Verkley et al.therefore be proposed for it. Sphaerul<strong>in</strong>a rubi Demaree & Wilcoxis already <strong>in</strong> use for another fungus with a Cyl<strong>in</strong>drosporium sexualstate (C. rubi Ellis & Morgan, conidia 40–55 × 2.5 µm cf. Saccardo),so Sphaer. westendorpii is proposed here as nomen novum.Sphaerul<strong>in</strong>a rehmiana has been associated with Sep<strong>to</strong>ria rosae<strong>CBS</strong> 355.58, which has been identified as S. rosae, is geneticallydist<strong>in</strong>ct from Sphaer. westendorpii (Quaedvlieg et al. 2013).Insufficiently known <strong>species</strong>For the follow<strong>in</strong>g <strong>species</strong> no host material was available and thesehave only been studied <strong>in</strong> culture, mostly based on older isolates,for which details are not described when the stra<strong>in</strong> is regarded asdegenerate.Sep<strong>to</strong>ria hippocastani Berk. & Broome, Ann. Mag. nat.Hist., Ser. 2, 5: 379. 1850.Material exam<strong>in</strong>ed: Germany, Pfälzer Wald, on Aesculus hippocastanum, Sep1961, deposited Nov 1961, W. Gerlach, liv<strong>in</strong>g culture <strong>CBS</strong> 411.61 (= BBA 9619).Note: <strong>CBS</strong> 411.61 is degenerated and sterile, but based onmultilocus sequence analysis it can be concluded that it is aSep<strong>to</strong>ria s. str. (Fig. 2).Sep<strong>to</strong>ria limonum Pass., Atti Soc. crit<strong>to</strong>g. ital., 2: 23. 1879.Description <strong>in</strong> vitro (18 ºC, near UV): Colonies on OA 20–29 mmdiam <strong>in</strong> 3 wk, with an even, colourless marg<strong>in</strong>; colonies plane,spread<strong>in</strong>g, immersed mycelium <strong>in</strong> the centre flesh, surrounded by abroad zone of dark v<strong>in</strong>aceous <strong>to</strong> brown-v<strong>in</strong>aceous, aerial myceliumabsent, or scarce, with few tufts of pure white aerial hyphae; reverseconcolorous. No sporulation observed. Colonies on MEA 25–32 mmdiam <strong>in</strong> 3 wk, with an even <strong>to</strong> somewhat ruffled, buff <strong>to</strong> colourlessmarg<strong>in</strong>; colonies spread<strong>in</strong>g, somewhat elevated <strong>in</strong> the centre,immersed mycelium appear<strong>in</strong>g grayish, the colony surface almostentirely covered by a dense mat of white <strong>to</strong> grey, woolly-floccoseaerial mycelium; reverse <strong>in</strong> the centre rust, surrounded by a broadzone of olivaceous-grey <strong>to</strong> greenish grey, which is sharply borderedby the narrow buff <strong>to</strong> luteous marg<strong>in</strong>. No sporulation observed.Material exam<strong>in</strong>ed: Italy, Citrus limonium, isolated Mar. 1951, deposited by G.Goidanich, liv<strong>in</strong>g culture <strong>CBS</strong> 419.51.Notes: In the multilocus sequence analysis (Fig. 2) this stra<strong>in</strong>groups with <strong>CBS</strong> 356.36 (S. citricola) and few other stra<strong>in</strong>s <strong>in</strong> aweakly supported clade close <strong>to</strong> the plurivorous Sep<strong>to</strong>ria protearumand isolates of Sep<strong>to</strong>ria citri. Due <strong>to</strong> the lack of morphological<strong>in</strong>formation l<strong>in</strong>ked <strong>to</strong> this stra<strong>in</strong>, its identity rema<strong>in</strong>s uncerta<strong>in</strong>.DISCUSSIONThe type <strong>species</strong> of the genus Sep<strong>to</strong>ria, S. cytisi, could not be<strong>in</strong>cluded <strong>in</strong> the multilocus analysis due <strong>to</strong> the fact that only LSU andITS sequences were available for this <strong>species</strong>. However, as shownby Quaedvlieg et al. (2011), the position of this taxon is beyonddoubt central <strong>to</strong> the clade <strong>in</strong>dicated here as the ma<strong>in</strong> Sep<strong>to</strong>riaclade. Several “typical” Sep<strong>to</strong>ria <strong>species</strong> <strong>in</strong>fect<strong>in</strong>g herbaceousplants proved genetically distant from S. cytisi and its relatives, andcan best be classified <strong>in</strong> separate genera, Sphaerul<strong>in</strong>a (Quaedvlieget al. 2013) and Caryophyllosep<strong>to</strong>ria.The identification of Sep<strong>to</strong>ria has thus far ma<strong>in</strong>ly relied on hosttaxonomy and morphological characters of the shape, size, andseptation of conidia (Jørstad 1965, Teterevnikova-Babayan 1987,Andrianova 1987, Vanev et al. 1997, Muthumary 1999, Sh<strong>in</strong> &Sameva 2004, Priest 2006). Taxonomists have noted that conidialwidth is generally a more reliable character for <strong>species</strong> identificationthan conidial length, which is more variable. Some also noticed thatSep<strong>to</strong>ria material collected from the same location and host <strong>species</strong>,but under different environmental conditions or at different times <strong>in</strong>the same season, can differ considerably <strong>in</strong> average conidial sizes,particularly length (Jørstad 1965). These f<strong>in</strong>d<strong>in</strong>gs are also confirmed<strong>in</strong> our study. Reliable identification based on morphologicalcomparison alone is not possible for many Sep<strong>to</strong>ria <strong>species</strong>, andreference sequences will have <strong>to</strong> be produced for many more taxa<strong>in</strong> future. This will require critical studies of type specimens andalso require the recollection of fresh material. It is crucial that thetypes of the oldest names available for Sep<strong>to</strong>ria on certa<strong>in</strong> hostswill need <strong>to</strong> be studied as part of such work, and where necessaryepitypes designated <strong>to</strong> fix the genetic application of these names.Although hardly practised thus far by taxonomists, isolation and study<strong>in</strong> culture is a valuable and <strong>in</strong>dispensable <strong>to</strong>ol for Sep<strong>to</strong>ria <strong>species</strong><strong>delimitation</strong> and identification. We noted that the shape of conidiaon OA generally agree best with those <strong>in</strong> the source material onthe natural substrate. Under standardised <strong>in</strong>cubation conditions onstandard media cultures orig<strong>in</strong>at<strong>in</strong>g from deviant voucher material,for example because it developed under adverse conditions, showaga<strong>in</strong> their “normal” phenotypes which is better for comparisonpurposes. Extract<strong>in</strong>g DNA from axenic cultures is straight-forwardand less prone <strong>to</strong> errors caused by contam<strong>in</strong>ants, a problem oftenencountered when extract<strong>in</strong>g DNA from plant tissue.The K2P results show that the five prote<strong>in</strong> cod<strong>in</strong>g genes useddur<strong>in</strong>g this research should all theoretically be able <strong>to</strong> dist<strong>in</strong>quishevery <strong>species</strong> <strong>in</strong> this dataset as their average <strong>in</strong>ter- <strong>to</strong> <strong>in</strong>traspecificdistance ration is over 10:1. The problem is that these are averagenumbers, not absolute numbers. For example, the Btub K2P graph<strong>in</strong> Fig. 1 starts at 0 and not at at 0.29, mean<strong>in</strong>g that there actuallyare a few <strong>species</strong> <strong>in</strong> our dataset that are not dist<strong>in</strong>guishable by Btubalone (although obviously by far most <strong>species</strong> <strong>in</strong> fact are). To avoidthis, we recommend us<strong>in</strong>g at least two of the prote<strong>in</strong> cod<strong>in</strong>g lociused <strong>in</strong> this study for identification of Sep<strong>to</strong>ria and allied genera.Because EF and Btub both have very high PCR success ratesand have the highest <strong>species</strong> resolution percentage of all the lociused <strong>in</strong> this study, we recommend us<strong>in</strong>g these two loci for <strong>species</strong>identification purposes. It is advisable, however, <strong>to</strong> first sequencethe ITS and LSU for a prelim<strong>in</strong>ary genus identification by blast<strong>in</strong>g <strong>in</strong>GenBank and other useful databases.The multilocus sequence dataset generally provided goodresolution, with maximum <strong>to</strong> high bootstrap support for almost allterm<strong>in</strong>al and most of the deeper nodes of the phylogenetic tree. The<strong>in</strong>traspecific variation <strong>in</strong> the genes <strong>in</strong>vestigated is limited for mosttaxa, even if specimens orig<strong>in</strong>ate from such distant geographicorig<strong>in</strong>s as New Zealand, Korea and Europe (S. convolvuli, S.leucanthemi, S. polygonorum). Stra<strong>in</strong>s assigned <strong>to</strong> Sep<strong>to</strong>ria citripossibly represent a <strong>species</strong> complex, one of few groups with<strong>in</strong>the ma<strong>in</strong> Sep<strong>to</strong>ria clade that was not resolved. One case of crypticspeciation is revealed <strong>in</strong> the S. chrysanthemella complex, whereat least two genetically discrete entities can be found that arephenotypically difficult <strong>to</strong> dist<strong>in</strong>guish.Our results confirm that most <strong>species</strong> of Sep<strong>to</strong>ria have narrowhost ranges, be<strong>in</strong>g limited <strong>to</strong> a s<strong>in</strong>gle genus or a few genera of thesame plant family. There were a few notable exceptions, however.We demonstrated that the supposed s<strong>in</strong>gle-family host ranges302
A <strong>new</strong> <strong>approach</strong> <strong>to</strong> <strong>species</strong> <strong>delimitation</strong> <strong>in</strong> Sep<strong>to</strong>riaof Sep<strong>to</strong>ria paridis (Liliaceae) and S. urticae (Urticaceae), eachactually <strong>in</strong>cluded one additional family (Violaceae and Lamiaceae,respectively). More surpris<strong>in</strong>gly Sep<strong>to</strong>ria protearum, previouslyonly associated with Proteaceae (Protea) (Crous et al. 2004), wasnow found <strong>to</strong> be also associated with Araceae (Zanthedeschia),Aspleniaceae (Asplenium), Rutaceae (Boronia), Borag<strong>in</strong>aceae(Myosotis), Oleandraceae (Nephrolepis), and Rosaceae (Geum).To our knowledge this is the first study <strong>to</strong> provide DNA-basedevidence confirm<strong>in</strong>g that multiple family-associations occur for as<strong>in</strong>gle <strong>species</strong> <strong>in</strong> Sep<strong>to</strong>ria. It is <strong>to</strong> be expected that collect<strong>in</strong>g andsequenc<strong>in</strong>g of more material will show more taxa <strong>to</strong> be plurivorous,and perhaps S. paridis and S. urticae will be among those.Coevolution of plant pathogenic fungi and their hosts hasbeen documented for several groups. Other possible patterns ofevolution have already been suggested for sep<strong>to</strong>ria-like fungi <strong>in</strong>previous studies but the data available were not sufficient <strong>to</strong> fullyunderstand the evolution of these fungi (Feau et al. 2006). Therobust phylogeny we <strong>in</strong>ferred revealed polyphyletic distributionpatterns over the entire range of the Sep<strong>to</strong>ria clade for no less than10 (s<strong>in</strong>gle<strong>to</strong>ns excluded) of the host families represented. Theseresults clearly reject the coevolution hypothesis for Sep<strong>to</strong>ria, as<strong>species</strong> do not seem <strong>to</strong> consistently coevolve with hosts from as<strong>in</strong>gle host family but frequently jump successfully <strong>to</strong> hosts <strong>in</strong> <strong>new</strong>families. Caryophyllosep<strong>to</strong>ria seems an exceptional genus <strong>in</strong> that i<strong>to</strong>nly comprises <strong>species</strong> <strong>in</strong>fect<strong>in</strong>g Caryophyllaceae, but it should benoted that it now only conta<strong>in</strong>s four <strong>species</strong>, as three other <strong>species</strong><strong>in</strong>fect<strong>in</strong>g this family cluster distant with<strong>in</strong> the Sep<strong>to</strong>ria clade (S.cucubali, S. cerastii, and S. stellariae). In the other clades somes<strong>in</strong>gle-host family clusters can be found, but they do not comprisemore than six fungal <strong>species</strong> (S. chrysanthemella and closerelatives of Asteraceae with<strong>in</strong> subclade 4b).We conclude that trans-family host jump<strong>in</strong>g must be a majorforce driv<strong>in</strong>g the evolution of Sep<strong>to</strong>ria and Sphaerul<strong>in</strong>a. Specieslike S. paridis and S. urticae <strong>in</strong>fect<strong>in</strong>g (at least) two plant familiesmay <strong>in</strong> fact be cases <strong>in</strong> po<strong>in</strong>t, as they could be <strong>in</strong> a transitionalperiod of gradually chang<strong>in</strong>g from one pr<strong>in</strong>cipal host family <strong>to</strong>another, unrelated one. The genetic basis for successful hostjump<strong>in</strong>g is unclear. It may <strong>in</strong>volve horizontal gene transfer, transientphases of endophytic <strong>in</strong>fections <strong>in</strong> “non-hosts” as a first step <strong>in</strong> aprocess of genetic adaptation <strong>to</strong> <strong>new</strong> optimal hosts, or perhaps acomb<strong>in</strong>ation of both. Plant pathological research may shed morelight on the mechanisms driv<strong>in</strong>g Sep<strong>to</strong>ria evolution which would beimportant, as it may <strong>in</strong> future allow accurate assessment of risks<strong>in</strong>volved with the <strong>in</strong>troduction of <strong>new</strong> crops <strong>in</strong> areas where Sep<strong>to</strong>ria<strong>species</strong> occur on the local flora.Host family <strong>in</strong>dexThe taxa fully described <strong>in</strong> the Taxonomy section of this study arelisted below accord<strong>in</strong>g <strong>to</strong> the host family.AceraceaeSphaerul<strong>in</strong>a acerisApiaceaeSep<strong>to</strong>ria aegopodiiS. aegopod<strong>in</strong>aS. anthrisciS. apiicolaS. heracleiS. petrosel<strong>in</strong>iS. siiAraceaeSep<strong>to</strong>ria protearumAspleniaceaeSep<strong>to</strong>ria protearumAsteraceaeSep<strong>to</strong>ria chromolaenaeS. chrysanthemellaS. ekmannianaS. erigerontisS. hypochoeridisS. lactucaeS. leucanthemiS. matricariaeS. putridaS. senecionisSphaerul<strong>in</strong>a sociaBetulaceaeSphaerul<strong>in</strong>a betulaeBorag<strong>in</strong>aceaeSep<strong>to</strong>ria protearumCampanulaceaeSep<strong>to</strong>ria campanulaeS. citri complexCaryophyllaceaeCaryophyllosep<strong>to</strong>ria lychnidisC. silenesC. spergulaeSep<strong>to</strong>ria cerastiiS. cucubaliS. stellariaeConvolvulaceaeSep<strong>to</strong>ria convolvuliCornaceaeSphaerul<strong>in</strong>a cornicolaCucurbitaceaeSep<strong>to</strong>ria cucurbitacearumDipsacaceaeSep<strong>to</strong>ria scabiosicolaFabaceaeSep<strong>to</strong>ria astragaliHypericaceaeSep<strong>to</strong>ria hypericiIridaceaeSep<strong>to</strong>ria sisyr<strong>in</strong>chiiLamiaceaeSep<strong>to</strong>ria galeopsidisS. lamiicolaS. melissaeS. stachydisLiliaceaeSep<strong>to</strong>ria paridisOleandraceaeSep<strong>to</strong>ria protearumOnagraceaeSep<strong>to</strong>ria epilobiiPassifloraceaeSep<strong>to</strong>ria passifloricolaPolemoniaceaeSep<strong>to</strong>ria phlogisPolygonaceaeSep<strong>to</strong>ria polygonorumwww.studies<strong>in</strong>mycology.org303