Estimates of global fungal numbers rely heavily on the ratios of unique fungi to host plants. Evidence for host-specificity, which is basic to our understanding of host to fungus ratios, is therefore explored in this review. There is considerable evidence that some endophytes, pathogens and mycorrhizal fungi are host-specific. Host-specificity however, may be an unwise term for saprobic fungi, and ‘host-exclusivity’ and ‘host-recurrence’ may be more appropriate terms. Concepts of host-specificity, host-exclusivity and host-recurrence are therefore defined and discussed. Suggestions for future work needed in order to establish host-exclusivity and host-recurrence in saprobic fungi are made.

The effects of a prescribed fire on the presence of ectomycorrhizal fungi on the root tips of ponderosa pine were investigated one year after a prescribed ground fire. Ectomycorrhizas were sampled within 1 m2 plots before and one year after the fire, and in nearby control plots that were not burned. The cores were divided into litter/organic, upper mineral, and lower mineral layers. The total ectomycorrhizal biomass in the control plots did not differ between year one and year two samples for any core layer, while in the fire plots the destruction of the litter/organic layer resulted in an eight-fold reduction in total ectomycorrhizal biomass. Mycorrhizal biomass in the two mineral layers was not significantly reduced by the fire. We used molecular tools to identify fungi directly from the ectomycorrhizas. In unburned plots members of the Russulaceae and Thelephoraceae were among the most frequent and abundant ectomycorrhizal types; species of most other taxa were rare. In the control plots these two families were among the dominant species in both years, but patchiness on a fine spatio-temporal scale caused some major changes in ranking of individual species between sample years. Rhizopogon subcaerulescens was the most pronounced example; its biomass in the control plot samples was seven times greater in year two than in year one because of an exceptionally large cluster of mycorrhizas encountered in a single core. The effect of fire on individual species was difficult to assess because of this patchiness and because all species were low in abundance after the fire. The most abundant pre-fire species were reduced to undetectable post-fire levels, while several less abundant species, including R. subcaerulescens, Cenococcum geophilum, and several unknown types were not substantially reduced by the fire. We speculate that the more abundant species, Martellia sp., thelephoroid 1, and Tomentella sublilacina, were differentially affected because their dominance was most prominent at the litter and organic layers. In any case, a short-term effect of this fire appears to be increased species evenness.

The infection of wheat spikes by Fusarium culmorum, one of the agents responsible for wheat head blight, was examined by light and electron microscopy. Macroconidia of the pathogen germinated 6–12 h after inoculation on all host surfaces they contacted. Developing germ tubes did not infect host tissues immediately, but gave rise to hyphae that grew and branched on host surfaces. Hyphal networks were usually formed 2 d after inoculation (dpi) on the inner surfaces of lemma, glume and palea, but not on the outer surfaces of lemma, glume and rachis. Hyphae on the outer surfaces of lemma and glume often grew over their edges to reach their inner surfaces. Penetration of host tissues occurred by infection hyphae on the inner surfaces of lemma, glume and palea, and on the upper part of the ovary. Occasionally, the pathogen invaded the host tissues through stomatal openings on the inner surface. Thereafter, the pathogen spread downwards to rachilla and rachis node by inter- and intracellular growth from above infected tissues. From the rachis node, hyphae extended downward to rachis and upward to peduncle through vascular bundles and cortical parenchyma tissue. When the pathogen arrived at the rachis 4–5 dpi, hyphae grew upwards and downwards inter- and intracellularly in vascular bundles and cortical parenchyma tissue of the rachis. During colonisation of the wheat spike, a series of alterations occurred in host tissues, including degeneration of host cytoplasm and organelles, collapse of parenchyma cells, disintegration or digestion of host cell walls and appearance of electron-dense coating materials on vessel walls.

Basidiomycete mycelia are ubiquitous in forest soils where they fulfil a range of key ecological functions. Population studies, based largely on basidiome collections, indicate that mycelia of many ectomycorrhizal and saprotrophic basidiomycetes can spread vegetatively for considerable distances through soil, but the extent to which these become physically or physiologically fragmented is unclear. This review considers aspects of the distribution, dynamics and translocatory activities of individual basidiomycete mycelia in forest soil, highlighting current gaps in our understanding and possible ways to address these.

Consumption of locoweeds, legumes endemic in arid western USA, has long been associated with locoism, a disease of ruminant animals. To explore the relationship between fungi associated with locoweed and locoweed toxicity, 11 locoweed populations from various sites in New Mexico were assessed for endophytic fungi. Endophytes were isolated from the leaves, stems, seeds, and flowers of eight populations of the toxic locoweeds Astragalus mollissimus, Oxytropis lambertii, and O. sericea. Fungal cultures grew very slowly and sporadically produced subcylindrical conidia with very dark transverse septa. All cultured endophytes produced the alkaloid swainsonine, which causes locoism. Endophyte-infected locoweed populations produced swainsonine, and the swainsonine level of endophyte strains in vitro was highly correlated with the swainsonine level of their host plant populations. The rDNA ITS from mycelia from four endophyte isolates and β-tubulin encoding regions from mycelia of 18 fungal endophyte isolates were amplified using PCR and the nucleic acid sequences were analyzed. The nucleic acid sequences of the β-tubulin encoding regions were essentially identical among all the endophytes regardless of plant genus and locations. Morphological evidence and sequence analysis of the ITS region suggest that the endophytes are most closely related to Embellisia. However, with the paucity of Embellisia species represented in sequence databases, precise taxonomic placement will await further study.

This review summarizes existing information about mushroom lectins, with an emphasis on those from the following species which have been most extensively characterized including various Agaricus species, Amanita pantherina, Boletus satanas, Coprinus cinereus, Ganoderma lucidum, Flammulina velutipes, Grifola frondosa, Hericium erinaceum, Ischnoderma resinosum, Lactarius deterrimus, Laetiporus sulphureus, Tricholoma mongolicum and Volvariella volvacea. It is noted that the mushroom lectins exhibit a diversity of chemical characteristics. Some of them are monomeric, whereas others are dimeric, trimeric or tetrameric. Their molecular weights range from 12 to 190 kDa, and the sugar contents from 0 to 18%. Carbohydrate specificities involve mainly galactose, lactose and N-acetylgalactosamine. A small number of mushroom lectins are specific for fucose, raffinose, N-glycolyneuraminic acid and N-acetyl-d-lactosamine. Studies on immunomodulatory and antitumour/cytotoxic activities have been carried out on lectins from Agaricus bisporus, Boletus satanas, Flammulina velutipes, Ganoderma lucidum, Grifola frondosa, Tricholoma mongolicum and Volvariella volvacea.

Random Amplified Microsatellite (RAMS) markers were used to detect variation among Phlebiopsis gigantea isolates from seven European countries. The scoring of 26 bands revealed five markers common to all isolates and 21 variable markers, of which six occurred in only one strain. A considerable amount of variation was observed among the progeny isolates of single strains. The analysis of variable markers resulted in 75 different banding patterns among the 86 isolates studied. The high degree of polymorphism and heterozygosity observed in RAMS markers suggests that the amount of genetic variation within this fungus is considerable. The equal distribution of markers in strains from different locations and the lack of country-specific markers suggests that genetic differentiation among the populations is low. The lack of distinct groups of similar banding patterns supports the idea of P. gigantea being a true biological species consisting of a single intersterility group.

Previously undescribed features on the surfaces of powdery mildew conidia were revealed by the scanning electron microscope (SEM), reinforcing differences observable by light microscopy (LM). Four distinct patterns were observed on the septa and 10 on the outer wall, thus categorizing 15 anamorph taxa. Anamorphs of Leveillula, Phyllactinia and Pleochaeta, all had smooth to moderately verrucose septa, but each bore a different distribution of warts on their verrucose outer wall. Annular septa and an echinulate outer wall were unique for the Oidium anamorph of Blumeria. The remaining powdery mildews, all with Oidium states, had either whorled or fibrillar septa. Those with whorled septa are separated into three newly proposed subgenera: Fibroidium for the anamorphs of Podosphaera and Sphaerotheca having conidia with fibrosin bodies and a smooth outer wall; Octagoidium for the dendritic patterned, eight-sided conidia of Sawadaea; and Setoidium for the anamorph of Cystotheca reflecting the setae on the mycelium. Three new subgenera are proposed for those with fibrillar septa: Reticuloidium for conidia with a roughcast outer wall of Erysiphe sect. Golovinomyces; Graciloidium for the unswollen conidia of Arthrocladiella; Striatoidium for conidia of E. sect. Galeopsidis with a striate outer wall. The name Pseudoidium is retained as a subgenus for the anamorphs of E. sect. Erysiphe, Microsphaera and Uncinula which had conidia with a scaly, fibrous outer wall. The patterns on the outer wall of conidia were frequently modified by secondary creasing. These wrinkling patterns, also observable under the LM, proved useful in identification of anamorphs, especially on herbarium specimens. SEM, LM and host range data were used to construct keys for the prediction of the anamorphs of all the powdery mildew genera found in the U.K. The findings, together with teleomorph criteria, also prompted a revision of the Erysiphaceae.

Verticillium wilt in winter-sown oilseed rape (Brassica napus L. ssp. oleifera), which has not yet been reported in the U.K. but is widespread in Europe, was shown to be caused by the host-specific, ‘near-diploid’ Verticillium longisporum comb. nov. Thirty-one cruciferous isolates (plus one from sugarbeet) of V. longisporum, mainly from oilseed rape from Germany, Sweden, Japan, France and Poland, were characterized and compared with nine typical isolates of V. dahliae and five of V. albo-atrum. Isolates of V. longisporum were distinguished from those of V. dahliae by three morphological characters, i.e. elongate microsclerotia, long conidia (7·1–8·8 μm) and mainly three phialides per node on conidiophores, whereas those of V. dahliae had ±spherical microsclerotia, short conidia (3·5–5·5 μm), and 4–5 phialides per node. Isolates of V. longisporum lacked extracellular polyphenol oxidase activity (p.p.o.); they showed mean conidial nuclear diam. (DAPI fluorescence) of ca 1·85 μm, and ‘near-diploid’ standardized arbitrary DNA values (Feulgen DNA microdensitometry) of 0·89–1·17 (mean, 1·02). For isolates of V. dahliae, extracellular p.p.o. activity was detectable, and the corresponding figures for conidial nuclear diam. were ca 1·16 μm and DNA values of 0·45–0·65 (mean, 0·57), respectively. Using three oligonucleotide primers, isolates of V. longisporum were clearly distinguishable from those of V. dahliae and V. albo-atrum by their RAPD band profiles. Greenhouse pathogenicity tests, employing five winter oilseed rape cvs, confirmed the pathogenicity of V. longisporum, whereas V. dahliae was non-pathogenic. On the basis of all the above characters, this host-specific pathotype, V. longisporum, should now be considered as a distinct species. Evidence is presented to suggest that it may have evolved by parasexual hybridization between a strain of V. albo-atrum and a strain of V. dahliae, thus explaining its ‘near diploid’ state and the origin of four recombinants detected.

Liquid media with differing carbon concentrations and carbon-to-nitrogen ratios were tested for production of desiccation tolerant blastospores of Paecilomyces fumosoroseus. While all media tested supported sporulation in submerged culture, high blastospore concentrations (5·8 × 108 spores ml−1) were produced in media containing 80 g glucose l−1 and 13·2 g Casamino acids l−1 (MS medium) and a significantly higher percentage (79%) of these blastospores survived air drying. Media containing glucose concentrations greater than 20 g l−1 and Casamino acid concentrations between 13·2 and 40 g l−1 supported maximal production of desiccation tolerant blastospores. All 23 isolates of P. fumosoroseus grown in MS media produced high concentrations of desiccation tolerant blastospores. When stored at 4 °C, more than 60% of the lyophilized blastospores produced in MS medium were still viable after 7 months storage while less than 25% of the air-dried blastospores survived after 90 d storage. Standard whitefly bioassays were performed to compare air-dried blastospores of P. fumosoroseus ARSEF 4491 with solid substrate-produced conidia of Beauveria bassiana ARSEF 252. Air-dried blastospores of P. fumosoroseus gave LD50S of 60 and 113 blastospores mm−3 for the silverleaf whitefly (Bemisia argentifolii) in two separate bioassays with potency ratios (LD50 B. bassiana/LD50 P. fumosoroseus) of 3·9 and 3·8, respectively. These results have demonstrated that high concentrations of blastospores of P. fumosoroseus can be rapidly produced in liquid culture, remain viable following drying, and infect and kill silverleaf whitefly.

In order to estimate phylogenetic relationships in the Auriculariales sensu Bandoni (1984) and allied groups we have analysed a representative sample of species by comparison of nuclear coded ribosomal DNA sequences, applying models of neighbour joining, maximum parsimony, conditional clustering, and maximum likelihood.

Analyses of the 5′ terminal domain of the gene coding for the 28 S ribosomal large subunit supported the monophyly of the Dacrymycetales and Tremellales, while the monophyly of the Auriculariales was not supported. The Sebacinaceae, including the genera Sebacina, Efibulobasidium, Tremelloscypha, and Craterocolla, was confirmed as a monophyletic group, which appeared distant from other taxa ascribed to the Auriculariales. Within the latter the following subgroups were significantly supported: (1) a group of closely related species containing members of the genera Auricularia, Exidia, Exidiopsis, Heterochaete, and Eichleriella; (2) a group comprising species of Bourdotia and Ductifera; (3) a group of globose-spored species of the genus Basidiodendron; (4) a group that includes the members of the genus Myxarium and Hyaloria pilacre; (5) a group consisting of species of the genera Protomerulius, Tremellodendropsis, Heterochaetella, and Protodontia. Additional analyses of the internal transcribed spacer (ITS) region of the species contained in group (1) resulted in a separation of these fungi due to their basidial types.

Epichloë spp. (Clavicipitaceae) and their close asexual relatives, Neotyphodium spp., form systemic endophytic associations with Pooideae grasses. Interactions between Lolium and Festuca host grasses and fungal endophytes were examined in studies focusing on leaves of natural associations and also of plants into which the endophyte was introduced by seedling inoculation. Light microscopy as well as fungal isolation was used to locate the position of hyphae, while transmission electron microscopy was used to examine host/fungus interactions at the cellular level. These studies provide support for synchronised plant and endophyte growth. This characteristic pattern of growth was maintained when both Epichloë and Neotyphodium spp. were introduced by seedling inoculation into new host species. Hyphae, with few exceptions, grew rapidly as leaves grew and ceased when leaf growth ceased. This pattern of growth offers an explanation for the characteristic appearance of hyphae in leaf sheaths of host grasses; seldom branched and for all species other than N. occultans, aligned parallel to the leaf axis. Hyphal growth of a second group of endophytic fungi, referred to as p-endophytes, was not regulated in the same way, with growth continuing as leaves aged. This pattern of growth gives rise to high concentrations of branched ramifying hyphae in old leaf sheaths. Although host genotype did not effect the basic pattern of hyphal growth of Epichloë/Neotyphodium endophytes, it strongly influenced the concentration and distribution of hyphae throughout leaves. Examination by TEM revealed no evidence that penetration of hyphae into dense tissue was aided by the secretion of pectic enzymes that loosen the middle lamella connecting host cells. Instead it appeared that penetration occurred by hyphae physically pushing between cells. The absence of enzymatic loosening of the middle lamella during penetration could explain why hyphae of these endophytes apparently do not elicit host defence reactions in natural associations. However, subtle interactions between the hyphae of Epichloë/Neotyphodium and host cells in natural associations were observed.

Production of the carcinogenic mycotoxins fumonisins has been reported in several Fusarium species, most of which are members of the Gibberella fujikuroi (Gf) complex. In this study, we examined 15 Fusarium species in the Gf complex and 12 other species for fumonisin production and the presence of fumonisin biosynthetic genes (FUM). Among the species within the Gf complex, fumonisin production was detected only in F. fujikuroi, F. globosum, F. proliferatum, F. nygamai, F. oxysporum and F. verticillioides. These five species include members of two of the three major clades delineated in the Gf complex. The FUM genes were detected in these same five species and in F. anthophilum, a member of the third clade. Among the species outside the Gf complex, fumonisin production and FUM genes were detected only in F. oxysporum. Phylogenetic analyses of nucleotide sequences from two FUM gene fragments inferred relationships similar but not identical to those inferred from previous analyses of other genes. The results indicate the FUM genes are discontinuously distributed in the Gf complex and that this distribution gives rise to the differences in the abilities of closely related Fusarium species to produce fumonisins.

The genetic structure within and between USA and European populations of the emerging phytopathogen Phytophthora ramorum was examined. Four primer combinations were used for amplified fragment length polymorphism (AFLP) fingerprinting of 67 USA isolates from California and Oregon, and 18 European isolates from Belgium, Germany, The Netherlands, Spain and the UK. In addition, three DNA regions (ITS, cox II, and nad 5) of additional Phytophthora species were amplified by polymerase chain reaction, sequenced, and analysed to provide better phylogenetic understanding of P. ramorum within the genus Phytophthora. AFLP banding patterns indicate that the 85 isolates form two distinct lineages within a monophyletic group, distinct from the closely related outgroup species P. lateralis. With the exception of two isolates from an Oregon nursery, European and USA isolates clustered separately within individual clades. The AFLP profiles also indicate that a single clonal lineage dominates the North American population, while the European population consists of an array of mainly unique, closely related AFLP types. Sequences from the three DNA regions were identical among all P. ramorum isolates, and phylogenetic analysis indicates that P. ramorum is closely related to P. lateralis and P. hibernalis.

Hybrids of Melampsora medusae and M. occidentalis are herein described as M. ×columbiana. This hybrid taxon is characterized by isolates that generally exhibit morphological intermediacy in uredinial and telial traits, and mixed virulence/avirulence on Populus trichocarpa and P. deltoides, the natural hosts of M. occidentalis and M. medusae, respectively. Hybrids frequently contain ribosomal ITS sequences from both Melampsora species, but discordant isolates were also found, like M. medusae in having a urediniospore equatorial smooth spot, and like M. occidentalis in spine density, and intermediate in urediniospore length. Isolates with hybrid urediniospore morphology but homozygous for ITS sequences from M. medusae or M. occidentalis were also found, suggesting that the primary F1 hybrids have produced F2 and/or backcross progeny. In a 1997 survey of the leaf rust population on hybrid poplar in the Pacific Northwest, this ‘new population’ of Melampsora ×columbiana was the only taxon found. Specimens from nearly a century ago indicate that an older hybrid population is of wide distribution in North America. This ‘older population ‘of Melampsora ×columbiana was found on specimens of all species of Populus in sections Aigeiros and Tacamahaca, and on many of their hybrids, in areas outside the type localities of M. occidentalis and M. medusae. These three Melampsora taxa, M. medusae, M. occidentalis, and M. ×columbiana, are the only ones that occur naturally on sections Aigeiros and Tacamahaca of Populus in North America.

Epichloë typhina is a biotrophic symbiont of grasses with its sexual state on immature host inflorescences. The ectophytic, presexual structure (stroma) of E. typhina produces spermatia and also serves as the female structure in mating. After fertilization there is proliferation of a dense mycelium which has been suggested to be heterokaryotic. Perithecia form in the thickened stroma. We investigated the possible formation of heterokaryons in matings and the role of ascospores in contagious spread of E. typhina. In almost all instances the transfer to a stroma of spermatia of opposite mating type leads to thickening of the stroma; however, ascospores were produced only if the parents were of the same mating population. Fertilization of part of a stroma by one spermatial parent often inhibited fertilization by another strain elsewhere on the stroma depending upon the spermatial isolates used. Tests for heterokaryon formation were performed by culturing stromata after matings that produced no ascospores, to avoid analysing meiotic products. In two instances the female was cultured, and in two the nuclear and mitochondrial haplotypes were primarily of the male. In another instance mitochondrial haplotypes and mitochondrial plasmids from the female became associated with the male nuclear haplotype. These results suggest that, following mating, male hyphae proliferate and heterokaryons may sometimes form and also proliferate. To test how ascospores mediate infection, inflorescences of uninfected perennial ryegrass plants were surrounded by fertilized stromata from which E. typhina ascospores were being ejected, seeds were collected and grown, newly infected progeny plants were identified, and isolates from five progeny plants were analysed genetically. As expected for ascospore progeny the rDNA haplotype of the stromal and spermatial parents had segregated, whereas all five isolates had the stromal profile of mitochondrial DNA. This observation demonstrated that ascospores mediate infection of new host plants.

A toxic protein, bassiacridin, was purified from a strain of the entomopathogenic fungus Beauveria bassiana isolated from a locust, using chromatographic methods. The final toxic fraction contained between 0.1 and 0.3% of the proteins of the crude extract. Bassiacridin showed no affinity for ion exchangers, was characterised as a monomer with a mol. wt of 60 kDa and an isoelectric point of 9.5, and exhibited β-glucosidase, β-galactosidase and N-acetylglucosaminidase activities. Injection of fourth instar nymphs of Locusta migratoria with the pure protein at relatively low dosage (3.3 μg toxin g body wt−1) caused a rate of mortality near to 50%. The effects of the crude and pure fractions were characterized at tissular and cellular levels. The formation of melanised spots on tracheae and air sacs and of melanised nodules in contact with the fat body was observed in injected locusts. Alterations of the fine structure of epithelial cells of tracheae, air bags, and integument were also revealed. The insecticidal protein showed a specific activity against locusts. Bassiacridin is different from the other macromolecular toxins of entomopathogenic fungi already described. Microsequencing of peptides generated by trypsic digestion of bassiacridin confirmed that it is a novel molecule and showed that it exhibits a probably limited similarity with a chitin binding protein from yeast.

Leptosphaeria maculans (anamorph Phoma lingam), the ascomycete causing stem canker of crucifers, is a species complex that can be separated into at least seven distinct subgroups using a combination of biochemical and molecular criteria. In the present study sequences of the entire ITS region, including the 5.8S rDNA, of 38 isolates representing the seven subgroups, along with specimens from culture collections, were analysed, compared to those of closely related Leptosphaeria species, and the phylogeny inferred using parsimony and distance analyses. A well-supported clade encompassed all isolates of the seven subgroups along with L. conferta, a known saprobe of dried crucifer stems. The L. maculans isolates were further separated into two well-supported clades corresponding to L. maculans s. str. and the recently named L. biglobosa. Parsimony and distance analyses further separated groups within both species, usually corresponding to specific host plants or geographic origin, e.g. L. maculans ‘brassicae’ from cultivated Brassica, L. maculans ‘lepidii’, from Lepidium sp., L. biglobosa ‘brassicae’, from various Brassica species, L. biglobosa ‘thlaspii’ from Thlaspi arvense, L. biglobosa ‘erysimii’ from Erysimum sp., and L. biglobosa ‘canadensis’ mostly found in central Canada. The oldest L. maculans specimens maintained in international collections clustered with either L. maculans ‘brassicae’, L. biglobosa ‘brassicae’, or a still different group closely related to L. biglobosa ‘thlaspii’. The evolutionary relationships between the seven infraspecific groups are discussed in terms of phytopathological relevance and species isolation linked with specific life cycle, geographic isolation or host specificity.

Fungi developing on dung samples, from a wide range of locations and incubated in moist chambers, were recorded. Highly significant differences were found among the mycobiota of different dung types, from different latitudinal ranges, and collected at different seasons. Sheep, cattle, deer, rabbit, hare and grouse provided 86 % of the 425 samples. Highly significant differences in the community composition of the mycobiota of these six dung types were observed. Coprinus stercoreus was the commonest basidiomycete, and most frequent on sheep, cattle, deer and rabbit; C. miser was the only common species, of any taxonomic group, which showed no significant difference in its frequency on different dungs. Apothecial fungi comprised about a quarter of the records, with Ascobolus, Saccobolus and Thelebolus spp., Iodophanus carneus and Lasiobolus cuniculi the commonest. A. carletonii occurred uniquely, but frequently, on grouse. Perithecial species, especially of Schizothecium, Podospora, Coniochaeta and Sporormiella, accounted for ca 50% of all records. The grouse dung mycobiota was markedly different, with a lower species richness (mean 3.2 per sample), from that of the mammalian herbivores, which had means of 9–12 spp. per sample. In temperate latitudes more species were recorded on samples in winter. Species richness was greater in samples from lower latitudes. Significant differences in the composition of different dung types are discussed in relation to their fungal communities. The usefulness of studying the fungi of the dung microcosm as a means of rapid biodiversity assessment is considered, by comparing the results with those of a similar, but smaller, study 30 years earlier, and by examining the diversity of a sample from a subjectively poor habitat in Morocco. There was no indication of any major change in the coprophilous mycobiota that might be associated, e.g. with climatic change or changed farming practice over the last three decades. The Moroccan samples had a much lower diversity (80 spp./50 samples) than would be expected of samples from that latitude (146 spp./50 samples) based on the worldwide data set.

Endophytic fungi were isolated from 7500 samples of wild Musa acuminata collected from five sites at Doi Suthep Pui National Park, Thailand during December 1998 to July 1999. Overall colonization rates from surface sterilized tissues were 56·5, 48·9, 48, 47·9 and 41·7% for the Medicinal Plant Garden, Ban Suthep, Queen Sirikit Botanic Garden, San Gu, and Montatarn waterfall sites respectively. Sixty-one different fungal taxa were isolated. Fewer isolates were recovered from younger than older samples. Xylariaceous taxa and Guignardia cocoicola were the most frequently isolated endophytes from leaves and were either absent or rare in midrib, petiole and pseudostem. Colletotrichum gloeosporioides, C. musae, Guignardia cocoicola, various sterile mycelia and xylariaceous spp. were common at all sites. The endophyte fungal communities at the five sites were found to differ. Deightoniella torulosa was the most frequent isolate at the Ban Suthep site and was either absent or rare at other sites. Colletotrichum species were most common in the midribs and petioles at all sites, while Pyriculariopsis parasitica and Dactylaria sp. were most common in the pseudostems. The endophyte communities isolated from M. acuminata in this study are compared with those from previous studies on tropical hosts. Several of the endophytes isolated are established pathogens of banana and provide support for the hypothesis that some endophytes are latent pathogens. The diversity of fungi on banana is discussed in relation to global estimates of numbers of fungus species.