Our work leads us to recognize several species, many undescribed. In the following we have anticipated the formal taxonomy by adopting those names in order to facilitate the presentation of the results.
PHYLOGENETIC ANALYSES OF SEQUENCE DATA
The position of T. koningii-like species on the Hypocrea/Trichoderma genus phylogeny is shown on the radial tree obtained after the analysis of partial rpb2 sequences (Fig. 1). This complex species takes the terminal position on the sect. Trichoderma branch, which consists of "Pachybasium A" and "Viride Clades" ("Rufa Clade" in Chaverri & Samuels 2004, and Druzhinina et al. 2005). It is interesting to note the relatively short genetic distances within clades and species on this branch. The neighbouring H. voglmayrii, which was recently described from the Austrian Alps (Jaklitsch et al. 2006a), or species from "Hypocreanum" and "Lutea Clades" are separated by longer evolutionary distances.
Visual inspection of ITS1 and 2 sequences of strains of the T. koningii complex show a very low degree of variability (max. 6 % of variable sites), corresponding to findings in other studies (Lübeck et al. 2004, Druzhinina et al. 2005). Therefore, this locus was not used in phylogenetic reconstructions. However, we were able to develop a species-specific oligonucleotide barcode from ITS sequences for some of these species. It was integrated into the upgraded version of TrichOKEY previously published by Druzhinina et al. (2005). The program allows the identification of four individual species with T. koningii-like morphology and one group of seven species (for details see below).
The high degree of similarity of teleomorphs and anamorphs within the T. koningii species complex led us to anticipate the higher level of sequence similarity of protein-encoding DNA sequences. Therefore we chose phylogenetic markers with relatively big introns such as (i) the partial sequence of the translation elongation factor 1-alpha (tef1) covering the fourth (large) and fifth (short) introns (Kopchinskiy et al. 2005), (ii) the partial actin (act) and (iii) the partial calmodulin (cal) genes with two introns each. Since tef1 is the most variable locus (>50 % of variable sites) it was selected as a reference phylogenetic marker and, consequently, sequenced for all investigated strains. The cal and act genes were used to apply the Genealogical Concordance Phylogenetic Species Recognition (GCPSR) concept of Taylor et al. (2000) to representatives of the main groups detected by the phylogenetic analysis of tef1.
In order to examine the phylogeny of T. koningii-like strains with respect to their position to the "Viride Clade" we aligned a portion of the tef1 gene for a large number of isolates. First, we attempted to analyse T. koningii-like strains against a background of few representatives of the nearest clades such as T. viride VD and T. viride VB (data not shown). However, the log probability plotted against the number of up to 5 million generations did not reach a stationarity. This indicated a low reliability of the resulting tree. Moreover, trees obtained in different runs with equal priors showed inconsistent topologies and were poorly resolved. Our strategy to solve this obstacle was based on the inclusion of the maximum known variability within the "Viride Clade" in the multiple sequence alignment, including T. viride VB, T. viride VD, H. stilbohypoxyli, T. erinaceus, T. atroviride and several potentially new taxa. The repeated consecutive use of intermediate phylogenetic analyses, rearrangements of sequences in MSA and realignments, particularly of the highly variable forth (large) intron of tef1, made it possible to produce the most correct final MSA file (available at www.isth.info/phylogeny/koningii). In this file, sequences of the T. koningii and T. viride complexes were aligned to representatives of T. asperellum and T. hamatum as members of the next neighbouring phylogenetic clade. As expected, likelihood estimations reached stationarity over generations, indicating reproducibility of the MCMC analyses. Fig. 2 represents a radial Bayesian phylogenetic tree obtained after 5 million chain generations.
Analyses of cal and act sequences did not produce problems during repeated MCMC runs; results are shown in Fig. 3.
All three trees show clear separation with high statistical support between fungi of the "Viride Clade"and species from "Pachybasium A" (Kullnig-Gradinger et al. 2002), of which the latter was represented by T. asperellum and T. hamatum. The monophyletic origin of the entire "Viride Clade" was confirmed by phylogenetic analyses of all three loci. As may be seen in the tef1 radial tree (Fig. 2), the majority of strains with a T. koningii morphology appear on a single proliferating branch, which is well-separated from other large species aggregates such as T. viride VB, T. viride VD, T. atroviride, H. stilbohypoxyli and T. erinaceus. The same tree topology is supported by both cal and act trees. This lineage was named "Large Koningii Branch"(LKB) (Figs 2, 3).
In the tef1 tree, the most basal position of the LKB is occupied by the highly supported multifurcating clade of T. koningiopsis. Trichoderma koningiopsis presents a genetically variable species because only few internal nodes within that species are well-supported and are of numerous long intraspecific genetic distances. Although strains of T. koningiopsis have the same position on the LKB on the trees of Fig. 3, its identity as a distinct monophyletic clade is not statistically supported in analyses of the cal and act genes. These findings could indicate the presence of a relatively intensive recombination process due to sexual reproduction, despite the fact that the vast majority of the several strains of T. koningiopsis studied were derived directly from substrata with no known teleomorph.
In the LKB of the tef1 tree, T. ovalisporum and T. koningii s. str. possess two equally supported clades. The insignificant support of nodes where both of these species diverge from the main stem suggests approximately simultaneous speciation even for both taxa. However, the divergence was clearly allopatric, because T. koningii is common in North America and Europe, while T. ovalisporum is an endophyte from South America. The phylogenetic position of these two species is not contradicted by cal or act trees, although these genes did not always provide significant support. Compared to other species such as T. koningiopsis, T. koningii s. str. appears to be a relatively homogeneous taxon represented by strains from North America and Europe. Trichoderma koningii strains have almost identical sequences in the hypervariable large intron of tef1. Only one strain (DAOM 167073 from Québec) appeared distinct from numerous other strains, which had very similar or identical sequences, irrespective of their broad geographic distribution.
The upper part of the LKB on all three trees has a stepped structure with well-supported internal nodes (except the cal tree). Based on the concordance between the three loci trees, it consists of at least six phylogenetic species. In general, their phylogeny may be attributed to allopatric speciation because T. taiwanense is Asian, both T. dorotheae and T. dingleyae are isolated, known only from Australia and New Zealand, while strains of T. caribbaeum var. aequatoriale have a South American origin. The terminal position is occupied by T. petersenii. It consists of strictly North American and European clades. Trichoderma intricatum, which is located basal to the species listed above, may be an exception because it is represented by one Asian and one Caribbean strain.
Results of phylogenetic analysis show that additional species having T. koningii-like morphology have evolved independently from the taxa of the LKB. The majority of these species appear on the "Small Koningii Branch" (SKB), which is segregated from the LKB by taxa of the "Viride Clade" (Figs 2, 3). The first species on the SKB is T. rogersonii, which is represented by mainly North American and a few European strains. The terminal part of the SKB (tef1 tree) consists of a number of long lineages that lead to geographically separated strains. The divergence among these strains may be explained by allopatric speciation. Three of these six strains originated from Australia and New Zealand, one from Europe, one from the U.S.A., and one from Taiwan. Based on both tef1 and cal loci, Taiwan and North American strains form the most terminal well-supported clade, although on the act tree this clade also includes a European strain. Thus, there is no concordance between topologies of the act tree and trees inferred from sequences of the other two loci. This finding makes it difficult to draw conclusions about phylogenetic species on the terminal part of the SKB.
The third lineage that is characterized by the koningii morphology is represented by the single species T. stilbohypoxyli.
For summary of continuous characters see Table 3.
A total of eighty-six strains were studied. Typical of Trichoderma, very little aerial mycelium forms on CMD or SNA, and mycelial production on PDA is typically lush. There is variation among the strains of individual clades as to whether conidiophores form in aerial mycelium or in complex cottony pustules on CMD as well as in relative amounts of conidial production. Conidial production on CMD and SNA tends to occur at the margin of the colony. Discrete conidial pustules sometimes form on CMD and SNA, but on PDA pustules are not formed, rather, conidia form in dense, effused areas. On CMD and SNA pustules are at most 1.5 mm diam and usually smaller, hemispherical, uniformly cottony. Entirely fertile, somewhat plumose conidiophores can often be seen within pustules (e.g. Figs 176, 197). Usually projecting sterile hairs or conidiophores that are only fertile at the apex are absent, but occasionally long, apically fertile conidiophores are seen in T. austrokoningii (Figs 55, 97), T. dingleyae (Fig. 138), and T. koningiopsis (Fig. 212). Conidiophores also form in the aerial mycelium. One isolate of Trichoderma dorotheae (G.J.S. 99-202) formed hemispherical pustules in addition to conidiophores in the aerial mycelium. Conidiophores in the pustules were not easily discerned (Figs 156, 161). In older cultures of this species phialides appeared to proliferate percurrently to form a second phialide (Fig. 158). The newly formed phialides were often abruptly swollen in the middle. This aspect is also seen in Eidamia viridescens A.S. Horne & H.S. Williamson (1923), which is T. viride VD of recent publications (e.g. Lieckfeldt et al. 1999, Dodd et al. 2003, Holmes et al. 2004) and distinct from "true"T. viride (VB in Figs 2, 3). Pustules generally are compact, formed of intertwined hyphae that tend to branch dichotomously near the surface and to produce short branches that sometimes act as phialides, or for the cells near the surface of the pustules to swell and produce two or more short cells. In addition, verticillium-like conidiophores arise from near the surface of the pustules. Conidia are dark green (27E–F7–8). There is variation in the time and temperature at which conidia appear. Most of the isolates of T. dingleyae and T. dorotheae lost their ability to produce conidia after storage on cornmeal agar slants at ca. 8 °C.
None of the isolates produces sterile hairs, although, as noted above, occasionally long conidiophores that are fertile only at the tip form in some cultures of some strains, as is described for Trichoderma sect. Pachybasium (Bissett 1991b). Often individual branched conidiophores can be seen within pustules when viewed with the stereo microscope. Conidiophores reaching the surface of pustules in members of all clades form a discernable major axis, from which primary lateral branches arise. Primary branches arise at or near 90° with respect to the main axis often singly but also often they arise in pairs or three at a node, with the members at a single node equal in length and progressively longer with distance from the tip of the main axis. Primary (1°) branches rebranch to form secondary (2°) branches. 2° branches follow the same pattern of branching as the 1° branches with longer side branches closer to the main axis and short branches more distal. Phialides arise singly, directly from the main axis near its tip and the 1° branches; they also terminate 1° and 2° branches in whorls of 3 or 4. 1° and 2° branches, respectively, of conidiophores reaching the surface of the pustule tend to be widely spaced from each other. Branches arising from conidiophores found in the interior of the pustule tend to be crowded, with short internodal distances, and phialides tend to be held in dense heads of several.
Conidiophores of T. taiwanense were unusual in often being conspicuously enlarged and verrucose at the base (Figs 290–293).
Phialides were nearly cylindrical, only slightly swollen in the middle, when formed on widely spaced branches and shorter and conspicuously swollen in the middle when crowded. Often phialides were densely clustered with a very short internode between phialides (e.g. T. koningii Fig. 201, T. koningiopsis Fig. 214), we have termed these dense clusters pseudowhorls. Within any culture there can be considerable variation in the size and shape of phialides but there was no difference among the clades in the degree of variation in any of the continuous attributes of the phialides, the mean variation of phialide length for all collections being 6.1 ± 1.8 µm. The mean continuous measurements for phialides in the 86 strains studied was 7.9 ± 1.0 µm long, 3.0 ± 0.3 µm at the widest point, 2.0 ± 0.2 µm at the base, L/W = 2.7 ± 0.6. The cell from which phialides arise was 2.8 ± 0.4 µm and the ratio of the phialide length to the cell from which it arose was 2.8 ± 0.6. Mean phialide length in most collections ranged 7.5–8.5 µm. In two species, T. dingleyae and T. caribbaeum var. aequatoriale, mean phialide length was 9.5–10 µm and in T. intricatum and T. stilbohypoxyli mean phialide length was ca. 7.0 µm.
Conidia of all collections included in this study were oblong to ellipsoidal or ovoidal and smooth; the mean length was 4.0 µm, the mean width was 2.7 µm, and the mean L/W was 1.4. The mean length of conidia in most collections ranged 3.8–4.0 µm; conidia in collections of T. dingleyae and T. koningii were longest, 4.1–4.3 µm, while the mean length of conidia of T. austrokoningii, T. ovalisporum, T. intricatum and T. stilbohypoxyli ranged 3.4–3.5 µm. The mean width of conidia of most collections ranged 2.7–3.0; conidia of T. dingleyae, T. dorotheae, T. intricatum and T. ovalisporum were somewhat wider, the mean ranging 3.0–3.2 µm, while conidia of collections of T. koningii, T. koningiopsis and T. stilbohypoxyli were somewhat narrower, ranging 2.6–2.7 µm. The mean length/width ratio of most collections ranged 1.4–1.5; the mean L/W ratio of conidia in collections of T. koningii was 1.6 and in T. ovalisporum and T. intricatum, which have broadly ellipsoidal to ovoidal conidia, the mean L/W was ca. 1.2.
Chlamydospores are produced sporadically by members of the various species and of the clades, only the two collections of T. intricatum failed to produce chlamydospores. Chlamydospores are typical of Trichoderma, being terminal or intercalary within hyphae, and globose to subglobose.
Most of the strains that we studied were derived from ascospores of Hypocrea specimens. The most notable exception was T. koningiopsis, a common tropical species that was most often encountered as direct isolations from substrata, including as an endophyte from trunks of trees of Theobroma cacao and Th. gileri, and only a few isolates were derived from ascospores. Trichoderma koningii s. str. was most often directly isolated from substrata but three cultures were derived from ascospores of Hypocrea collections made in the United States and one from the Netherlands. Trichoderma ovalisporum is known only from four isolations, all from natural substrata. Trichoderma caribbaeum and its variety aequatoriale (DIS 320c) are represented by three strains; of these, two were isolated from ascospores of Hypocrea specimens collected in, respectively, Guadeloupe (G.J.S. 97-3) and Puerto Rico (G.J.S. 98-43), where they were growing on fructifications of black ascomycetes; the variety (DIS 320c) was isolated from the trunk of a live tree of Theobroma gileri in Ecuador and may be an endophyte. Despite strong phylogenetic similarity between DIS 320c, on one hand, and G.J.S. 97-3/G.J.S. 98-43 on the other, DIS 320c is phenotypically quite different from the other two and we regard it as a variety of T. caribbaeum.
Stromata (when dry, Figs 24–35, 36–50) were typically 6C–D8, brownish orange to light brown, but in T. petersenii stromata are darker, 7–8E–F8, reddish brown; stromata were typically pulvinate and broadly attached or at most slightly free at the margins. Perithecial elevations, or mounds, were not visible; the stroma surface was plane or wrinkled. Ostiolar openings were not visible, at least in dry specimens, or were barely visible as viscid circular areolae or dots on the stroma surface. There was no reaction to KOH in any tissue. When young, stromata were semi-effused, light brown or tan and villose; the developing stroma retained the villose aspect, which eventually was lost. The villose aspect is the result of short hairs that arise from the cells of the surface of the stroma; these are 5–10 µm long, 2.5–3.5 µm wide, septate, unbranched, often spinulose. The stroma surface, seen in face view, appeared mottled with unevenly deposited brown pigment in the cell walls. The cells at the surface of the stroma, when seen in face view, were angular, 3–7 µm diam, with walls slightly thickened. The stroma comprised three anatomically distinct regions. The surface region was 15–25(–35) µm thick and pigmented, in section appearing yellow when mounted in lactic acid. Cells of the surface region were angular, 2.5–5 µm diam, with slightly thickened walls. The tissue immediately below the stroma surface consisted of compact to loosely disposed hyphae. The tissue below the perithecia was pseudoparenchymatous, the cells measured 5–10(–15) x 3–7(–10) µm; their walls were slightly thickened or not visibly thickened; cells were oriented perpedicular to the surface of the substratum. The stromata of T. taiwanense (G.J.S. 95-93) are atypical in the group because they are luteous, lack hairs and have conspicuous ostiola; however this specimen is old and possibly has lost the traits that are typical of this group.
Perithecia were elliptic in section, 160–280 µm tall, 100–185 µm wide, ostiolar canal 53–90 µm long, cells of the perithecial apex not sharply differentiated from the cells of the surrounding stroma.
Asci were cylindrical, 60–70 x 4–5.7 µm, completely filled with ascospores; there was a thickening at the tip of each ascus. Ascospores were bicellular; they disarticulated at the septum into two part-ascospores early in development.
There was little variation in ascospore morphology or measurements among the 46 teleomorph collections that were studied. Differences are noted as follows: Part-ascospores were hyaline, spinulose, dimorphic; distal parts ranging (3.0–)3.5–4.0(–4.5) x (2.0–)3.1–3.8(–4.0) µm; proximal parts ranging (2.7–)3.7–4.5(–5.0) x (2.2–)2.7–3.2(–3.7) µm. The means of distal part-ascospores of most collections ranged 3.6–4.0 x 3.2–3.7 µm and of proximal part-ascospores 4.0–4.6 x 2.8–3.2 µm. The ascospores of H. intricata were somewhat smaller overall than in the other species. The distal part-ascospores of H. intricata were somewhat shorter and narrower than in the other species, ca. 3.3 x 3.1 µm. The proximal part-ascospores of H. intricata were also smaller, falling in the lower end of the range of spore dimensions overall. The distal part-ascospores of H. koningii were somewhat longer than most species (mean 4.1 µm) while those of H. petersenii were somewhat wider than in most species (mean = ca. 3.3 µm).
PHENOTYPE: COLONY MORPHOLOGY AND GROWTH RATE
Colony morphology is described from PDA at 25 and 30 °C in light or darkness after 72–96 h. Colony morphology is more or less consistent within a species. The cultures illustrated in Figs 6–14, 15–23 are representative of the respective species. There is a tendency for conidia to form in concentric rings that are more or less pronounced; this is especially clear in T. petersenii. With the exception of T. dingleyae and T. dorotheae, conidia tended to form in abundance and to be dark green; conidial production in these two species is poor.
A summary of growth rate curves is shown in Fig. 5. In general, these are rather slow-growing species of Trichoderma. The colony radius is typically less than 50 mm and none reaching a colony radius of 70 mm on PDA, and most less than 40 mm on SNA, when grown for 72 h at optimum temperature of 25–30 °C in darkness (Fig. 5). The temperature optimum for most species is 25–30 °C; the temperature optimum for DIS 203c (T. ovalisporum) and T. dingleyae is lower, 20–25 °C. There is little (radius typically 35 °C for any of the species. On SNA, only T. caribbaeum, T. koningiopsis and T. ovalisporum reach a radius of 40 mm at the optimum temperature; the rest of the species reach a radius of 20–30 mm. On PDA after 72 h darkness, the mean colony radius of most species was T. caribbaeum var. caribbaeum and T. ovalisporum was 55–60 mm and the mean radius of T. koningiopsis was 60–65 mm. Most species grow faster at 30 °C than at 20 °C. However, T. dingleyae grew very poorly at 30 °C on both PDA and SNA (radius 72 h in darkness), whereas at 20 °C colony radius was ca. 20–25 mm, and T. caribbaeum var. aequatoriale grew considerably more slowly at 30 °C (10 mm) than at 20 °C (35 mm).
Green conidia were first observed in PDA cultures of most species within 48–72 h on PDA at 25–30 °C, although individual isolates of a species varied in this regard. In T. caribbaeum var. caribbaeum, T. ovalisporum and T. dorotheae conidia of most isolates were first observed at 20 °C within 48–72 h. All but a few isolates of all species, except T. caribbaeum var. aequatoriale and T. dingleyae, formed conidia within 96 h. There was a correlation between species and temperature and time of the first appearance of green conidia, with the exception of T. austrokoningii, in which conidia overall appeared after 72 h; but there was considerable variation among the isolates as to the time of first appearance of conidia, which ranged from 48–96 h. In T. dorotheae, T. ovalisporum, and T. petersenii first green conidia were seen beginning after 48 h at 20 °C.
No distinctive odour was detected in any cultures, or rarely a coconut odour in T. koningii.
With few exceptions, a biogeographic bias was seen in the respective clades (Fig. 2, Table 3). The many reports in the literature of wide distribution not withstanding, distribution of T. koningii is limited to eastern North America and Europe. Trichoderma koningiopsis is a common and cosmopolitan species, but it is more common at tropical than at temperate latitudes.
Most of our isolates originated in the American tropics, but the species occurs in Canada (Ontario) and Germany, and its teleomorph has been found in the U.S.A. (Kentucky). It was also found in the rhizosphere of Coffea arabica from the main coffee-growing area in Ethiopia, where sampling was done from elevations of 1300–2000 m (T. Belayneh, pers. comm.). Trichoderma stilbohypoxyli was also revealed in this work to be a common tropical species, being widespread in tropical America and found in one location in Ghana, but it also occurs in the U.K.
Trichoderma petersenii and T. rogersonii are common and sympatric in eastern North America and central Europe; we have only seen T. rogersonii as isolations from ascospores but we have a single soil isolate of T. petersenii; stromata of T. petersenii have been collected also in Costa Rica. Trichoderma intricatum is known only from two ascospore-derived cultures that originate, respectively, in Puerto Rico and Thailand.
The most problematic clade from the point of biogeography is the clade comprising strains identified here as T. austrokoningii. This clade includes six isolates with unclear phylogenetic position because topologies of corresponding branches are nonconcordant among three loci. Most divergent are two strains, respectively, from Florida and Taiwan. The basal lineage on the tef1 tree comprises two strains (only one shown in Fig. 2) from the South Island of New Zealand. There is a single lineage/isolate from Russia and two closely related collections from tropical Australia. The sequence divergence in this clade suggests that additional sampling would resolve it into two or more species. Although collections in the two Australasian clades are physically relatively close to each other, their actual locations are climatically very different, tropical in the case of the two Australian cultures and south-temperate in the case of the collections from New Zealand.
We cannot say that any geographic region is more diverse than any other as regards the genetic diversity represented in Fig. 2. As was noted above, T. petersenii and T. rogersonii, T. koningii and T. koningiopsis are sympatric in eastern North America. Three species are found in Australia and New Zealand, viz. T. dingleyae, T. dorotheae and T. austrokoningii, although the latter species was found on the northern, tropical Queensland coast of Australia, whereas the other two were collected in south-temperate Nothofagus forests of Australia (Victoria) and New Zealand (S. Island).
The Hypocrea specimens from which most of the cultures were derived were found either directly on ascomata of, often, members of the Xylariaceae or on indeterminate black fungi on rotting decorticated wood or bark of rotting trees. In only a few cases was a fungal substratum not seen. The isolates taken directly from the substratum were taken from soil, less frequently from fallen leaves and mushroom casing. Several isolates were recovered as endophytes from the sapwood of stems of Theobroma species or, in one case (T. ovalisporum, DIS 70a), a liana. The Trichoderma endophytes of Theobroma gileri were reported by Evans et al. (2003). Five isolates of T. koningiopsis from Ecuador, represented in the cladogram by G.J.S. 01-07, were isolated directly from pods of Theobroma cacao infected with Moniliophthora roreri that had been placed on cacao leaf litter in a search for parasites of the Moniliophthora. Perhaps most interesting of the Trichoderma endophytes of woody plants was T. stilbohypoxyli. We have found (Samuels, unpubl.) that Trichoderma stem endophytes tend strongly to be specific to host genus and to biogeography, but T. stilbohypoxyli was isolated as an endophyte from trunks of ancient Fagus sylvatica in the United Kingdom and Theobroma species in Ecuador and Brazil.
RESULTS OF PRINCIPAL COMPONENTS ANALYSIS (PCA)
PCA was performed to determine the correlation between phenotype traits and clades; the phylogenetic clades were used as the grouping factor. Only characters of the anamorph and of colony morphology and growth rates and geographic distribution were used in the analysis because they were common to all isolates. Characters of the teleomorph were not utilized in PCA because those characters were not common to all strains. Analysis of the teleomorph characters did not resolve groups. In the PCA of the geographic and phenotype characters listed in Table 3, 51 % of the variation is accounted for by the first three axes. While the results of the Eigen analysis (Table 2) do not indicate a strong fit of the data to the model, a scatter plot of the Eigenvalues reveals that isolates of the same clade/species tend to group together (Fig. 4). The two geographically distinct isolates of T. intricatum were separated because of the slow growth of G.J.S. 97-88. Slow growth also pulled DIS 94c from the rest of the isolates of T. koningiopsis, and CBS 979.70 from the rest of T. koningii. The four Puerto Rican gatherings of T. stilbohypoxyli clustered together and distant from the other 8 cultures; this may be because the Puerto Rican strains have, on average, slightly longer and wider conidia than the others. The phylogenetic diversity of T. koningiopsis and, especially, T. austrokoningii is reflected in their wide dispersion on the scatter plot of eigenvalues.
IDENTIFICATION OF SPECIES
Three methods of Hypocrea/Trichoderma species identification based on the analysis of DNA sequences have been developed. Most recently, an automated identification system using oligonucleotide DNA barcodes of ITS1 and 2 sequences was developed. If it is already available for the group under investigation, a barcode is the easiest method to obtain an absolute result. The second possibility is to perform a similarity search (BLAST) against a pool of voucher sequences. This method is very useful because a search can be made using multiple loci; however, results from this technique are unavoidably uncertain, because the user must subjectively weigh every mismatch in the resulting sequence alignment. Moreover, since gene evolution does not always reflect the speciation process, it is highly recommended to obtain a concordant result of several unlinked loci. The third method of molecular species identification is the most reliable one but, at the same time, the most laborious because it implies phylogenetic analyses and the application of the Gene Concordance Phylogenetic Species Recognition (GCPSR) concept of Taylor et al. (2000). A detailed description of the application of each of these methods to the T. koningii aggregate species is given below. Molecular identification is available via a dedicated online "T. koningii morphological species project", which is located at www.isth.info/phylogeny/koningii.
Using ITS1 and 2 and the oligonucleotide barcode program TrichOKEY v. 1.1
The first version of DNA oligonucleotide barcode integrated in TrichOKEY v. 1.0 (www.isth.info, Druzhinina et al. 2005) is able to recognize Trichoderma sect. Trichoderma and all species from the "Viride Clade" that were known prior to this study. Thus, the barcode distinguished the T. koningii aggregate species as a triplet of T. koningii/T. ovalisporum/H. muroiana. We have investigated the inter- and intraspecific variability of ITS1 and 2 sequences from the complex based on the present larger sample size. Unique species-specific oligonucleotide hallmarks for T. petersenii and T. rogersonii and T. koningii s. str. have been discovered. Because all three species are known from many specimens, all of which were considered in the development of the barcode, the resulting identification is reliable ("standard" in TrichOKEY v.1.1). In addition, a characteristic DNA signature that is common to both isolates of T. intricatum is incorporated in the program. Although, due to the low number of available isolates, the barcode identification is of low reliability and needs to be confirmed by other methods of sequence analysis. Other species from the T. koningii aggregate species such as T. koningiopsis, T. caribbaeum, T. ovalisporum, T. dingleyae, T. dorotheae, T. taiwanense and T. austrokoningii are not distinguishable based on ITS1 and 2 sequences, at least based on the observed diversity. Therefore, they will be identified as "T. koningiopsis or 6 rare species with T. koningii morphology" because T. koningiopsis is the most abundant and cosmopolitan species, known from more isolates than the total number of specimens from other species with ITS1 and 2 haplotype identical to it. Help involving biogeography is provided for distinguishing these species. Thus, the updated version of the ITS1 and 2 barcode (TrichOKEY v. 1.1) is able to distinguish all sympatric species from the complex of the T. koningii aggregate species.
Using tef1 and sequence similarity search program TrichoBLAST
The TrichoBLAST tool for Hypocrea/Trichoderma sequence identification installed on www.isth.info (Kopchinskiy et al. 2005) determines the sequence from the database to which the query sequence is most similar. The main TrichoBLAST database consists of sequences of five phylogenetic markers (two introns and one exon of tef1, partial exon of rpb2 and ITS1 and 2). With respect to the present group, the first version of this database included only two sequences of each tef1 intron from two strains of T. koningii s. str. and one strain of T. ovalisporum. The remaining biodiversity was not considered. In order to facilitate the identification of species from the T. koningii morphological species, we have extracted both most variable tef1 phylogenetic markers (forth large and fifth short introns) from the type sequences of each new species and inserted them in the main database, which is named "Nucleotide DB of Phylogenetic Markers" (http://isth.info/tools/blast/blast.php). Because of the high degree of intraspecific variability, we assume some difficulty in species identification based on tef1 introns and similarity search, because the user would need to charge the weight of multiple mismatches on the pairwise alignment. Therefore, in order to minimize the possible dissimilarity between the query and subject sequences in a blast result, we have composed a separate database of both tef1 introns for all available sequences from the T. koningii morphological species. This database is named "koningii tef1" and it is directly available from the T. koningii project page (www.isth.info/phylogeny/koningii) or as a selectable database in the main TrichoBLAST. Thus, the user has a possibility to perform the primary round of species identification using the default, main, database. A positive result ("koningii-positive") leads to the possibility of searching for the most similar haplotype among highly homologous sequences and making allowance in the final identification according to it. The conclusion about species identification may be drawn if the query sequence is significantly more similar (significantly higher bit score) to sequence(s) of one species compared to the similarity to others and if all precautions explained in Kopchinskiy et al. (2005) are taken into account.
Using multilocus phylogenetic analysis
As has been shown in this study, the "Viride Clade" is an extremely species-rich group. Therefore, we anticipate the further discovery of new taxa with the T. koningii morphology. In this case, both methods of molecular species identification would provide uncertain results, e.g. the ITS1 and 2 based barcode will lead to an identification at the level of clade and/or "T. koningiopsis and six rare species" (see above), while the result of similarity search will show the same relation to the group of species instead of only a single taxon. Such a situation may be resolved only by the use of phylogenetic analysis based on several unlinked phylogenetic markers. In order to facilitate the task, we have included type sequences of newly recognized species in the multiloci database of phylogenetic markers (http://www.isth.info/tools/blast/show_all_seq.php). This database is especially designed to assist in retrieving of type sequences for the subsequent phylogenetic analyses. In addition, as has been mentioned above, the T. koningii project page also contains a table listing all the ITS1 and 2, and both tef1 sequences of species from the T. koningii aggregate species, and the geographic origin of the corresponding strain is given. Thus, the final species identification or the detection of the new species may be done based on the phylogenetic analysis, with phylogenetic markers retrieved from www.isth.info/phylogeny/koningii.
Using phenotypic characters
Geography and reproductive isolation have played a large part in our species concept. PCA revealed phenotype-based groups that combine with GCPSR in species delimitation. While there is significant homoplasy in the phenotypic characters, the species that we recognize in most cases are not sympatric. Although the species that we recognize in this work are characterized to some extent by phenotypic characters, we recognize that often the characters are at best subtle and difficult to observe. It will be difficult to recognize a species if it is out of the currently known geographic range. Nevertheless we have provided a key for species identification based on phenetic characters, which should resolve doubts that may derive from sequence similarities.