We studied 86 strains of Trichoderma, any of which could have been identified as T. koningii following the schemes of Rifai (1969) or Gams & Bissett (1998). In the absence of reproductive isolation and clear morphological differentiation for detecting species boundaries, the GCPSR concept (Taylor et al. 2000) remains the only currently applicable choice. It requires the concordant phylogenetic position of a taxon among closely related other taxa based on at least three unlinked loci. The concordant phylogenetic position should also not be contradicted by analyses of other loci. In this study we found few phylogenetic markers that could reliably resolve groups of closely related, apparently recently evolved Hypocrea/Trichoderma species. The choice of phylogenetic markers for a particular group of fungi is a delicate task. Druzhinina & Kubicek (2005) have listed eleven phylogenetic markers attributed to eight DNA loci used in phylogenetic analyses of Hypocrea/Trichoderma species. The internal transcribed spacers 1 and 2 (ITS1 and 2), which provide considerable diagnostic properties in Hypocrea/Trichoderma (Druzhinina et al. 2005), are insufficient for phylogenetic modelling even at the intercladal level. Therefore, for the analysis of T. koningii-like species, we have selected intron-rich fragments of the protein-encoding genes tef1, act, and cal that deliver higher levels of variation. However, sufficiently high resolution was obtained only from both tef1 introns, while the phylogenetic signal of act and cal was moderate. Tree topologies based on act and cal loci are concordant with the selected tree based on tef1, however, overall statistical support of species nodes was relatively low. Therefore, an integrated approach for the identification of Trichoderma species was developed that adopts multivariate analyses of phenotypic characters and patterns of geographical distributions as well as phylogenetic inferences and oligonucleotide barcoding (see also Kraus et al. 2004, Harrington & Rizzo 1999).
The molecular phylogenetic analysis based on three protein-encoding genes revealed several weakly or well supported clades representing taxa that could be characterized and therefore recognized at least partly by phenotypic characters and patterns of geographic distributions. With the help of this integrated approach, a taxonomy was developed that currently accepts twelve species and one variety.
Trichoderma, like many other fungi, suffers from homoplasy of morphological characters, which have been the basis of species descriptions since the genus was described more than 200 years ago. In the absence of characters derived from DNA, all of the isolates that we studied could have been identified as T. koningii in the sense of Rifai (1969). In the present work, geographic distribution and rate of growth in agar culture were among the most significant characters separating species. PCA that included only the few available characters of the anamorph per se (i.e. measurements of conidia and phialides) did not result in a clustering of strains that was consistent with the results of phylogenetic analysis. Certain characters of the anamorph are variable within a phylogenetic species; others suggest homoplasy in different phylogenetic species. There are usually no finite conidiophores; instead, conidiophores are aggregated within more or less well-developed pustules and individual elements cannot be measured and are difficult to characterize. The formation and extent of pustules, or degree of aggregation of "conidiophores", in most species is highly variable and the ability to form pustules may decline with length of time of strain preservation or after successive transfers. Phialides in most groups are longer or shorter, wider or narrower, depending upon whether they are formed near the surface of the pustule, where they are less crowded, or at the interior of the pustule where they are more crowded. However, crowded and less crowded phialides can hardly be analyzed separately from each other. Conidia present the most consistent morphological character because they can be measured and because their morphology remains constant over successive transfers. What compounds the difficulty in finding phenotypic characters is the general lack of pigments in cultures, a character that has been used in taxonomy of Trichoderma sect. Longibrachiatum (Samuels et al. 1998) or other species-rich genera such as Fusarium.
The teleomorph is not helpful in species recognition and is only of limited use in recognizing phylogenetically distinct clades. In general, individual species within a clade cannot be distinguished on the basis of the morphological characters of the teleomorph. However, clades may or may not have a distinctive teleomorph morphology. Chaverri & Samuels (2004) found that among species with green ascospores, the same anatomy of the Hypocrea stroma was found in phylogenetically distant groups, however, all of those teleomorphs were very different from the teleomorphs reported here and from those formed in T. sect. Longibrachiatum (Samuels et al. 1998). In gross morphology, most of the Hypocrea collections studied here cannot be distinguished from H. rufa (Pers.: Fr.) Fr., the teleomorph of the closely related T. viride, and the type species of Hypocrea. Doi (1972) subdivided Hypocrea of Japan on the basis of ascospore colour and stroma anatomy, but phylogenetic analysis has not upheld those subdivisions.
With the adoption of molecular phylogenetics, it has become obvious that a species delineated by morphological characters is likely to comprise multiple phylogenetic species. If we had limited our study of phenotype characters to those of the conidia, conidiophores and ascospores, we might have concluded (as did Rifai in 1969) that there were potentially more than one "cryptic species" within the morphological species T. koningii. We would not have been able, however, to identify most of them unambiguously. This phenomenon is common in the ascomycetes and led Hawksworth (2001) to revise upward the estimated number of species of fungi. Minute or subtle characters that would have, in an earlier time, been disregarded as insignificant, are now accepted and sought for characterization of clades. For example, the common species T. harzianum was distinguished from the cause of green mould of mushrooms, T. aggressivum Samuels & W. Gams, by the inability of the latter species to grow at 35 °C (Samuels et al. 2002).
The paucity of phenotypic characters to distinguish clades is not unique to Trichoderma but rather is likely to become increasingly found in the ascomycetes in general as genera are subjected to phylogenetic analyses. In the example of Coccidioides, Fisher et al. (2005) found that salt tolerance was the only phenotypic character to distinguish Coccidioides posadasii from the closely related C. immitis. Botryosphaeria has received much attention recently. Differences in yellow pigment and host distinguished between the phylogenetically closely related Botryosphaeria lutea and B. acaciae (Slippers et al. 2004). The Fusarium solani complex remains an important challenge to taxonomic revision (O'Donnell 2000).
As increasing numbers of members of species-rich genera such as Trichoderma, Fusarium or Botryosphaeria are included in phylogenetic analyses, increasing numbers of morphologically defined species will be found to be paraphyletic or comprise numerous cryptic species. As much as we would wish for a taxonomy that would permit accurate species identification using only the microscope, we must face the reality that there may not be enough characters in morphology and growth to reflect the differences revealed in diverging DNA sequences. Homoplasy of morphological characters may well be the result if genetically distinct lineages occupy the same niche for a long time. This is also confirmed by many cases where data for the overall carbon utilization profiles (Biolog) did not correspond to the phylogenetic analysis, while the analysis of the utilization of certain single carbohydrates did (Kubicek et al. 2003). A similar situation was observed in marine species of the unrelated genus Dendryphiella, collected from different marine sites (Dela Cruz & Druzhinina, unpublished).
One could argue that our species concept in Trichoderma is too narrow, too strongly influenced by phylogeny, but an example from the present study argues against that. Trichoderma rogersonii and T. petersenii are well separated in the phylogenetic analysis (Figs 2, 3). They are common and sympatric in the Eastern U.S.A. However, as can be seen from the PCA (Fig. 4), the two species are incompletely separated by a suite of phenotype characters. The characters separating the two species (Table 3) are the L/W and length of conidia and the length of the proximal part-ascospores. In addition, growth rates on SNA are also characteristic of the respective species. While these differences are statistically significant, there will be many cases that cannot be identified on the basis of their phenotype.
In Trichoderma, at least, traditional characters may be too few to provide practical identification of the apparent large number of species. Carbon utilization estimated using a Phenotype MicroArray technique may provide additional characters (Kubicek et al. 2003, Kraus et al. 2004). In Trichoderma we do not yet have the ability to perform in vitro mating experiments, which would help immensely in determining whether strains belong to the same or different biological species.
The species that we have studied in the current work represent a small part of the diversity of the species-rich "Viride Clade." Work is continuing on taxonomy of this group.
The refined definition of T. koningii given by Lieckfeldt et al. (1998) was reinforced in this study; the species is distinguished by its longer and narrower conidia and slow rate of growth on SNA. Although T. koningii is among the most commonly cited species in the genus, our results suggest that it is an uncommon species of Europe and North America. The far more common species isolated directly from natural substrata and the species often used in biological control applications is the closely related T. koningiopsis.
Trichoderma koningiopsis is essentially a tropical species, known from South America and Africa (Ethiopia), but its Hypocrea teleomorph has been found as far north as New York State. Trichoderma koningiopsis is the most commonly encountered species having a T. koningii-like morphology. This species was reported earlier as "Hypocrea sp. (8)" in part (Lieckfeldt et al. 1998), "T. koningii II" (Dodd et al. 2003) and "T. koningii Tkon 21" (Holmes et al. 2004).
Webster (1964) described "Hypocrea sp. 1" from the United Kingdom. The T. koningii-like anamorph described by Webster (1964) strongly resembles T. petersenii in the morphology of conidia, conidiophores and in the formation of concentric rings in agar culture. We have obtained the cultures cited by Prof. Webster from CBS. Based on anamorph morphology as observed in these cultures and sequences of tef1, we can see that Hypocrea sp. 1 was based on a mixture of two species. Webster 2534 = CBS 257.62 = T. harzianum, the anamorph of H. lixii (Chaverri & Samuels 2004); the other cultures (Webster 2545 = CBS 258.62, 2617 = CBS 259.62 and 2644 = CBS 260.62) are all T. minutisporum Bissett, the anamorph of H. minutispora (Lu et al. 2004). Neither of these species is closely related to members of the "Viride Clade" (Samuels 2006).
Doi (1974) reported a T. koningii-like anamorph for Japanese collections of H. muroiana Hino & Katumoto. However, the range of conidial types – including those with surface ornamentation – described by him lead us to suspect that more than one species was involved. Moreover, none of the collections cultured by Doi was taken from bamboo, which is the substratum of the type collection of H. muroiana. In our experience, bamboo is an unusual substratum supporting fungi that are not usually found on other substrata, at least not on woody substrata, which was the source of specimens reported by Doi. We have examined the type collection of H. muroiana (YAM) and conclude on the basis of its morphology that it is a member of the "Viride Clade," but there is no material with a living culture available. We are not able to identify to species the T. koningii-like anamorph(s) reported by Doi (1974) for H. muroiana. Two cultures isolated from rhizomorphs of, respectively, Armillaria mellea (IFO 31288) and Lentinula edodes (IFO 31293) and identified by Y. Doi as H. muroiana are T. atroviride, the anamorph of H. atroviridis.
Several isolates of T. koningiopsis, represented by G.J.S. 01-10 and G.J.S. 01-11 but not included in the phylogenetic analysis, were isolated in Ecuador from pods of Theobroma cacao that were infected by the destructive parasite Moniliophthora roreri. These isolates are currently in field trials to protect cacao from the Moniliophthora (C. Suarez, pers. comm.). Trichoderma koningiopsis isolates G.J.S. 04-10 and 04-11 are effective in protecting cotton plants from infection by Thielaviopsis basicola in Texas (C.R. Howell, pers. comm.), and isolate G.J.S. 05-462 (received too late to be included in the present study) is showing potential for control of Fusarium verticillioides in maize (I. Yates, pers. comm.). A single isolate of this species, G.J.S. 97-273 (= BBA 65450), was isolated from soil in Germany. We tested several isolates of T. koningiopsis for their ability to parasitize the cacao pathogen Moniliophthora roreri in vitro (results not shown) following the "preinoculated plate test" described by Evans et al. (2003) and found that several were able to parasitize the mycelium of M. roreri, with the German isolate being especially effective.
Trichoderma koningiopsis is probably cosmopolitan but perhaps more common in tropical regions. In Figs 2, 3 the species can be seen to comprise several well-supported internal branches; however, we were not able to detect any geographic or phenotypic bias to any of the clades. Six strains (designated as "DIS" in Fig. 2) of T. koningiopsis were isolated as endophytes from freshly exposed, living sap-wood of trunks of species of Theobroma in Brazil, Ecuador and Peru. Following the protocol described in Holmes et al. (2004), strains DIS 172ai (from Theobroma grandiflorum) and DIS 229d (from Th. gileri) could be introduced into seedlings of Theobroma cacao and were reisolated from woody tissue but not from the apical meristem. The isolate DIS 339c (from Th. gileri) could be reisolated from all stem sections of Th. cacao seedlings, including the apical meristems, and it could be reisolated from inoculated pods of Th. cacao after 12 weeks, indicating a potential for protecting pods against infection by M. roreri (K. Holmes, pers. comm.). Ecuadorian strains (G.J.S. 01-07–G.J.S. 01-12, Table 1) were isolated from pods of Th. cacao that were naturally infected with the parasite M. roreri, the cause of frosty pod rot, and have been included in a field trial in Ecuador against that pathogen (C. Suarez, pers. comm.).
Trichoderma koningiopsis occupies the most basal position of the Large Koningii Branch (LKB) in the tef1 tree (Fig. 2), although the statistical support of this species on both act and cal trees is particularly low. This finding in combination with confirmed wide distribution of the species in tropical countries may indicate a relatively intensive recombination process due to sexual reproduction. However, the majority of T. koningiopsis strains were isolated as anamorphs from natural substrata. Teleomorph specimens are only known from the Caribbean region and from the U.S.A. Alternately, the paraphyly of T. koningiopsis could be explained if the species were relatively old. Partially sympatric old, clonal lineages could occur sympatrically and, over evolutionary time, accumulated mutations in the introns and other parts of the DNA could explain the variation in the species.
Trichoderma petersenii and T. rogersonii are common and sympatric in eastern North America. Trichoderma petersenii was reported earlier as "Hypocrea sp. (8)" in part (Lieckfeldt et al. 1998), "T. koningii Tkon 3" and "Tkon 22" in part (Holmes et al. 2004). Trichoderma rogersonii was reported as "Hypocrea sp. (4) and (5)" (Lieckfeldt et al. 1998). The similarity between these two species was noted above. The most obvious difference between these very similar but phylogenetically relatively distantly related species is that T. rogersonii grows more slowly on SNA than does T. petersenii; moreover, conidia of T. petersenii are slightly shorter and broader than those of T. rogersonii (95 % CI of L/W respectively 1.35–1.39, 1.40–1.46). Their Hypocrea morphs are indistinguishable from each other and, at least in gross morphology, they are indistinguishable from H. rufa (anamorph: T. viride). However, H. rufa is an uncommon species, albeit sympatric with the other two, despite the many reports of its occurrence. It differs from T. petersenii and T. rogersonii in having slightly larger ascospores (distal part-ascospores approx. 4.5–5 x 4–4.5 µm; proximal part-ascospores approx. 5–6 x 3–4 µm). Trichoderma viride is readily distinguished from T. petersenii and T. rogersonii by its subglobose, warted conidia.
Trichoderma koningii is also sympatric with T. petersenii and T. rogersonii. Because these highly similar species are common, despite their phylogenetic distance from each other, it is important that they may be reliably distinguished by the ITS1 and 2 oligonucleotide barcodes.
Trichoderma ovalisporum is distinguished by its subglobose to more ovoidal conidia (Figs 235–236). It was found as an endophyte of Theobroma species and was also isolated from a woody stem of the liana Banisteropsis caapi that was infected by Moniliophthora (Crinipellis) perniciosa, the cause of Witches' Broom disease of cacao in tropical America (Holmes et al. 2004). The fifth isolate was isolated from soil in Panama, where cacao is grown; it was not included in the phylogenetic analysis. The liana isolate (DIS 70a) reinfected and was reisolated from meristematic tissue of Th. cacao, and inhibited radial growth of the frosty pod rot pathogen (Moniliophthora roreri) in vitro. It also persisted on the surface, and within tissues, of cocoa pods in the field for at least 10 weeks. Initial field trials in Costa Rica, where conidia were applied as a spray, indicated an ability to protect pods against infection by M. roreri (Holmes et al. 2004).
With the exception of the one isolate of
T. ovalisporum (DIS 70a), all of the DIS isolates studied for this work (
Table 1) were isolated as endophytes from woody stems of South American
Th. cacao, Th. gileri and other
Theobroma species. These and other
Trichoderma isolates were reported previously by Evans
et al. (
2003) and Holmes
et al. (
2004) to be endophytes of
Th. gileri. In the current work we identify several additional endophytes as members of the "Viride Clade" of
Trichoderma sect.
Trichoderma. Trichoderma koningiopsis was especially well represented in the endophyte isolations but they did not fall into an endophyte-specific lineage in this species. In contrast to
T. ovalisporum and
T. caribbaeum var.
aequatoriale, which are known only as endophytes of cacao and cacao relatives but which are not common even in that niche, it is not surprising that a species that is as common as
T. koningiopsis should be found as an endophyte of a common tropical tree. Judging by the large number of isolates of
T. koningiopsis that we received from many sources, it would be surprising not to find it as an endophyte of stems of other tropical trees. Equally, it would not be surprising to find additional isolates that have a biological control potential for fungus-induced plant diseases. Interesting is that Arnold & Herre (
2003) did not report
Trichoderma species as leaf endophytes in Panama. Outside of the
T. koningii aggregate species,
T. erinaceus (DIS 7, DIS 8 in
Fig. 2) was extended to Peru. This species was previously known only from Southeast Asia (Thailand, Cambodia, Malaysia).
Trichoderma stilbohypoxyli was originally described as a parasite of the xylariaceous fungus
Stilbohypoxylon muelleri in Puerto Rico in 1996. Since then several isolates considerably expand the biological and geographic distribution of this species by discovery of its teleomorph in Costa Rica and Ghana on bark and perithecia of
Neonectria jungneri, and endophyte isolations from woody tissue of
Theobroma species in Ecuador and Brazil and from
Fagus sylvatica in the United Kingdom. The diffusing yellow pigment, especially seen in colony reverse on PDA, and fast growth rate characterize this species.
Trichoderma caribbaeum var. caribbaeum is represented by two collections (G.J.S. 97-3, G.J.S. 98-43); both are derived from ascospores of Hypocrea specimens collected in, respectively, Guadeloupe and Puerto Rico. As can be seen from the PCA (Fig. 4), these two strains are closely similar in phenotype and also genotype (Fig. 2). The isolate DIS 320c, T. caribbaeum var. aequatoriale, forms a highly supported clade with the other two isolates but is phenotypically and apparently biologically distinct. It was isolated as an endophyte from stems of Th. gileri in Ecuador. The considerable differences in phenotype, biogeography and habit, despite its phylogenetic proximity to the ascospore isolates, lead us to recognize the endophyte as a variety, var. aequatoriale. The apparent close relationship between the two varieties is possibly an artifact of sampling; additional sampling could support their separation at the species rank.
At least three species occur in New Zealand and Australia, viz. T. dorotheae, T. dingleyae and T. austrokoningii. Trichoderma dingleyae is the slowest-growing species in the present study; its temperature optimum is 20–25 °C and the colony radius is 5 mm after 72 h at 30 °C. The first two species were collected in Nothofagus forests whereas the third, T. austrokoningii, was found in the tropical Queensland coast and in Nothofagus forests of New Zealand. Hypocrea vinosa Cooke was described from New Zealand (Cooke 1879) and is reported often in the literature, or on the World Wide Web, from diverse geographic regions (e.g. Brazil, Bresadola 1896; Japan, Komatsu & Hashioka 1966; New Guinea, Doi 1971). Ascospores in the type specimen of H. vinosa (K!) are unusually large (distal part-ascospores 5.1–6.7 x 5.0–5.5 µm; proximal part-ascospores 5.7–7.2 x 4.6–5.3 µm), suggesting that most or all of the reports of this species outside of New Zealand are based on misidentifications. CBS 247.63, H. austrokoningii, is derived from a Hypocrea specimen received from New Zealand (J.M. Dingley No. 3, Auckland, Te Aroha). However, we cannot locate that specimen in CBS or PDD to confirm its identity. We have collected specimens in New Zealand that conform to the type collection of H. vinosa and redescribe the species in another publication (Jaklitsch et al. 2006b).
Unlike most clades, the one that includes T. austrokoningii is geographically diverse, including lineages (Figs 2, 3) from tropical Australia (Queensland, Figs 51–59), temperate New Zealand (Figs 80–88), Russia (Figs 71–79), and a single lineage that includes one collection from the United States (Florida) and one from Taiwan (Figs 89–101). Subtle phenotypic differences characterize each clade (e.g. growth rates, Fig. 102, and ascospore measurements). The phylogenetic and phenotypic diversity of the isolates in this "austrokoningii" clade, which occupies the terminal position of the SKB clade (Fig. 2), suggests that more than one taxon could be involved and that additional sampling would resolve this clade.
Some of the species are represented by one or two strains. We would not normally describe a species based on such a small amount of material because there is no way to estimate intraspecific variability. Nonetheless,
T. taiwanense, based on a single collection from Taiwan (G.J.S. 95-93), and
T. intricatum, based on two collections (G.J.S. 97-88 from Thailand and G.J.S. 96-13 from Puerto Rico), are phylogentically distinct from all other species that we have included. The two strains of
T. intricatum are phenotypically distinct as shown by PCA (
Fig. 4).
Trichoderma intricatum was formerly reported as "
H. cf.
muroiana/
Hypocrea sp. (6)" in Lieckfeldt
et al. (
1998).
Answers to all your Biology Questions
