Cryptic Species in the Puccinia Monoica Complex Author(S): Barbara A

Home , Puccinia monoica, Schoenocrambe linifolia

Mycological Society of America

Cryptic Species in the Puccinia monoica Complex Author(s): Barbara A. Roy, Detlev R. Vogler, Thomas D. Bruns and Timothy M. Szaro Reviewed work(s): Source: Mycologia, Vol. 90, No. 5 (Sep. - Oct., 1998), pp. 846-853 Published by: Mycological Society of America Stable URL: http://www.jstor.org/stable/3761326 . Accessed: 10/12/2012 12:55

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at . http://www.jstor.org/page/info/about/policies/terms.jsp

. JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship. For more information about JSTOR, please contact [email protected].

Mycological Society of America is collaborating with JSTOR to digitize, preserve and extend access to Mycologia.

http://www.jstor.org

This content downloaded by the authorized user from 192.168.52.70 on Mon, 10 Dec 2012 12:55:26 PM All use subject to JSTOR Terms and Conditions Mycologia,90(5), 1998, pp. 846-853. © 1998 by The New YorkBotanical Garden, Bronx,NY 10458-5126

Crypticspecies in the Puccinia monoicacomplex

Barbara A. Roy' asites and theirhosts are intimatelyconnected, it is Geobotanical Institute, Swiss Federal Institute of assumed that theirevolution should also be parallel. Technology(ETH), 8044 Zurich,Switzerland Second, it is ofteneasier to distinguishhosts than to Detlev R. Vogler visuallyseparate differentspecies of parasites since ConservationGenetics Laboratory, Biology Department, the shared parasitichabit leads to morphologicalre- San FranciscoState University,San Francisco, duction and convergence(Price, 1980; Barrett,1986; California,USA 94132 Brooks and McLennan, 1993). However,there are a Thomas D. Bruns numberof reasonswhy host and parasitephylogenies TimothyM. Szaro may not alwaysbe the same, including: (i) a lack of variationin the host for resistanceor in the Departmentof Plant and MicrobialBiology, University genetic of California,Berkeley, California, USA 94708 pathogen forvirulence (Barrett, 1986; Futuymaet al., 1995), (ii) differentrates of evolutionbetween host and parasite (Savile, 1971), (iii) adaptation in only Abstract: The Puccinia monoicacomplex is an enig- one memberof the pair (Thompson, 1986), and (iv) matic group of rust fungi. They are flowermimics, host shiftsto unrelatedspecies (Savile, 1971; Funk et and theygreatly reduce host reproductionand sur- al., 1995; Radtkeyand Singer,1995). vival. These fungi are relativelycommon, attacking Modern techniquessuch as DNA sequencing com- approximately960 species in 11 genera of crucifers bined withphylogenetic analysis make it possible to as well as at least fivegenera of grasses.In modern bypassthe problem of morphologicalconvergence in taxonomictreatments the Puccinia monoicacomplex the pathogens, and permit an independent assess- is treated as four species that are differentiatedby ment of the assumptionthat host associationsdefine the number of spore statesin theirlife cycles.How- specificlevel differencesin pathogens (Brooks and ever, other systematictreatments have divided the McLennan, 1993). In this studywe used sequence group into species or formsbased on host associa- data to ask whetherthere are likelyto be crypticspe- tion. Withinthe species based on spore statethere is cies in a particularlyenigmatic group of rusts,the morphologicalvariation, but it has not been readily flower-mimickingPuccinia monoicacomplex. Cryptic assignable to either host species or geographic area. species are morphologicallyindistinguishable but are We used DNA sequencing and phylogeneticanalysis not interfertile,and they often differin nonmor- to determinewhether there are crypticspecies in this phological characterssuch as DNA sequence diver- thatare not evidentwhen is group onlymorphology gence (Sivarajan and Robson, 1991; Soltis et al., used. We sequenced the nuclear rDNA region con- 1992). Crypticspecies differfrom biologic races as + 5.8S tainingthe internaltranscribed spacers (ITS-1 defined by Caten (1987) or Jaenike (1981) by being + gene ITS-2) of isolates fromdifferent hosts. Our reproductivelyisolated. there in the resultsindicate that are crypticspecies The biologyof the cruciferrusts suggests that these Puccinia monoica and that in this complex, species rustsare strongselective agents on the hostsand that cannot be identified life group strictlyby cyclestage. sexual reproductionin the fungusis likelyto be im- Words: molecular Key Arabis,mimicry, phyloge- portant in the evolution of the fungi. Infectionof ny, Puccinia thlaspeos,rust fungi, Tranzschel's Law Brassicaceoushosts occurs in late summerfrom wind- borne basidiospores (Roy and Bierzychudek,1993). Within a few months of basidiospore germination INTRODUCTION and penetrationof host leaves,fungal hyphae invade meristematictissues and cause infection.In- Classificationof parasites and pathogens is often systemic based on host association.There are at least tworea- fectedhost plants produce one to severalflower-like sons forthis mode of classification.First, because par- rosettes(pseudoflowers) in the spring,but rarelypro- duce true flowers(Roy, 1993a, 1994). Spermogonia, Accepted for publicationApril 6, 1998. the outcrossingstructures of the rust funguswhich 1 Email: [email protected] containspermatia and receptivehyphae, form on the

846

This content downloaded by the authorized user from 192.168.52.70 on Mon, 10 Dec 2012 12:55:26 PM All use subject to JSTOR Terms and Conditions RoY ET AL.: CRYPTIC SPECIES OF PUCCINIA 847

TABLEI. Life cycleinformation for the Puccinia monoicacomplexa

AlternatesAlternates Spore stages Pathogen spp. to a grass? Basidia Telia Spermatia Aecia Uredia P monoicaArth. yes X X X X X P consimilisEllis & Everh. unknown X X X X 0 P thlaspeosC. Schub. no X X X 0 0 P. holboelliiHornem.b no X X 0 0 0

a Data from:Gaumann, 1959; Arthurand Cummins,1962; Savile 1974a, b and personal observationsby B. A. Roy. X = present,0 = absent. b Note that there are both a host and a funguswith the same specificepithet: P holboelliiand A. holboellii.In thispaper we presentdata on the fungiP monoicaand P thiaspeoson A. holboellii,but presentno data on the fungusP holboelliion any hosts. surface of the pseudoflowers. Insects are required to monoicacomplex are differentiatedtaxonomically by transfer spermatia to receptive hyphae belonging to the numberof spore stagesin theirlife cycles (TABLE a different mating type, a process that is analogous I); in the past theyhave been divided into species or to flowerpollination by insects (Roy,1993a; Roy and formsbased on host association (e.g., Jorstad,1932; Raguso, 1997). Observations and experiments on Gaumann, 1959). Morphological variation within three of the four species (P. monoica, P thlaspeos,and species, as defined by spore states, exists but it is not P consimilis)show thatinfection substantially increas- easilyclassifiable. For example, in referenceto P hol- es host mortalityand almost alwayseliminates host boellii,D. B. 0. Savile wrote, "In this materialmor- reproduction (Roy, 1993a, b; Roy and Bierzychudek, phological variabilityseems to be random ratherthan 1993). geographic or by host" (Savile, 1974a). The Puccinia monoicacomplex is currentlycom- prised of 4 named species (TABLE I). Hosts include MATERIALS AND METHODS approximately960 species in 11 genera of the mus- tard family,Brassicaeae, and at least five genera of Fungal isolates and extraction.-TABLE II lists the rust grasses (Gaumann, 1959; Arthur and Cummins, fungus species, isolate number, host species, place of 1962; Farr et al., 1989). Today members of the P collection,collection source, and Genome Sequence

TABLEII. Fungal collectionsexamined, host species of collection,genome sequence data bank accession number (GSDB#), and collectioninformation

Rust species GSDB# Host species County State Collector/# Date Puccinia thlaspeos S:1216225 Arabis breweriS. Watson Contra Costa California Roy/F549 10 May 96 S:1216218 Arabisdemissa Greene Gunnison Colorado Roy/F427 24 May 93 L76183 Arabisgunnisoniana Rollins Gunnison Colorado Roy/F489 9June 93 L76177 Arabis holboelliiHornem. Gunnison Colorado Roy/F363a 9 June 93 S:1216227 Polycteniumfremontii(Watson) Horney Oregon Roy/F610b 6 May 96 Greene Puccinia consimilis L76181 Arabis crandalliiRobinson Gunnison Colorado Roy/F443 25 May 93 Puccinia monoica S:1216223 Arabisdemissa Greene Pitkin Colorado Roy/F510 1 May 96 L76180 Arabis drummondiiGray Summit Utah Vogler/Pumol 1994 S1216228 Arabisdrummondii Gray Gunnison Colorado Roy/F620 June 96 S:1216222 Arabishirsuta (L.) Scopoli Gunnison Colorado Roy/F432 9 June 93 L76182 Arabis holboelliiHornem. Gunnison Colorado Roy/F315' 5 June 93 S:1216224 Arabisperennans S. Watson Grand Utah Roy/F534 2 May 96 S:1216226 Arabissparsiflora Torrey & Lassen California Roy/F587 11 May 96 Gray L76176 Schoenocrambelinifolia (Nutt.) Gunnison Colorado Roy/F289 1 June 93 Greene

a Host correspondsto clone "B4" described in Roy,1993b. b Collected by B. Ertter,May 6, 1996; the collectionnumber is Ertter#14623. Voucher specimen deposited at UC. c Host correspondsto clone "Dl" describedin Roy,1993b.

This content downloaded by the authorized user from 192.168.52.70 on Mon, 10 Dec 2012 12:55:26 PM All use subject to JSTOR Terms and Conditions 848 MYCOLOGIA

Data Base (GSDB) accession numbers. Infected disticanalyses. Sequences were deposited in the Ge- leaves were collected into paper coin envelopes, nome Sequence Data Base (TABLE II). dried at room temperature,then stored at 4 C in Phylogeneticanalysis.-Phylogenetic analysis of the plasticcontainers lined withDri-Rite. For extraction, Puccinia isolates from crucifersincluded sequence we crushed 4 mm2 pieces of infectedleaves in 2X data fromITS spacers 1 and 2 and the associated5.8S CTAB buffer,following the method of Gardes and and 28S genes. For comparativeand outgroup pur- Bruns (1993). poses we included in our firstanalysis published se- Zambino and Szabo fromthe ce- PCR amplification,cleaning, and sequencing.-The quences by (1993) nuclear rDNA region containing the internal tran- real rusts.We designated the cereal rustsas a para- We felt scribed spacers (ITS-1 and ITS-2) and the 5.8S gene phyleticoutgroup to the cruciferrusts. justi- fied in this choice because the was amplified by the polymerase chain reaction outgroup sequences were and (PCR) on a Techne thermalcycler with the following alignable, preliminaryanalyses including tree rusts shown) indicated thatthe cereal rusts primerpairs: ITS-5 (White et al., 1990) and ITS-4b (not clustered from the cruciferrusts. Unfor- (Gardes and Bruns, 1993) and the internalprimers separately tunately,the published sequences forthe cereal rusts ITS-2r (Vogler and Bruns, 1998) and ITS-3r (Vogler includes the last 40 bases of ITS-1. Because the and Bruns,1998). A waterblank was included in each only unsequenced section is coded in Paup DNA analysis PCR reactionset. The thermocyclerprogram for DS- as unknowns,it causes a branch-lengthartifact in the amplificationwas as described by Gardes and Bruns final phylograms.We thereforeused all of the avail- (1993). able data to estimatethe tree,but then correctedthe All of the cruciferrusts except one were sequenced branch lengthsby using only the alignable data com- on an ABI (=Applied 310 automaticse- Biosystems) mon to all taxa when the tree was printed (i.e., the one sample (P monoica,Pumol fromUtah) quencer, last 40 bp of ITS-1, and all of the 5.8S and ITS-2). was on an ABI 377. Prior to sequenced sequencing, This approach correctsthe branch-lengthartifact in the final PCR was cleaned witha product QIAquick the phylogram,and preservestopology based on the PCR kit cleaning (Qiagen #28104).Cycle sequencing available data fromthe ITS-1. The alignmentis avail- reactions used 8 FIL TerminatorPremix (Perkin El- able fromTreeBASE. mer), 1 FLLprimer (3.2 mM), 11 FiL clean ds DNA In a second set of analyseswe used data onlyfrom + H20 for a total volume of 20 FL (some reactions the cruciferrusts. We used this subset because, al- were done with 1/2these amounts). The cycle se- though it was possible to align the cereal rusts,the quence programwas: 25 cyclesof 94 C (30 sec)/50 alignmentnecessitated the inclusion of many gaps. C (15 sec)/60 C 4 min. To removeexcess terminator In addition,we had more data for the cruciferrusts: dye, the cycle sequence reactionswere run through all of the ITS-1, 5.8S, and ITS-2 as well as the first CentrisepSpin Columns. 120 bp. of the 28S gene. By analyzingthe crucifer We used the programsSequence Navigatorv. 1.0.1 rusts separatelywe were able to place more confi- and SeqEd. v.1.0.3 (Applied Biosystems)to align se- dence on the branches because the alignmentswere quences through a combination of clusteringalgo- tighterand more complete. rithmsand visual editing.Sequences were verifiedby Heuristic searches were performedwith both the checkingeach sequence againstits complement (i.e., random-addition sequence option (10 replicates) ITS 5 + 2r, 3r + 4b), and by sequencing multiple and with the simple addition option in Paup 3.1 samples of the same accession (all cases), and by se- (Swofford,1993). Both methods of searching re- quencing multiplespecimens. We sequenced twoiso- vealed the same trees. Bootstrap support (Felsen- lates of P monoicaon Arabisdrummondii from differ- stein, 1985) for branches was based on 1000 repli- ent populations. The Utah sample differedfrom the cations.To determinethe relativefrequency of trans- Colorado sample by havingTC at bases 493 and 494 versions(tv) and transitions(ti) we calculated the ac- instead of YY (ambiguous,Y = T or C). We also se- tual ratios in MaClade (Maddison and Maddison, quenced five samples of P thlaspeoson A. demissa 1995). The average ratio of ti/tvwas veryclose to 1: fromone population. The sequences fromA. demissa 1 for all trees. We thus performed unweighted were identical,with the exception of one ambiguous searches. (T or C, coded as Y) base in one sample. Because therewas littledifference among multiplesequences RESULTS fromthe same host/funguscombination from different individuals (1-2 bp/approx. 570 bp, << 0.01% In phylogeneticanalyses of the cruciferrusts includ- difference),we used a consensussequence in the cla- ing the cereal rustsas an outgroup,both the ITS-1

This content downloaded by the authorized user from 192.168.52.70 on Mon, 10 Dec 2012 12:55:26 PM All use subject to JSTOR Terms and Conditions RoY ET AL.: CRYPTIC SPECIES OF PUCCINIA 849

P. thlaspeos on Arabisgunnisoniana P. thlaspeos on Arabis holboellii P. monoica on Arabisperennans P. thlaspeos on 58 P. monoica on Arabis demissa Arabis demissa P. consimilison . P. thlaspeos on Arabis holboellii Arabis crandallii P. thlaspeos on Arabis demissa GroupI P. monoica on P. consimilis on Arabis crandallii Arabis demissa 85 monica on Arabis sparsiflora r P. _P. monoica on ( 3roupI 63 . Arabis perennans L_ P. thlaspeos on Arabis breweri I P. thalaspeos on Arabis gunnisoniana - P. thlaspeos on Polycteniumfremontii P. on I P. monoica on Arabis holboellii thlaspeos Arabis breweri 100 P. monoica on Arabis hirsuta 36 P. monoicaon |56F| P. monoica on Arabis sparsiflora Arabis drummondii P. monoica on Schoenocrambe linifolia 'P. thlaspeos on Polycteniumfremontii cciniahordei P. monoica on ' Uromycesscillarum Arabis drummondii P. monoicaon PuJccinia graminis Cereal Rusts 1.....0..... Arabisholboellii Group 11 Puccinia mesnieriana 100 _P. monoicaon Arabis hirsuta Puccinia coronata 'P. monoica on Schoenocrambe linifolia FIG. 1. Phylogramfrom heuristic analysis of cruciferrust DNA data includingsome cereal rustsas an outgroup.The FIG.2. Phylogramfrom heuristic analysis of cruciferrust topologyand bootstrapanalyses were based on the entire DNA data including the entire ITS-1, 5.8S, and ITS-2 ITS region,but the tree was generatedusing only the data regionsand 120 bp of the 28S gene. The analysisyielded 3 fromthe last 40 bp of the ITS-1 plus the entire 5.8S and trees of 50 steps, CI = 0.94, RI = 0.95, RC = 0.89. Here ITS-2 regionsto correctfor branch length artifacts resulting we show one of the three mid-pointrooted trees that re- from incomplete data from the cereal rusts.The analysis sulted from a heuristicsearch with the random addition = = = yielded6 treeswith length 156, CI 0.87, RI 0.87, RC sequence option. The particularphylogram shown is iden- = 0.75. Here we show the one of the six from phylograms ticalto the singlephylogram that results from the strictcon- a heuristicsearch. Bootstrapvalues for the branches are sensus option. Bootstrapvalues for the branchesare based based on 1000 replicates. on 1000 replicates. and ITS-2 contained phylogenetically informative lograms, Group 2, which contains only P monoica iso- sites. Our alignment of all taxa yielded six trees of lates, is characterized by deletions relative to the oth- 156 steps, and a CI of 0.87. In FIG. 1 we show one of er taxa and these sequences are the shortest (552- the six phylograms. The illustrated tree has the same 558 bp). Collections originally identified as P. thlas- topology as the tree produced by the strictconsensus peos were all 572-575 bp. All of the variation in option. The topology of the six trees differed in the length among isolates was in the spacers; the 5.8S arrangement of the poorly differentiated terminal gene was always 158 bp (TABLE III). In content, the taxa in Group 1, and in the placement of P monoica 5.8S sequence was very similar among all of the cru- on Schoenocrambe linifolia (which either clustered cifer rusts, but there were a few informative differ- with Group II, or was found alone). ences, mostly in Group 2 (FIG. 2). In phylogenetic analyses of just the crucifer rusts, Puccinia monoica is probably not monophyletic. both the ITS-1 and ITS-2 were informative. Paup Two well supported clusters of taxa contain different DNA analysis including all characters, and in which entities called P monoica (FIGS. 1, 2). One well sup- gaps were treated as missing data, yielded 3 trees of ported cluster of "P monoica" (FIGS. 1, 2, Group 2) 50 steps and a CI of 0.94. In FIG. 2 we show one of is likely to contain at least three species: one on Ar- the three trees. The illustrated tree has the same to- abis drummondii,one on Schoenocrambelinifolia, and pology as the tree produced by the strict consensus one that occurs on both A. holboelliiand A. hirsuta. options. The three trees differed only in the arrange- This group of Puccinia species is differentiated from ment of the poorly differentiatedterminal taxa in the the rest of the sequences by deletions of 10-15 bp top clade of group one. (TABLE III). In addition to the well defined group of In addition to variation in sequence content, there species with deletions (FIG. 2, Group 2), there are was variation in sequence length, with sequences also "Puccinia monoica" rusts that are indistinguish- ranging from 559 to 573 bp (TABLE III). In the phy- able from P thlaspeosin length and that cluster tight-

This content downloaded by the authorized user from 192.168.52.70 on Mon, 10 Dec 2012 12:55:26 PM All use subject to JSTOR Terms and Conditions 850 MYCOLOGIA

TABLEIII. Variationin sequencelength (bp) amongthe crucifer rusts Collection# Host Rust ITS 1 5.8S ITS 2 Total

F549 Arabis breweri P thlaspeos 211 158 203 572 F427 Arabisdemissa P thlaspeos 212 158 203 573 F489 Arabisgunnisoniana P thlaspeos 212 158 203 573 F363 Arabisholboellii P thlaspeos 212 158 203 573 F610 Polycteniumfremontii P thlaspeos 212 158 204 574 F443 Arabiscrandallii P consimilis 211 158 205 574 F510 Arabisdemissa P monoica 211 158 203 572 F620 Arabisdrummondii P monoica 201 158 199 558 F432 Arabishirsuta P monoica 201 158 197 556 F315 Arabisholboellii P monoica 196 158 198 552 F534 Arabisperennans P monoica 211 158 203 572 F587 Arabissparsiflora P monoica 211 158 203 572 F289 Schoenocrambelinifolia P monoica 202 158 205 565

lywithin P thlaspeos(Group 1). The phylogramssug- produce chemically distinctfloral fragrances(Roy gest that one of these P monoicathat is not charac- and Raguso, 1997; Raguso and Roy,in press). terizedby deletionsis also a separatespecies: the tax- It is often asserted that pathogens are good tax- on thatoccurs in Californiaon A. sparsifloraand that onomists and that thereforedifferent host species is indistinguishableby ITS sequence from the "P should have a differentcomplement of pathogensas- thlaspeos"on Arabis breweri(FIGS. 1, 2). sociated with them (Barrett, 1986). Although our Puccinia thlaspeosmay also be polyphyleticbut data shows that there is some association between there are smaller sequence differencesamong the cruciferhost and fungus species, it is not perfect. short-cycledspecies than among the long-cycledP That is, each cruciferhost species does not have a monoicataxa (FIG. 2). In FIG. 2, Group 1 containsall separate fungal species associated withit. For exam- the microcyclicP thlaspeosisolates plus some isolates ple, we were unable to differentiateamong isolates thatare morphologicallyP monoica.Distances within of P. thlaspeoson three differentspecies of Arabis Group 1 are all short,suggesting recent derivation or (FIG. 2: A. holboellii,A. gunnisoniana,and A. demis- a slow rate of evolution.Group 2, on the other hand, sa), or betweenisolates of P monoicaon twodifferent which contains only macrocyclicisolates of P monoi- Arabishosts (FIG. 2: A. hirsutaand A. holboellii). ca, is separate from Group 1 and is relativelymore When pathogens attack more than one host it is differentiated. likelythat both hosts influence the evolutionof the pathogen.The pathogensin the Puccinia monoicalife cycle group (Group 2, FIGS. 1, 2) are both macro- DISCUSSION cyclicand heteroecious and thus have unrelated ae- Crypticspecies ?-The ITS sequence data suggestthat cial and telial hosts,in this case withinthe Brassica- there are indeed crypticspecies withinboth P. mon- ceae and Poaceae, respectively.Some of the observed oica and P thlaspeossince there are significantse- differentiationin the macrocyclicisolates of P mon- quence differencesamong collectionstaken from dif- oica may be associated with differencesin the telial ferenthosts (FIGS. 1, 2; TABLEIII). Not only were (i.e., grass hosts); if true,this would parallel the pat- there differencesin sequence length and composi- tern found in the cereal rusts. Within P monoica tion among rust isolates on differenthost genera therewas differentiationamong isolateson A. drum- (FIG. 2: isolateson Polycteniumand Schoenocrambever- mondiiversus A. holboelliiand A. hirsuta(FIG(. 2). This sus Arabis) but there were also differencesamong differentiationwas not entirelyunexpected as it is rustisolates on the same host genus (e.g., differences likely that rust isolates colonizing A. drummondii between isolates on ArabisFIGS. 1, 2). Althoughmat- have a differenttelial host.This suspicionis based on ing studieswere not performedto verifyspecies sta- three pieces of information:(i) In the RockyMoun- tus, the sequence divergencebetween some of our tains,A. drummondiiis found in the montane-alpine samples was similar to that between well character- zones in associationwith the grass Trisetumspicatum ized cereal rust species of commercial importance (L.) Richter,whereas A. holboelliiand A. hirsutaare (FIG. 1). In addition to differencesin ITS sequences found in submontane-montaneareas in association and host use, manyof these flower-mimickingfungi with the grass,Koeleria macrantha (Ledeb.) Schultes

This content downloaded by the authorized user from 192.168.52.70 on Mon, 10 Dec 2012 12:55:26 PM All use subject to JSTOR Terms and Conditions ROY ET AL.: CRYPTIC SPECIES OF PUCCINIA 851

(B. A. Roy,pers. obs.). (ii) Where the distributionof species verywell, (ii) these particularcollections rep- the Arabishosts overlap,infection of A. drummondii resent very recent speciation events and sufficient onlyoccurs when Trisetumis present (B. A. Roy,pers. ITS differenceshave not yet accumulated to distin- obs.). (iii) Arthurperformed inoculation studies with guish them,or (iii) some of the collectionswere mis- spores fromTrisetum, Koeleria, and Arabis.His results identified,(iv) there is polymorphicexpression of led him to suggestthat there were "biological races" spore states. Of these hypotheses,1, 2 and 4 seem (Arthur,1912, 1915). Unfortunately,Arthur was not the most likely.We think hypothesis3, misidentifi- confidentabout his Arabisidentifications, so it is dif- cation, is unlikelyin our studybecause most of the ficultto ascertainwhich Arabisspecies were infected infectedplants analyzed were collected by the first by spores fromwhich grass.However, from the local- author who carefullyexamined the plants for spore itydata given in Arthur'spapers, the host pairs are stages at the time of collectionand again beforepre- likelyto have been A. holboellii+ Koeleriamacrantha, paring the material for sequencing. However, mis- and A. drummondii+ Trisetumspicatum. identificationis possible if a cruciferwas infectedby Some of the variationin sequences can also be cor- P. consimilis,and there were veryfew telia present. related with geography and habitat. For example, In this case, telia mightbe overlooked,and the fun- thereis a reasonablywell supportedcluster of P. mon- gus would be misidentifiedas P monoica.Alternative- oica on A. sparsifloraand P thlaspeoson A. breweri ly, there may be polymorphicexpression of spore (FIGS. 1, 2). These collectionswere made in North- stages in P consimilis.That is, if sometimesP consi- ern Californiawhereas the rest of the cruciferrust milis produces just aecia, sometimesjust telia and collectionswere made in the RockyMountains. It is, sometimes a mixture of the two. Polymorphicex- of course, difficultto unlink the influencesof geog- pressionof spore stagesin the demicyclicP. consimilis raphyfrom host range, and the sample size is small, seems very likely because the expression of spore but we should bear in mind that other evolutionary stagesin thistaxon is veryvariable. We have observed processes besides host, such as genetic drift,may in- considerablevariation in the ratio of telia: aecia pro- fluence differentiationamong taxa. duced on plants infectedby P consimilis(B. A. Roy and M. Pfunder,unpubl. data). Sometimesonly a sin- in Life-cycles taxonomyand evolution.-The conven- gle teliummay be produced among hundreds of ae- tion of differentiatingcrucifer rust species by life-cy- cia. Thus it seems likelythat under some environ- cle is not supportedby the molecularphylogeny. The mental or host compatibilityconditions only aecia macrocyclic"Puccinia monoica" and microcyclic"P maybe expressed on a plant,making it appear to be thlaspeos"species do not formmonophyletic groups infectedby P. monoica,when in factit is infectedby (FIGS. 1, 2). Nor did macrocyclicand microcycliciso- P consimilis. lates from the same host species always cluster to- If there is polymorphicexpression of spore states, gether (e.g. P monoicaand P thlaspeoson Arabishol- it probablydoes not occur in all taxa of the P. mon- boellii,FIG. 1). However, this last result should be oica complex. For example, the firstauthor has ob- treated with some caution because, although the served Puccinia monoica-infectedArabis drummondii same "species" of hostwas involved,the isolateswere throughoutits range over a seven-yr.period and has taken fromdifferent clones of an apomictic species, never seen anythingbut aecia produced on it. Other Arabis holboellii.The two clones could be treatedas hosts,however, are likelyto be infectedby polymor- separate microspeciesfor the followingreasons: (i) phic rustsmost, or all, of the time.For example, the Because these clones are apomictic theyare repro- firstauthor oftensees all three common "life cycle" ductivelyisolated fromeach other; there is no inter- rust species (P monoica,P thlaspeos,and P consimi- breeding among clones , (ii) clones of the D and B lis) in the same population of Arabis holboellii.And, isozyme phenotypes have differentchromosome A. crandalliiand A. lignifera,which are ofteninfected numbers (Roy, 1995), (iii) clone D is never (within by P consimilis,exhibit varying degrees of expression statisticalerror of clone identification)infected by P. of aecia and telia. thlaspeos,and Clone B is rarelyinfected by P monoica In summary,we have shown that there is enough (Roy and Bierzychudek,1993). variationat the ITS sequence level withinthe named Species withdifferent life cycles clustered together species of P. monoicaand P thlaspeosto suggestthat because there was littleor no sequence divergence additional species-leveltaxa exist. In addition, we correlated with life cycle type. Similar resultswere have called into question the practice of separating also recentlyfound forother cruciferrusts (Kropp et rust fungi into species based on life cycles. Rather al., 1997). We have severalhypotheses for why there than formingan evolutionaryseries, life cyclediffer- may be a lack of differentiationamong life cycle ences may be relativelyunstable and polymorphicin types: (i) life cycle differencesmay not define rust expressionin some taxa. Beforeundertaking the nec-

This content downloaded by the authorized user from 192.168.52.70 on Mon, 10 Dec 2012 12:55:26 PM All use subject to JSTOR Terms and Conditions 852 MYCOLOGIA essary taxonomic revisions in this group we are fol- Gardes, M., and T. D. Bruns. 1993. ITS primerswith en- lowing up on the study described herein with more hanced specificityfor basidiomycetes-applicationto extensive surveys, both morphological and molecu- the identificationof mycorrhizaeand rusts.Molec. Ecol. lar, of the crucifer rusts and their hosts across their 2: 113-118. Gaumann,E. 1959. Die Mitteleuropasmit besonderer geographic ranges. We plan to use the resulting phy- Rostpilze Beriicksichtigungder Schweiz. Buchdruckerei Buchler & logenetic reconstructions to address questions con- Co., Bern, Switzerland.1407 pp. cerning life-cycle plasticity,speciation, and the evo- Jaenike,J. 1981. Criteriafor ascertainingthe existence of lution of fungal floral mimicry. host races. Amer.Naturalist 117: 830-834. Jorstad,I. 1932. Notes on Uredinae. Nyt.Mag. Naturvidensk ACKNOWLEDGMENTS 70: 325-408. Kropp, B. R., D. R. Hansen, P. G. Wolf,K. M. Flint,and S. The authors would like to thank M. Gardes and B. V. Thomson. 1997. A studyon the phylogenyof the Ertter for collections, David Williams for locality in- Dyer'sWoad rustfungus and other species of Puccinia formation and enthusiasm, D. Brada, S. Burda, A. fromcrucifers. Phytopathology 87: 565-571. Kretzner, and M. Pfunder for their help with se- Maddison,W. P., and D. R. Maddison. 1995. MacClade, ver- quencing, anonymous reviewers for their comments, sion 3.05. Sinauer,Sunderland, Massachusetts. and the financial assistance of the Swiss National Price, P. W. 1980. Evolutionarybiology of parasites.Mono- in Fund (3100-046865.96 to B. A. Roy), and the U. S. graphs Population Biology. Princeton University New 237 National Science Foundation, SystematicBiology Pro- Press,Princeton, Jersey. pp. R. R., and M. C. 1995. reversals gram (BSR-8918454 and DEB-9221472 to T. D. Radtkey, Singer. Repeated of host-preferenceevolution in a specialistinsect her- Bruns). bivore.Evolution 49: 351-359. Raguso, R. A., and B. A. Roy.In press. "Floral" scent pro- LITERATURE CITED duction by Puccinia rustfungi that mimic flowers. Mo- lec.Ecol. Arthur, C. 1912. Culturesof Uredineae in 1911. J. Mycologia Roy,B. A. 1993a. Floral mimicryby a plant pathogen. Na- 4: 49-65. ture362: 56-58. . 1915. Cultures of Uredineae in 1913, and 1912, . 1993b. Patternsof rust infectionas a functionof 1914. Mycologia7: 61-89. host genetic diversityand host densityin naturalpop- , and G. B. Cummins. 1962. Manual of therusts in ulationsof the apomicticcrucifer, Arabis holboellii. Evo- UnitedStates and Canada, witha newsupplement. Haf- lution47: 111-124. ner Pub. Co., New York.438 pp. . 1994. The effectsof pathogen-inducedpseudoflow- Barrett,J. A. 1986. Host-parasiteinteractions and systemat- ers and buttercupson each other's insect visitation. ics. Pp. 1-17. In: Coevolutionand systematics.Eds., A. 75: 352-358. R. Stone and D. L. Hawksworth.Clarendon Press,Ox- Ecology . 1995. The breedingsystems of six species of Arabis. ford,United Kingdom. Amer Bot. 82: 869-877. Brooks,D. R., and D. A. McLennan. 1993. Parascript.Smith- J. , and P. 1993. The for rust sonian Series in Comparative EvolutionaryBiology. Bierzychudek. potential infectionto cause naturalselection in Arabis SmithsonianInstitution Press, Washington. 429 pp. apomictic holboellii 95: 533-541. Caten, C. E. 1987. The concept of race in plant pathology. (Brassicaceae). Oecologia and R. 1997. versusvisual cues Pp. 21-37. In: Populationsof plant pathogens:their dy- , Raguso. Olfactory in a floral 109: 414-426. namicsand genetics.Eds., M. S. Wolfeand C. E. Caten. mimicrysystem. Oecologia D. B. 0. 1971. Coevolution of the rust and BlackwellScientific Publications, Oxford, United King- Savile, fungi dom. theirhosts. Quart. Rev. Biol. 46: 211-218. Farr,D. F., G. F. Bills, G. P. Chamuris,and A. Y Rossman. 1974a. Puccinia holboellii.Fungi Canadensis47: 1-2. 1989. Fungi on plants and plant productsin theUnited 1974b. Some new or poorlyknown rusts of Brassi- States.American PhytopathologicalSociety Press, St. caceae. Can. J. Bot. 52: 1501-1507. Paul, Minnesota. 1252 pp. Sivarajan,V. V., and N. K. B. Robson. 1991. Introductionto the Felsenstein,J. 1985. Confidence limitson phylogenies:an principlesof plant taxonomy.Cambridge University approach using the bootstrap.Evolution 39: 783-791. Press,Cambridge, United Kingdom.292 pp. Funk, D. J., D. J. Futuyma,G. Orti and A. Meyer.1995. A Soltis,D. E., P. E. Soltis,and B. G. Milligan.1992. Intraspe- historyof host association and evolutionarydiversifi- cificchloroplast DNA variation:systematic and phylo- cation for Opharella(Coleoptera: Chrysomelidae):new geneticimplications. Pp. 117-150. In: Molecularsystem- evidencefrom mitochondrial DNA. Evolution49:1008- atics of plants. Eds., P. S. Soltis,D. E. Soltis,and J. J. 1017. Doyle. Chapman and Hall, New York. Futuyma,D. J., M. C. Keese, and D. J. Funk. 1995. Genetic Swofford,D. L. 1993. PAUP: Phylogeneticanalysis using par- constraintson macroevolution:the evolution of host simony. affiliationin the leaf beetle genus Ophraella.Evolution Thompson,J. N. 1986. Patternsin coevolution.Pp. 119-143. 49: 797-809. In: Coevolutionand systematics.Eds., A. R. Stone and

This content downloaded by the authorized user from 192.168.52.70 on Mon, 10 Dec 2012 12:55:26 PM All use subject to JSTOR Terms and Conditions RoY ET AL.: CRYPTIC SPECIES OF PUCCINIA 853

D. L. Hawksworth.Clarendon Press, Oxford, United genes forphylogenetics. Pp. 315-320. In: PCR protocols: Kingdom. a guide to methodsand applications.Eds., M. Innis, D. Vogler,D. R., and T. D. Bruns. 1998. Phylogeneticrelation- Gelfand,J. Sninsky,and T. White.Academic Press,New ships among the pine stemrust fungi (Cronartium and York. Peridermiumspp.). Mycologia90: 244-257. Zambino, P.J., and L. J. Szabo. 1993. Phylogeneticrelation- White,T. J., T. Bruns, S. Lee, and J. Taylor.1990. Amplifi- ships of selected cereal and grassrusts based on rDNA cation and directsequencing of fungalribosomal RNA sequence analysis.Mycologia 85: 401-414.

This content downloaded by the authorized user from 192.168.52.70 on Mon, 10 Dec 2012 12:55:26 PM All use subject to JSTOR Terms and Conditions