad record, primitiveforms possibly assignable toSphen- circumborealdistribution s, their latitude ranging be- opsida appearedin the lateDevonian (Stewartand tween 408 and 608 north (Hauke1963, 1978). The ex- Rothwell 1993;T aylor and Taylor 1993).Sphenops ida ceptions are E. bogotense , E. giganteum , and E. myri- then reached maximum diversity during the Carbon- ochaetum,from Central and SouthAmerica, E. laeviga- iferous. Afterwards,major extinction episodes took tum,which is restricted toN orth America, E. diffusum, placeduring the early Permian and the lateJurassic. endemicofthe Himalayas,and E. ramosissimum , which Since the beginning ofthe Cenozoic,all known Sphen- ranges from Europe,Africa,and Asia tosome Oceanic opsida havebeen herbaceousforms thatare indistin- Islands. On accountof overlappin gdistributions, guishable from living horsetails (Brown 1975;McIver manyinterspecic hybrids, involving allspecies except and Basinger 1989;Stew artand Rothwell 1993).F ossils E. bogotense ,are found in the wild. These hybrids are attributedto the genus Equisetites thatclosely resemble consideredtobe sterile (butsee Krahulec et al.1996) modern Equisetum date from the Permian (Boureau and torely on vegetativereproduction forpersistenc e 1964;Stew artand Rothwell 1993),and possibly even and growth. Most ofthe hybrids have alsobeen ex- from the Carboniferous (Emberger 1968;T aylor and perimentally synthesized bycontrolled crosses (Duck- Taylor 1993).On accountof the similarity between ett 1979b).N otably,the pattern ofhybridiza tion in the Equisetites fossils and living Equisetum species, the dis- genus Equisetum canbe used todistinguish twodis- tinction made between these twogenera has been tinctgroups ofspecies. Within eachgroup, species are questioned tothe point thatArnold (1947)proposed connected by hybridization, whereas hybridization that Equisetum might beregarded asthe oldest surviv- cannotoccur between groups. This division based on ing vascularplant genus in the world. hybrid occurrence perfectly matches with the taxonom- Morphologicalvariability within living species and icdivision ofth egenus into the twosubgenera Equi- extensive hybridizationamong them has long ob- setum and Hippochaete (Duckett1979b). scured the taxonomy ofthe genus Equisetum. Since the The sectional classication of the subgenera Equi- workofH auke(1963,1978), which signicantly clari- setum and Hippochaete is more controversial. Different ed the picture,15species of Equisetum have been divisions ofthe subgenus Equisetum have been pro- widely accepted.The taxonomy ofthe genus Equisetum posed, depending on the charactersprioritized fortax- recognizes twosubgenera ( Equisetum and Hippochaete ; onomy. These charactersinclude (i)stem dimorphism Milde 1861),on the basisof stomatal position: super- (Braun 1839),(ii) en dodermal patterns (Ptzer 1867), cial in Equisetum (E. arvense, E. bogotense , E. diffusum, (iii) surfacemorpholo gy ofsilica deposits (Page 1972), E. uviatile , E. palustre, E. pratense, E. sylvaticum , and E. and (iv)antheridiu mmorphology (Duckett1973). Al- telmateia),butsunk en below the epidermal surfacein though there wassome correlationbetween Duckett’s Hippochaete (E. giganteum , E. hyemale, E. laevigatum , E. observations concerning antheridiummorphology and myriochaetum , E. ramosissimum , E. scirpoides , and E. var- the classication proposed by Page (1972),Ha uke iegatum).Other characters,such asthe morphology of (1974)considere dthatthe availableinformati on was 251 252 SYSTEMATIC BOTANY [Volume 29 TABLE 1. Characters differingbetw een the twosubgenera Eq- and Nickrent 1999),or (ii)both n uclearand mitochon- 1 2 uisetum and Hippochaete. except E. bogotense and E. diffusum. drial small-subunit rDNAsequences(Renzaglia et al. except E. laevigatum, and E. ramosissimum incolder climates. 3 ex- cept E. ramosissimum,E. myriochaetum and E. giganteum. 4 except E. 2000),or (iii) plastid atpB, rbcL, and rps4 and nuclear myriochaetum and E. laevigatum. Data fromDuckett (1973, 1979a), small-subunit ribosomal DNAsequences(Pryer et al. Hauke (1963, 1968, 1969, 1974, 1978) and Page (1972). 2001)also concluded thathorsetails and ferns are a monophyletic group and the closest living relatives to Characters Equisetum Hippochaete seed plants.The exactposition of Equisetum with re- Sporophyte spectto ferns varies among the studies, some placing stem Equisetum asa sister clade toleptosporan giate ferns persistence annual1 perennial2 morphology branched unbranched3 (Duffand Nickrent 1999;Renzaglia et al.2000), some cone apex rounded pointed4 weakly supporting the grouping of Equisetum with silica deposits ornamented amorphous Marattiopsida(Pry er et al.2001), and some placing Eq- position ofstomata supercial sunken uisetum asthe sister group toall ferns (Kenrick and Gametophyte Crane 1997). lamellae morphology unistratose column In contrast,phylogen etic relationships within the antheridia genus Equisetum have remained largely unexplored. position protruding sunken Evolutionin Equisetum wasdiscussed by Schaffner cover cell number usually .2 usually 2 (1925,1930) and Hauke(1963),although acladistic Chromosomesize small large analysis ofm orphologicalcharacters has not yet been undertaken in horsetails. Considering that E. giganteum has large stems and persistent sheath teeth, asdo most insufcient toform ulatea sectional classication of the fossil members ofthis group,and becauseit has reg- subgenus Equisetum.Later,Hauke(1978)constructed a ularwhorls ofbranches and stomatain manylines, as phenetic dendrogrambased on pairwise comparisons found throughoutthe subgenus Equisetum, both au- ofspecies forman ycharacters.A ccording toHauke ’s thors concluded that E. giganteum wasth emost prim- results, only onegroup is worth recognizing; this in- itive member ofthe genus.On the basisof the sup- cludes E. arvense, E. pratense, and E. sylvaticum . In con- posed primitive statusof E. giganteum ,the most ad- trast,from the pattern ofh ybridizationwithin the sub- vanced members ofthe subgenus Hippochaete were genus Equisetum,Duckett (1979b)suggested that E. proposed tobe E. laevigatum and E. scirpoides (Schaffner pratense and E. sylvaticum would form aunit separate 1925;Hauke 1963). from the remaining species. Besides the evolutionofsporophy ticcharacters, the Aclassication of the subgenus Hippochaete has also phylogeneticstatusof E. giganteum has implications for been proposed by Hauke(1963).Section Incunabula the evolution ofsexuality in Equisetum gametophytes. contains the regularly branched E. giganteum , section Gametophytes ofhorsetails are usually unisexual: ga- Hippochaete contains species with evergreen, un- metophytes rst develop asmale or female,with fe- branched aerial stems ( E. hyemale, E. variegatum , and E. male gametophytes later producing antheridia,when scirpoides),and the section Ambigua contains those spe- not fertilized (Duckett 1970,1972). Interestingl y,ga- cies with intermediate morphology ( E. myriochaetum , metophytes of E. giganteum are functionally bisexual E. ramosissimum , and E. laevigatum ).This division is (i.e.,producing antheridia and archegoniasimulta- partly supported byexperimen taldata on hybridiza- neously), aphenologythatis unique in the genus tion (Duckett 1979b),but the morphology of Hippo- (Hauke1969, 1985). Because he considered E. gigan- chaete gametophytes shows no clear discontinuities teum tobe primitive ,Haukehypothesizedthatbisex- that tthe sectional division (Duckett 1979a).M ore- uality couldhave been the primitive condition in Eq- over, the frequent occurrence ofhybrids within sub- uisetum.In contrast, E. bogotense gametophytes seem to genus Hippochaete suggests thatthe taxaare more bestrictly unisexual (i.e.,never showing sex reversal closely interrelatedthan in subgenus Equisetum. from archegoniato antheridia production),aphenol- The