Weir et al.grass-inhabiting <strong>species</strong> of the C. graminicola group and thedevelopment of a m<strong>or</strong>e useful taxonomy f<strong>or</strong> this group of fungi(e.g. Hsiang & Goodwin 2001, Du et al. 2005, and Crouch etal. 2006). This group is now recognised as comprising severalhost-specialised, genetically well characterised <strong>species</strong>, but amodern taxonomy f<strong>or</strong> C. gloeosp<strong>or</strong>ioides has yet to be resolved.Von Arx (1970) and Sutton (1980) distinguished the C.gloeosp<strong>or</strong>ioides group using conidial shape and size. A few apparentlyhost-specialised, C. gloeosp<strong>or</strong>ioides-like taxa were retained by theseauth<strong>or</strong>s, but the basis of their identification was often difficult tounderstand. Pri<strong>or</strong> to the availability of DNA sequence data, taxonomicconcepts within <strong>Colletotrichum</strong> were based on features such as host<strong>species</strong>, substrate, conidial size and shape, shape of appress<strong>or</strong>ia,growth rate in culture, colour of cultures, presence <strong>or</strong> absence ofsetae, whether <strong>or</strong> not the teleom<strong>or</strong>ph develops, etc. Some studieshave found characters such as these useful f<strong>or</strong> distinguishinggroups within C. gloeosp<strong>or</strong>ioides (e.g. Higgins 1926, G<strong>or</strong>ter 1956,Hind<strong>or</strong>f 1973, and Johnston & Jones 1997). However, problemsarise because many of these m<strong>or</strong>phological features change underdifferent conditions of growth (dependent upon growth media,temperature, light regime, etc.), <strong>or</strong> can be lost <strong>or</strong> change with repeatedsubculturing. Host preference is po<strong>or</strong>ly controlled — even good, welldefinedpathogens causing a specific disease can be isolated bychance from other substrates (e.g. Johnston 2000). <strong>Colletotrichum</strong>conidia will germinate on most surfaces, f<strong>or</strong>m an appress<strong>or</strong>ium,remain attached to that surface as a viable propagule <strong>or</strong> perhaps asa min<strong>or</strong>, endophytic <strong>or</strong> latent infection, and grow out from there intosenescing plant tissue <strong>or</strong> onto agar plates if given the opp<strong>or</strong>tunity.In addition, the same disease can be caused by genetically distinctsets of isolates, the shared pathogenicity presumably independentlyevolved, e.g. the bitter rot disease of apple is caused by membersof both the C. acutatum and C. gloeosp<strong>or</strong>ioides <strong>species</strong> <strong>complex</strong>es(Johnston et al. 2005).Sutton (1992) commented on C. gloeosp<strong>or</strong>ioides that “Noprogress in the systematics and identification of isolates belongingto this <strong>complex</strong> is likely to be made based on m<strong>or</strong>phology alone”.A start was made towards a modern understanding of this namewith the designation of an epitype specimen with a culture derivedfrom it to stabilise the application of the name (Cannon et al. 2008).Based on ITS sequences, the ex-epitype isolate belongs in astrongly supp<strong>or</strong>ted clade, distinct from other taxa that have beenconfused with C. gloeosp<strong>or</strong>ioides in the past, such as C. acutatumand C. boninense (e.g. Abang et al. 2002, Martinez-Culebras etal. 2003, Johnston et al. 2005, Chung et al. 2006, Farr et al. 2006,Than et al. 2008). However, biological and genetic relationshipswithin the broad C. gloeosp<strong>or</strong>ioides clade remain confused and ITSsequences alone are insufficient to resolve them.In this study we define the limits of the C. gloeosp<strong>or</strong>ioides<strong>species</strong> <strong>complex</strong> on the basis of ITS sequences, the <strong>species</strong> weaccept within the <strong>complex</strong> f<strong>or</strong>ming a strongly supp<strong>or</strong>ted cladein the ITS gene tree (fig. 1 in Cannon et al. 2012, this issue). Inall cases the taxa we include in the C. gloeosp<strong>or</strong>ioides <strong>complex</strong>would fit within the traditional m<strong>or</strong>phological concept of the C.gloeosp<strong>or</strong>ioides group (e.g. von Arx 1970, M<strong>or</strong>due 1971, andSutton 1980). Commonly used <strong>species</strong> names within the C.gloeosp<strong>or</strong>ioides <strong>complex</strong> include C. fragariae, C. musae, and C.kahawae. Since the epitype paper (Cannon et al. 2008), severalnew C. gloeosp<strong>or</strong>ioides-like <strong>species</strong> have been described inregional studies, where multi-gene analyses have shown the new<strong>species</strong> to be phylogenetically distinct from the ex-epitype strain ofC. gloeosp<strong>or</strong>ioides (e.g. Rojas et al. 2010, Phoulivong et al. 2011,and Wikee et al. 2011).The regional nature of most of these studies, the often restrictedgenetic sampling across the diversity of C. gloeosp<strong>or</strong>ioides globally,and the minimal overlap between isolates treated and gene regionstargeted in the various studies, means that the relationship betweenthe newly described <strong>species</strong> is often po<strong>or</strong>ly understood.While some auth<strong>or</strong>s have embraced a genetically highlyrestricted concept f<strong>or</strong> C. gloeosp<strong>or</strong>ioides, many applied researcherscontinue to use the name in a broad, group-<strong>species</strong> concept (e.g.Bogo et al. 2012, Deng et al. 2012, Kenny et al. 2012, Parvin etal. 2012, and Zhang et al. 2012). In this paper we accept bothconcepts as useful and valid. When used in a broad sense, werefer to the taxon as the C. gloeosp<strong>or</strong>ioides <strong>species</strong> <strong>complex</strong> <strong>or</strong> C.gloeosp<strong>or</strong>ioides s. lat.This paper aims to clarify the genetic and taxonomicrelationships within the C. gloeosp<strong>or</strong>ioides <strong>species</strong> <strong>complex</strong> usinga set of isolates that widely samples its genetic, biological andgeographic diversity. Type specimens, <strong>or</strong> cultures derived fromtype specimens, have been examined wherever possible. Althoughwe do not treat all of the names placed in synonymy with C.gloeosp<strong>or</strong>ioides <strong>or</strong> Glomerella cingulata by von Arx & Müller (1954)and von Arx (1957, 1970), we treat all names f<strong>or</strong> which a possibleclose relationship with C. gloeosp<strong>or</strong>ioides has been suggested inthe recent literature, along with all subspecific taxa and f<strong>or</strong>maespeciales within C. gloeosp<strong>or</strong>ioides and G. cingulata.ITS sequences, the official barcoding gene f<strong>or</strong> fungi (Seifert2009, Schoch et al. 2012), do not reliably resolve relationshipswithin the C. gloeosp<strong>or</strong>ioides <strong>complex</strong>. We define <strong>species</strong> in the<strong>complex</strong> genetically rather than m<strong>or</strong>phologically, on the basis ofphylogenetic analyses of up to eight genes. Following Cannonet al. (2012, this issue) the generic name <strong>Colletotrichum</strong> is usedas the preferred generic name f<strong>or</strong> all <strong>species</strong> wherever possiblethroughout this paper, whether <strong>or</strong> not a Glomerella state has beenobserved f<strong>or</strong> that fungus, and whether <strong>or</strong> not the Glomerella statehas a f<strong>or</strong>mal name.MATERIALS AND METHODSSpecimen isolation and selectionAn attempt was made to sample the genetic diversity across C.gloeosp<strong>or</strong>ioides as widely as possible, with isolates from diversehosts from around the w<strong>or</strong>ld selected f<strong>or</strong> m<strong>or</strong>e intensive study. ABLAST search of GenBank using the ITS sequence of the epitypeculture of C. gloeosp<strong>or</strong>ioides (Cannon et al. 2008) provided acoarse estimate f<strong>or</strong> the genetic limit of the C. gloeosp<strong>or</strong>ioides<strong>complex</strong> and ITS diversity across the <strong>complex</strong> was used toselect a genetically diverse set of isolates. Voucher cultures wereobtained from the research groups who deposited the GenBankrec<strong>or</strong>ds. To these were added isolates representing the knowngenetic and m<strong>or</strong>phological diversity of C. gloeosp<strong>or</strong>ioides fromNew Zealand, isolated from rots of native and introduced fruits,from diseased exotic weeds, and as endophytes from leaves ofnative podocarps. Additional isolates representing ex-type andauthentic cultures of as many named taxa and f<strong>or</strong>mae specialeswithin the C. gloeosp<strong>or</strong>ioides <strong>complex</strong> as possible were obtainedfrom international culture collections. Approximately 400 isolatesbelonging to the C. gloeosp<strong>or</strong>ioides <strong>complex</strong> were obtained.GAPDH gene sequences were generated f<strong>or</strong> all isolates as an initialmeasure of genetic diversity. A subset of 156 isolates, selected t<strong>or</strong>epresent the range of genetic, geographic, and host plant diversity,116
The <strong>Colletotrichum</strong> gloeosp<strong>or</strong>ioides <strong>species</strong> <strong>complex</strong>was used in this research (Table 1).Most of the New Zealand isolates had been st<strong>or</strong>ed as conidialsuspensions made from single conidium <strong>or</strong> ascosp<strong>or</strong>e cultures andthen st<strong>or</strong>ed at -80 °C in a 5 % glycerol/water suspension. Additionalisolates from New Zealand were obtained from the ICMP culturecollection, where isolates are st<strong>or</strong>ed as lyophilised (freeze-dried)ampoules <strong>or</strong> in a metabolically inactive state in liquid nitrogenat -196 °C. The st<strong>or</strong>age hist<strong>or</strong>y of most of the isolates receivedfrom other research groups is not known. Table 1 lists the isolatesstudied. All those supplying cultures are acknowledged at theend of this manuscript, and additional details on each culture areavailable on the ICMP website (http://www.landcareresearch.co.nz/resources/collections/icmp).Culture collection and fungal herbarium (fungarium)abbreviations used herein are: <strong>CBS</strong> = Centraalbureau vo<strong>or</strong>Schimmelcultures (Netherlands), ICMP = International Collectionof Micro<strong>or</strong>ganisms from Plants, MFLU = Mae Fah Luang UniversityHerbarium (Thailand) MFLUCC = Mae Fah Luang UniversityCulture Collection (Thailand), GCREC = University of Fl<strong>or</strong>ida,Gulf Coast Research and Education Centre (USA), HKUCC = TheUniversity of Hong Kong Culture Collection (China), IMI = CABIGenetic Resource Collection (UK), MAFF = Ministry of Agriculture,F<strong>or</strong>estry and Fisheries (Japan), DAR = Plant Pathology Herbarium(Australia), NBRC = Biological Resource Center, National Instituteof Technology and Evaluation (Japan), BCC = BIOTEC CultureCollection (Thailand), GZAAS = Guizhou Academy of AgriculturalSciences herbarium (China), MUCL = Belgian Co-<strong>or</strong>dinatedCollections of Micro-<strong>or</strong>ganisms, (agro)industrial fungi & yeasts(Belgium), BRIP = Queensland Plant Pathology Herbarium(Australia), PDD = New Zealand Fungal and Plant DiseaseCollection (New Zealand), BPI = U.S. National Fungus Collections(USA), STE-U = Culture collection of the Department of PlantPathology, University of Stellenbosch (South Africa), and MCA =M. Catherine Aime’s collection series, Louisiana State University(USA).DNA extraction, amplification, and sequencingMycelium was collected from isolates grown on PDA agar, andmanually comminuted with a micropestle in 420 μL of QuiagenDXT tissue digest buffer; 4.2 μL of proteinase K was added andincubated at 55 °C f<strong>or</strong> 1 h. After a brief centrifugation 220 μL ofthe supernatant was placed in a C<strong>or</strong>bett X-tract<strong>or</strong>Gene automatednucleic acid extraction robot. The resulting 100 μL of pure DNA inTE buffer was st<strong>or</strong>ed at -30 °C in 1.5 mL tubes until use.Gene sequences were obtained from eight nuclear gene regions,actin (ACT) [316 bp], calmodulin (CAL) [756 bp], chitin synthase(CHS-1) [229 bp], glyceraldehyde-3-phosphate dehydrogenase(GAPDH) [308 bp], the ribosomal internal transcribed spacer(ITS) [615 bp], glutamine synthetase (GS) [907 bp], manganesesuperoxidedismutase (SOD2) [376 bp], and β-tubulin 2 (TUB2)[716 bp].PCR Primers used during this study are shown in Table 2.The standard CAL primers (O’Donnell et al. 2000) gave po<strong>or</strong> <strong>or</strong>non-specific amplification f<strong>or</strong> most isolates, thus new primers(CL1C, CL2C) were designed f<strong>or</strong> <strong>Colletotrichum</strong> based on the C.graminicola M1.001 genome sequence. The standard GS primers(Stephenson et al. 1997) sequenced po<strong>or</strong>ly f<strong>or</strong> some isolates dueto an approx. 9 bp homopolymer T run 71 bp in from the end of theGSF1 primer binding site. A new primer, GSF3, was designed 41 bpdownstream of this region to eliminate the homopolymer slippageerr<strong>or</strong> from sequencing. The reverse primer GSR2 was designed inthe same location as GSR1 with one nucleotide change. Both newGS primers were based on similarity with a C. theobromicola UQ62sequence (GenBank L78067, as C. gloeosp<strong>or</strong>ioides).The PCRs were perf<strong>or</strong>med in an Applied Biosystems VeritiThermal Cycler in a total volume of 25 μL. The PCR mixturescontained 15.8 μL of UV-sterilised ultra-filtered water, 2.5 μL of 10×PCR buffer (with 20 mM MgCl 2), 2.5 μL of dNTPs (each 20 μM), 1μL of each primer (10 μM), 1 μL of BSA, 1 μL of genomic DNA, and0.2 μL (1 U) of Roche FastStart Taq DNA Polymerase.The PCR conditions f<strong>or</strong> ITS were 4 min at 95 °C, then 35cycles of 95 °C f<strong>or</strong> 30 s, 52 °C f<strong>or</strong> 30 s, 72 °C f<strong>or</strong> 45 s, and then7 min at 72 °C. The annealing temperatures differed f<strong>or</strong> the othergenes, with the optimum f<strong>or</strong> each; ACT: 58 °C, CAL: 59 °C, CHS-1: 58 °C, GAPDH: 60 °C, GS: 54 °C, SOD2: 54 °C, TUB2: 55°C. Some isolates required altered temperatures and occasionallygave multiple bands, which were excised separately from anelectroph<strong>or</strong>esis gel and purified. PCR Products were purified on aQiagen MinElute 96 UF PCR Purification Plate.DNA sequences were obtained in both directions on an AppliedBiosystems 3130xl Avant Genetic analyzer using BigDye v. 3.1chemistry, electropherograms were analysed and assembled inSequencher v. 4.10.1 (Gene Codes C<strong>or</strong>p.).Phylogenetic analysesMultiple sequence alignments of each gene were made withClustalX v. 2.1 (Larkin et al. 2007), and manually adjusted wherenecessary with Geneious Pro v. 5.5.6 (Drummond et al. 2011).Bayesian inference (BI) was used to reconstruct most ofthe phylogenies using MrBayes v. 3.2.1 (Ronquist et al. 2012).Bayesian inference has significant advantages over other methodsof analysis such as maximum likelihood and maximum parsimony(Archibald et al. 2003) and provides measures of clade supp<strong>or</strong>t asposteri<strong>or</strong> probabilities rather than random resampling bootstraps.jModelTest v. 0.1.1 (Posada 2008) was used to carry out statisticalselection of best-fit models of nucleotide substitution using thec<strong>or</strong>rected Akaike inf<strong>or</strong>mation criteria (AICc) (Table 3). Initialanalyses showed that individual genes were broadly congruent,thus nucleotide alignments of all genes were concatenated usingGeneious, and separate partitions created f<strong>or</strong> each gene withtheir own model of nucleotide substitution. Analyses on the fulldata set were run twice f<strong>or</strong> 5 x 10 7 generations, and twice f<strong>or</strong> 2 x10 7 generations f<strong>or</strong> the clade trees. Samples were taken from theposteri<strong>or</strong> every 1000 generations. Convergence of all parameterswas checked using the internal diagnostics of the standarddeviation of split frequencies and perf<strong>or</strong>mance scale reductionfact<strong>or</strong>s (PSRF), and then externally with Tracer v. 1.5 (Rambaut& Drummond 2007). On this basis the first 25 % of generationswere discarded as burn-in.An initial BI analysis treated all 158 isolates using a concatenatedalignment f<strong>or</strong> five of the genes, ACT, CAL, CHS-1, GAPDH, and ITS.<strong>Colletotrichum</strong> boninense and C. hippeastri were used as outgroups.A second BI analysis, restricted to ex-type <strong>or</strong> authentic isolatesof each of the accepted <strong>species</strong>, was based on a concatenatedalignment of all eight genes. A third set of BI analyses treatedfocussed on taxa within the Musae clade and the Kahawae clade.F<strong>or</strong> each clade, the ex-type <strong>or</strong> authentic isolates, together with 2–3additional selected isolates of each accepted taxon where available,were analysed using a concatenated alignment of all eight genes,with C. gloeosp<strong>or</strong>ioides used as the outgroup f<strong>or</strong> both analyses.www.studiesinmycology.<strong>or</strong>g117