International Journal of Food Microbiology 95 (2004) 341–349 www.elsevier.com/locate/ijfoodmicro

An integrated taxonomic study of langsethiae, Fusarium poae and Fusarium sporotrichioides based on the use of composite datasets

H. Schmidta, A. Adlerb, A. Holst-Jensenc, S.S. Klemsdald, A. Logriecoe, R.L. Machf, H.I. Nirenbergg, U. Thraneh, M. Torpc, R.F. Vogela, T. Yli-Mattilai, L. Niessena,*

a Technische Universitat-Mu¨nchen, Lehrstuhl fu¨r Technische Mikrobiologie, Weihenstephaner Steig 16, D-85350 Freising, Germany b Bundesamt fu¨r Agrarbiologie, Wieningerstrabe 8, A-4040 Linz, Austria c National Veterinary Institute, Section of Food and Feed Microbiology, P.O. Box 8156 Dep., N-0033 Oslo, Norway d The Norwegian Crop Research Institute, Plant Protection Centre, Høgskoleveien 7, N-1432 A˚ s, Norway e Institute of Science of Food Production, National Research Council, CNR, Viale Enaudi 51, I-70125 Bari, Italy f Institute for Chemical Engineering, Microbial Biochemistry and Gene Technology Group, Technical University of Vienna, Getreidemarkt 9/166, A-1060 Vienna, Austria g Institute of Plant Virology, Microbiology, and Biological Safety, Federal Biological Research Centre for Agriculture and Forestry, Ko¨nigin-Luise-Str. 19, D-14195 Berlin, Germany h Technical University of Denmark, BioCentrum-DTU, Søltofts Plads 221, DK-2800 Kgs. Lyngby, Denmark i Laboratory of Plant Physiology and Molecular Biology, Department of Biology, University of Turku, FIN-20014 Turku, Finland

Abstract

An integrated systematic study was carried out to clarify the taxonomical position and relationship of Fusarium langsethiae to other taxa within the Fusarium section Sporotrichiella. Strains of this species were compared with strains of the closely related species Fusarium poae and Fusarium sporotrichioides using a composite dataset. This set consisted of DNA sequences derived from the ribosomal internal transcribed spacer (ITS) regions, partial sequences of the ribosomal intergenic spacer (IGS) region, the h-tubulin and translation elongation factor-1 alpha (EF-1a) genes, AFLP fingerprints, chromatographic data on secondary metabolites and morphological data and growth characteristics. From these combined data, a consensus matrix was calculated by taking the mean of all pairwise distances between single isolates over all separate datasets. The consensus matrix was used as the basis for the construction of a UPGMA dendrogram and a multidimensional scaling, both of which revealed a clear separation of the three taxa. Partial IGS, EF-1a and h-tubulin sequence—as well as chromatography—and AFLP-derived similarities turned out to be comparably consistent, while ITS sequence- and morphology-derived similarity matrices were rather divergent. D 2004 Elsevier B.V. All rights reserved.

Keywords: Fusarium; Fungi; Polyphasic; ; Numerical; Composite dataset

1. Introduction * Corresponding author. Tel.: +49-8161-715496; fax: +49- 8161-713327. For the elucidation of taxonomic problems, E-mail address: [email protected] (L. Niessen). mycologists are still used to relying on the exami-

0168-1605/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.ijfoodmicro.2003.12.012 342 H. Schmidt et al. / International Journal of Food Microbiology 95 (2004) 341–349 nation of a more or less limited set of characters. On form a group separated from F. poae. Also, PCR- the one hand, morphological and other phenotypical based methods for the rapid detection and distinction observations are still essential for the valid descrip- of the species were developed (Konstantinova and tion of a fungal species, whereas on the other hand, Yli-Mattila, 2004; Mach et al., 2004; Niessen et al., molecular data such as DNA sequences and finger- 2004). Here we present a collaborative study aimed prints are about to dominate fungal systematics. at clarification of the taxonomic position of F. However, only by combining approaches can the langsethiae within the Fusarium section Sporotri- true relationship among different fungal groups be chiella. The objective of the present study was to fully elucidated, and multidisciplinary studies are the consider all available data, morphological, chromato- logical answer to meet this challenge. It has been graphic, and molecular (both fingerprints and nucle- discussed how mycologists could effectively com- otide sequences). By choosing this strategy, we hope bine these different data to produce reliable classifi- to contribute to the understanding of the true rela- cation and identification regimes (Seifert et al., tionships between the different species and put 2000). The prerequisites for this task are now met forward a clearer perception of the whole due to the rapid development in computer technolo- (Kendrick, 1979). gy and software engineering and the availability of diverse datasets. Because of their importance in agriculture and 2. Materials and methods human health, Fusarium species belong to the best studied group of fungal organisms. Their impact is 2.1. Fungal strains mainly based on the ability to cause a number of plant diseases and to synthesise a variety of myco- Information on the examined strains can be toxins which can contaminate food and feed. The retrieved from Torp and Adler (2004).Inthe trichothecenes are considered to be the most hazard- current study, only those strains for which a ous compounds produced by Fusarium species. Type complete set of data comprising all analytical A trichothecenes, especially T-2 toxin and HT-2 methods applied were selected. In particular, the toxin, are among the most toxic substances regularly number of strains included in the EF-1a dataset detected in cereal samples (Marasas et al., 1984).A (Knutsen et al., 2004) was considerably lower than recently discovered species, Fusarium langsethiae is the number of strains included in each of the other believed to be an important producer of these two datasets rendering the number of strains studied toxins in cereals. This species resembles Fusarium quite low. poae in several morphological features, but the latter species has only occasionally been shown to produce 2.2. Sequence data T-2 and HT-2 toxins. Thus, F. langsethiae has some affinity to Fusarium sporotrichioides and F. poae The DNA sequences of parts of the IGS, com- (Torp and Langseth, 1999; Torp and Nirenberg, plete sequences of the ITS region, and partial 2004). However, when toxin patterns are compared, sequences of the h-tubulin gene used are those the species seems to be more similar to F. sporo- described by Yli-Mattila et al. (2004). The partial trichioides than to F. poae, but segregated into a EF-1a sequences of the examined strains are those distinct cluster by statistical analysis of full spectra determined by Knutsen et al. (2004). An alignment chromatographic data (Thrane et al., 2004). Phylo- of the sequences was obtained by subjecting the genetic analyses of combined ribosomal internal sequences to BioNumerics software Version 2.50 transcribed spacer (ITS) regions, and partial sequen- (Applied Maths, Sint-Martens-Latems, Belgium). ces of the intergenic spacer (IGS) region and the h- For alignment and similarity calculation, no conver- tubulin (Yli-Mattila et al., 2003) and EF-1a genes sion costs were used and no gap penalty was (Knutsen et al., 2003), as well as cluster analysis of assigned. The similarity between the sequences AFLP fingerprints (Schmidt et al., 2004), strongly was calculated using the correction of Jukes and suggest that F. sporotrichioides and F. langsethiae Cantor (1969). H. Schmidt et al. / International Journal of Food Microbiology 95 (2004) 341–349 343

2.3. Amplified fragment length polymorphism From these combined data, a consensus matrix was calculated by taking the mean of all pair wise dis- AFLP data were obtained and processed as de- tances between single isolates over all separate data- scribed by Schmidt et al. (2004). Resemblance be- sets. No weights were assigned to the individual tween the fingerprints was calculated using the experiments. To visualise the obtained consensus Pearson correlation. matrix a dendrogram was calculated by the unweight- ed pair-group method using arithmetic means 2.4. Chromatographic data (UPGMA, Sokal and Michener, 1958). From the same matrix, a nonmetric multidimensional scaling (MDS, The chromatographic data on secondary metabo- Shepard, 1962) graph was computed. To check the lites were those obtained and processed by Thrane et robustness of the dendrogram and groups, the cophe- al. (2004) with the exception of data on beauvericin netic correlation and error flags for each node of the and enniatins which were not considered in the current dendrogram (standard deviation) were determined. study. The cubed correlation coefficient calculated by Pairwise resemblance between the data matrices was the COWTool software (Nielsen et al., 1998) was computed using the Pearson correlation. All calcula- directly imported into the BioNumerics software. tions were carried out on complete datasets using the BioNumerics software. 2.5. Other phenotypical data

Selected morphological data and growth character- 3. Results istics were recorded as described by Torp and Niren- berg (2004) into a binary character table (legal states 0 A set of strains was chosen for which data of each and 1). The characters used are throughout the text individual character set were available. From these referred to as phenotypical characters and described data a consensus matrix was calculated and trans- and summarized in the phenotypical character matrix formed into an UPGMA dendrogram (Fig. 1). This given in Table 1. dendrogram showed a clear separation of the strains into three clusters. The main groups of the dendro- 2.6. Composite dataset, congruence of characters and gram were well supported by a rather high cophe- software netic correlation. However, the standard deviations, which indicate the homogeneity of a group, were The consensus matrix was calculated using the quite high at the basal nodes. The error flags at the values from the similarities of the individual datasets. basal parental nodes did not overlap with those of the daughter nodes for any of the three main groups, Table 1 pointing to the conclusion that the three main clus- Taxonomic charactersa used to code phenotypical data and ters represent distinct entities (species). It should be characteristic I/O matrices of the taxa treated stressed that the F. langsethiae strain IBT 9959 Strains Characters formed a rather separate branch within the F. lang- 12345678 sethiae cluster. From the analysis of these data, it F. poae +++ÀÀ+++ was clear that F. sporotrichioides was the most F. langsethiae À + ÀÀ+ ÀÀÀ homogenous group of the three taxa. From the F. sporotrichioides + ÀÀ+ À ++À figure, it was not possible to conclude on the internal a Description of characters (+argument): (1) Sporodochial relationships between the three species. conidia (macroconidia) present under nUV. (2) Shape of conidia The result of comparative congruence analysis of of aerial mycelium (microconidia) globose/napiform only. (3) Only the individual datasets is presented in Fig. 2. The monophialides present. (4) Chlamydospores present. (5) Powdery comparison of the similarity matrices of two and two appearance on Czapec–Dox iprodione dichloran (CZID) agar. (6) Colony diameter on Potato sucrose agar (PSA), 25 jC exceeding 60 datasets shows that the highest concordance between mm after 6 days. (7) Height of aerial mycelium on PSA exceeding 3 two individual datasets was found between the simi- mm. (8) Fruity odour present in culture. larity matrices of the partial EF-1-a and h-tubulin 344 H. Schmidt et al. / International Journal of Food Microbiology 95 (2004) 341–349

Fig. 1. UPGMA dendrogram calculated from the combined similarity matrix of ITS, partial IGS, as well as partial DNA sequences of the EF-1-a and h-tubulin genes, chromatographic data, AFLPs and phenotypical data. Numbers at the nodes give the cophenetic correlation. Bars indicate the standard deviation for the corresponding cluster. The scale at the top indicates the percentage of similarity. gene sequences. The lowest concordance came from exclusion of the partial EF-1a gene sequences. Nota- the comparison of ITS and phenotypically derived bly, the resolution contributed from the EF-1a gene similarity matrices. It is interesting to note that the sequences was consequently also excluded. The new highest consilience of the composite matrix with an combined dataset included 30 strains of F. poae, F. individual matrix is with the one derived from the langsethiae and F. sporotrichioides.Theresultof AFLP experiments. reanalysing the data with the larger strain sample is As the set of strains for which a complete dataset shown in Fig. 3. Largely, the results are similar to was available turned out to be rather limited, in those of the first analysis, and the three main clusters particular due to the size of the EF-1a sample, we were well separated. However, the branching pattern reanalysed the data with a larger set of strains after observed at the basal nodes was not resolved because

Fig. 2. Congruence between the experiments leading to the clustering in Fig. 1. The similarity matrix derived from the pairwise comparison of the individual experiments (right side of the figure) was transformed to the UPGMA dendrogram at the left side. AFLP: data derived from the AFLP experiments; all: composite dataset; chrom: chromatographic data; IGS: DNA sequences of the IGS region; ITS: DNA sequences of the ITS regions; pheno: similarity data derived from the coding of phenotypical data (see Table 1); tef: partial DNA sequences of the EF-1a-gene; tub: partial DNA sequences of the h-tubulin gene. H. Schmidt et al. / International Journal of Food Microbiology 95 (2004) 341–349 345

Fig. 3. UPGMA dendrogram calculated from the combined similarity matrix of IGS, ITS as well as partial h-tubulin DNA sequences, chromatographic data, AFLPs and phenotypical data. Numbers at the nodes give the cophenetic correlation and the bars indicate the standard deviation for the corresponding cluster. The scale at the top indicates the percentage of similarity. 346 H. Schmidt et al. / International Journal of Food Microbiology 95 (2004) 341–349

Fig. 4. Two views of the multidimensional scaling (MDS) graph calculated from the similarity matrix leading to the dendrogram in Fig. 3. .: F. poae strains; 1: F. sporotrichioides strains; x: F. langsethiae strains. the standard deviation within the clusters was high various formats produced in the individual studies, and the error flags overlapped. e.g. molecular data as graphic format and as sequence Multidimensional scaling (MDS) is a non-hierar- strings, chromatographic data, phenotypical data as chical grouping technique, and was used to compute a binary matrix, were analysed. The algorithms used for three-dimensional graphic image of the relationships data computation measure the overall similarity be- between individual strains from the composite simi- tween individual strains and these matrices were larity matrix of the selected strains (i.e. excluding the transformed into graphs for visualization of the EF-1a data; Fig. 4). This ordination method allows results. The reason for choosing such an approach interpretation of the similarity of data leading to the was the difficulty of combining the different types of separation of a group as well as the homogeneity of data, since AFLP and chromatographic techniques do the group. However, conclusions about the phyloge- not produce distinct characters and character states netic relationship between groups are hard to draw and are therefore not suitable for phylogenetic analy- from the data presented. The three main groups sis, as are character-based techniques. corresponding to the three Fusarium species studied One of the problems encountered during the analy- were clearly separated. However, within the F. lang- sis of combined data was the limited number of sethiae group, strain IBT 9959 was again distant to all complete datasets. For only 14 of the 109 strains other F. langsethiae strains. studied were all data representing the different techni- ques available. The reason for this is that in most of the individual studies due to high complexity and costs, 4. Discussion only representative subsets of strains were analysed. However, a combination of individual similarity ma- The three taxa, F. poae, F. sporotrichioides and F. trices into a combined similarity matrix displayed clear langsethiae, were clearly separated by most of the separation of the taxa analysed with high resolution of various different individual techniques used within the the basal nodes in the resulting dendrogram (Fig. 1).A polyphasic approach presented here (Knutsen et al., similar clear separation of the taxa studied was found 2004; Schmidt et al., 2004; Thrane et al., 2004; Torp by Yli-Mattila et al. (2004) when sequence data for the and Nirenberg, 2004; Yli-Mattila et al., 2004). Use of h-tubulin gene, IGS, and ITS were combined in a a partial sequence of the tri5 gene of the strains neighbour-joining and POY analysis. However, none studied was the only method which failed to clearly of the individual sequences produced such a clear separate F. langsethiae as a species (Niessen et al., taxonomical separation. Especially the high intraspe- 2004). In the current comparative study, data of cific variability of the IGS sequences corroborated H. Schmidt et al. / International Journal of Food Microbiology 95 (2004) 341–349 347 taxonomic affiliation of the strains. The advantage of cases the grouping of operational taxonomic units (e.g. the IGS sequences was that the authors found a good single strains) coincide, and it has been pointed out that correlation between the sequence data and geographi- neither approach may truly reflect natural classification cal origin of the strains analysed. Konstantinova and (Sneath and Sokal, 1973). The differences in the Yli-Mattila (2004) used the sequence variation of the branching patterns shown in Figs. 1 and 3 can be IGS gene to separate two subgroups within F. langse- attributed to the different weights of the individual thiae by PCR. Also, sequences of the EF-1-a gene datasets. In Fig. 1, morphological and ITS sequence- clearly separated F. langsethiae as a separate taxon derived similarities contribute relatively less to the within the section Sporotrichiella of Fusarium (Knut- overall similarity compared to Fig. 3, because the sen et al., 2004). However, these authors state that there former includes EF-1a data. The EF-1a study of is a bias in classification of F. langsethiae as the closest Knutsen et al. (2004) is the only phylogenetic study sister taxon to F. sporotrichioides, which corresponds providing strong support for a sister taxon relationship to their mycotoxin profile (Thrane et al., 2004), and between F. sporotrichioides and F. langsethiae. The their morphological classification which indicate closer other phylogenetic studies indicate that one of the two relationship of F.langsethiae and F.poae. The subset of taxa is derived from the other (Yli-Mattila et al., 2004). strains analyzed for sequences of the EF-1-a gene was It was therefore not unexpected that exclusion of the particularly small. Therefore, a dendrogram which EF-1a data from the composite dataset had an impact contained all complete datasets without the EF-1-a on the degree of resolution observed. gene was calculated showing that the three taxa were In each composite dataset, the signal contributed still clearly separated (Fig. 3). However, the branching by morphology and ITS sequences seemed to be the pattern observed at the basal nodes was not so well least congruent, while the IGS, and partial EF-1a and resolved as compared to the dendrogram calculated h-tubulin gene sequence-derived similarities as well from complete datasets given in Fig. 1. as chromatographic- and AFLP-derived similarity The multidimensional scaling (MDS) tool of the matrices were largely congruent. The value of differ- BioNumerics software package was used to provide a ent DNA sequences for Fusarium taxonomy has been three-dimensional representation of the dendrogram discussed elsewhere (Knutsen et al., 2004; Yli-Mat- given in Fig. 3, where calculations were based on tila et al., 2004). Also, the usefulness of chemotyping datasets without the EF-1-a gene sequences (see Fig. for the identification and classification of fungal 4). The three taxa appear as distinct clouds in space species has been previously reported (O’Donnell et with the exception of strain IBT 9959, which is an al., 1998; Thrane et al., 2001). While ITS sequence outlier from F. langsethiae. The exceptional position of comparison has proven to be helpful in supporting that particular strain was also observed by Knutsen et teleomorph–anamorph connections and to articulate al. (2004), Thrane et al. (2004) and Yli-Mattila et al. phylogenetic relationships in certain groups of Tri- (2004) and can further be observed in the UPGMA choderma isolates, they were found to lack sufficient dendrograms given in the current study (Figs. 1 and 3). variation for general species delineation in Fusarium The strain was originally described as having interme- (Lieckfeldt and Seifert, 2000). An additional prob- diate morphology between F. poae and F. langsethiae lem, especially with ITS2 sequences, is the presence (Torp and Langseth, 1999). Also, Niessen et al. (2004), of two types of that particular sequence which may who used a combination of tri5 gene-based PCR assays occur as minor or major constituents in one isolate to differentiate between F. langsethiae and F. sporo- (Waalwijk et al., 1996; O’Donnell and Cigelnik, trichioides, placed this strain into the latter species. 1997). Also, this fact demands for very careful and The UPGMA algorithm is not suited for reconstruc- critical interpretation of ITS2 sequences obtained, tion of phylogeny unless all characters are ultrametric and even more when ITS2 fragments are used to and evolve at the same speed (Hillis et al., 1996).Such generate ITS-RFLP results. However, it has also been an assumption would not be valid for the data analysed demonstrated that ITS rDNA sequences may well be during the current study. A phenetic approach may be helpful to separate plant pathogenic forma speciales an indicator of phylogenetic relationship, but it is not in species of Fusarium, e.g. Fusarium solani (Suga et necessarily congruent with the latter. However, in most al., 2000, Lee et al., 2000). 348 H. Schmidt et al. / International Journal of Food Microbiology 95 (2004) 341–349

The difficulty in choosing and coding morpholog- points of view was published as a special issue of ical data to create numerical classification systems is Studies in Mycology (vol. 45, 2000). In their intro- immanent and requires much expertise. Analyses of the duction, the editors raise five central questions which similarity matrix of the combined data with that should duly be answered to cope with future develop- obtained from morphological inspection and growth ments in fungal taxonomy. One of these questions was characteristics show that the coding and grouping used how mycologists can effectively combine morpholog- is helpful for the morphological recognition of the three ical and molecular data to produce reliable classifica- species. Each species can be recognized on the basis of tion and identification regimes. Considering all papers a combination of distinct morphological characteristics presented, one of the conclusions drawn by the editors (see Table 1). However, it is difficult to apply such of that special issue was that multidisciplinary or characters when establishing evolutionary relation- polyphasic taxonomic studies are the way of the ships since very few of them are specific to lineages future. During the studies presented in the current revealed by molecular data, and vice versa. Since special issue, various working groups have elucidated morphological characters are also very plastic, and a group of fungal isolates using various up to date since few of them are covariant, it is tempting to molecular and classical morphological techniques to consider these characters less reliable than molecular characterize and delineate closely related species characters. within section Sporotrichiella of Fusarium.Inthe While sequences of protein coding genes, e.g. the current paper, attempts were made to use all the EF-1a and h-tubulin are generally better suited for different types of data supplied by various collabo- studies of higher level phylogeny, AFLP analysis is rators to calculate a similarity matrix from compound particularly suited to study differences in similarity datasets. The datasets used were characterised by down to the level of clones (Baayen et al., 2000). The considerable heterogeneity, containing character- high concordance of the overall similarities of the data based data like various coding and non-coding DNA with AFLP supports findings that F. langsethiae may sequences, chromatograms coded as XY values and have evolved as a sister species closely related to F. phenotypical data coded as a binary matrix. The value sporotrichioides rather than to F. poae. However, the of all these different types of data in separating taxa at type of data derived by AFLP analysis is inaccessible to different taxonomical levels differs between the var- phylogenetic analysis and therefore it is not possible to ious methods used and this may be one of the draw- further interpret the results obtained in that direction. backs of such a multidisciplinary approach. The tools Various attempts were made during the current used here are very powerful from the technological studies to develop pairs of PCR primers which enable point of view, but results and data have to be evalu- specific diagnosis of the taxa treated. Yet, none of the ated and interpreted very carefully in order to draw the primers derived from sources like IGS (Konstantinova right conclusions. However, the results obtained dur- and Yli-Mattila, 2004), ITS (S.S. Klemsdal, unpub- ing the current study may point out the way to deal lished data), or the tri5 gene (Niessen et al., 2004) with complex taxonomical problems in the future. were diagnostic for F. langsethiae. Most of them detected this species together with F. sporotrichioides, supporting their close phylogenetic relationship. Only Acknowledgements when combinations of two or more separate PCR assays were used or when techniques like ARMS- Holger Schmidt wants to thank M. Korakli and M. PCR and DGGE were applied with tub1 gene-based Tieking for fruitful discussions and N. Magan and S. primers in pure cultures, could the species be resolved Knoll for critical reading of the manuscript. We thank (Mach et al., 2003). C. Seeliger and S. Faulhammer for technical assis- tance. The scientific network behind this collaborative 4.1. Concluding remarks study was funded by the EC COST action 835. Parts of this work were supported by the EC FP5 ‘‘Quality In May 2000, a thorough treatise elucidating the of Life and Management of Living Resources’’ problems of modern fungal taxonomy from various program within the project ‘‘Early Detection of H. Schmidt et al. / International Journal of Food Microbiology 95 (2004) 341–349 349

Fusarium Species and Ochratoxigenic Fungi in Plant International Journal of Food Microbiology. doi:10.1016/ Products (DeToxFungi, contract QLK1-1999-01380), j.ijfoodmicro.2003.12.009. O’Donnell, K., Cigelnik, E., 1997. Two divergent intragenomic and in part by LMC Centre for Advanced Food rDNA ITS2 types within a monophyletic lineage of the fungus Studies (to UT). Fusarium are nonorthologous. Molecular Phylogenetics and Evolution 7, 103–116. O’Donnell, K., Cigelnik, E., Casper, H.H., 1998. Molecular phylo- References genetic morphological, and mycotoxin data support reidentifica- tion of the quorn mycoprotein fungus as Fusarium venenatum. Baayen, R.P., O’Donnell, K., Bonants, P.J.M., Cigelnik, E., Kroon, Fungal Genetics and Biology 23, 57–67. L.P.N.M., Roebroeck, E.J.A., Waalwijk, C., 2000. Gene gene- Schmidt, H., Niessen, L., Vogel, R.F., 2004. AFLP analysis of alogies and AFLP analyses in the Fusarium oxysporum complex Fusarium species in the section Sporotrichiella—evidence for identify monophyletic and nonmonophyletic formae speciales F. langsethiae as a new species. International Journal of Food causing wilt and rot disease. Phytopathology 90, 891–900. Microbiology. doi:10.1016/j.ijfoodmicro.2003.12.012. Hillis, D.M., Moritz, C., Mable, B.K. (Eds.), 1996. Molecular Sys- Seifert, K., Gams, W., Crous, P.W., Samuels, G.J. (Eds.), 2000. tematics Sinauer Associates, Sunderland Mass, USA 655 pp. Molecules, Morphology and Classification: Towards Monophy- Jukes, T.H., Cantor, C.R., 1969. Evolution of protein molecules. In: letic Genera in the Ascomycetes. Studies in Mycology, vol. 45. Munron, H.N. (Ed.), Mammalian Protein Metabolism. Academ- 230 pp. ic Press, New York, pp. 21–132. Shepard, R.N., 1962. The analysis of proximities: multidimensional Kendrick, B. (Ed.), 1979. The Whole Fungus: The Sexual–Asexual scaling with an unknown distance function I and II. Psychome- Synthesis. Proceedings of the Second International Mycological trika 27, 125–140, 219–246. Conference held at the Environmental Sciences.Centre of the Sneath, P.H., Sokal, R.R., 1973. Numerical Taxonomy Freeman, National Museum of Natura Sciences, vol. 1 and 2. National San Francisco California, USA 573 pp. Museums of Canada and the Kananaskis Foundation, Ottawa, Sokal, R.R., Michener, C.D., 1958. A statistical method for evalu- Canada 793 pp. ating systematic relationships. University of Kansas Science Knutsen, A.K., Torp, M., Holst-Jensen, A., 2004. Phylogenetic Bulletin 38, 1409–1438. analyses of the Fusarium poae, F. sporotrichioides and F. lang- Suga, H., Hasegawa, T., Mitsui, H., Kageyama, K., Hyakumachi, sethiae species complex based on partial sequences of the trans- M., 2000. Phylogenetic analysis of the phytopathogenic fungus lation elongation factor-1 alpha gene. International Journal of Fusarium solani based on the rDNA-ITS region. Mycological Food Microbiology. doi:10.1016/j.ijfoodmicro.2003.12.007. Research 104, 1175–1183. Konstantinova, P., Yli-Mattila, T., 2004. IGS-RFLP analysis and Thrane, U., Poulsen, S.B., Nirenberg, H.I., Lieckfeldt, E., 2001. development of molecular markers for identification of Fusa- Identification of Trichoderma strains by image analysis of HPLC rium poae, F. langsethiae, F. sporotrichioides and F. kyush- chromatograms. FEMS Microbiology Letters 203 (2), 249–255. uense. International Journal of Food Microbiology (this issue). Thrane, U., Adler, A., Clasen, P.-E., Galvano, F., Langseth, W., Lee, Y.M., Choi, Y.K., Min, B.R., 2000. Molecular characterization Lew, H., Logrieco, A., Nielsen, K.F., Ritieni, A., 2004. Diver- of Fusarium solani and its formae speciales based on sequences sity in metabolite production by Fusarium langsethiae, F. poae analysis of the internal transcribed spacer (ITS) region of ribo- and F. sporotrichioides. International Journal of Food Microbi- somal DNA. Mycobiology 28, 82–88. ology. doi:10.1016/j.ijfoodmicro.2003.12.005. Lieckfeldt, E., Seifert, K.A., 2000. An evaluation of the use of ITS Torp, M., Adler, A., 2004. Introduction. The European Sporotri- sequences in the taxonomy of the . Studies in My- chiella project: a polyphasic approach to the biology of a new cology 45, 35–44. Fusarium species. International Journal of Food Microbiology. Mach, R.L., Kullnig-Gradinger, C.M., Farnleitner, A.H., Reischer, doi:10.1016/j.ijfoodmicro.2003.12.015. G., Adler, A., Kubicek, C.P., 2004. Specific detection of Fusa- Torp, M., Langseth, W., 1999. Production of T2-toxin by a Fusa- rium langsethiae and related species by DGGE and ARMS-PCR rium resembling Fusarium poae. Mycopathologia 147, 86–96. of a (tub1) h-tubulin gene fragment. International Journal of Torp, M., Nirenberg, H.I., 2004. Fusarium langsethiae sp. nov. on Food Microbiology. doi:10.1016/j.ijfoodmicro.2003.12.011. cereals in Europe. International Journal of Food Microbiology. Marasas, W.F.O., Nelson, P.E., Toussoun, T.A. (Eds.), 1984. Toxi- doi:10.1016/j.ijfoodmicro.2003.12.014. genic Fusarium Species—Identity and Mycotoxicology. The Waalwijk, C., deKoning, J.R.A., Baayen, R.P., Gams, W., 1996. Pennsylvania State University Press, University Park, PA. Discordant groupings of Fusarium spp. from sections Elegans, 328 pp. Liseola and Dlaminia based on ribosomal ITS1 and ITS2 Nielsen, N.-P.V., Carstensen, J.M., Smedsgaard, J., 1998. Aligning sequences. Mycologia 88, 361–368. of single and multiple wavelength chromatographic profiles for Yli-Mattila, T., Mach, R.L., Alekhina, I.A., Bulat, S.A., Koskinen, chemometric data analysis using correlation optimised warping. S., Kullnig-Gradinger, C.M., Kubicek, C.P., Klemsdal, S.S., Journal of Chromatography A 805, 17–35. 2004. Phylogenetic relationship of Fusarium langsethiae to Niessen, L., Schmidt, H., Vogel, R.F., 2004. The use of tri5 gene Fusarium poae and F. sporotrichioides as inferred by IGS, sequences for PCR detection and taxonomy of trichothecene ITS, h-tubulin sequences and UP-PCR hybridization analysis. producing species in the Fusarium section Sporotrichiella. International Journal of Food Microbiology (this issue).