Introduction

The coelomycetous anamorphic genus Septoria represents more than 2000 described fungal species including important pathogens of crops and trees. Based on morphological and molecular studies, Septoria spp. appear unequivocally linked to the teleomorphic genus Mycosphaerella (Dothideales) (Verkley and Priest 2000; Verkley et al. 2004). More than 30 anamorphic genera have been associated with Mycosphaerella and recent molecular studies have underlined the difficulty in using morphological characters for generic and specific definition among these anamorphs (Crous et al. 2000, 2001; Verkley and Priest 2000; Verkley et al. 2004). Host specificities also have appeared inadequate to delimit species and reconstruct phylogenetic relationships among anamorphs within Mycospaherella. This is in disagreement with the co-evolutionary concept usually expected between parasites and host species (Feau et al. 2006; Crous et al. 2004).

Under these phylogenetic investigations, the small (SSU) and large (LSU) subunits of the nuclear ribosomal DNA (rDNA) gene have been sequenced in a wide variety of fungal taxa, but to date very little is known about nuclear rDNA size variation among species within the Mycosphaerella genus. In a recent survey of the 3′ end of the SSU-rDNA region in Septoria spp., we noted length polymorphisms in four Septoria species (Feau et al. 2006). Sequence alignments revealed that 484–500 nucleotides (nt) insertions were all located at the same rDNA position corresponding to base 1506 of the SSU-rDNA gene of Escherichia coli (Gutell 1993). Since sporadic insertions at this position in the nuclear rDNA correspond mainly to group I introns, we hypothesized that the four insertions encountered in Septoria spp. belonged to this class of introns. By convention, group I introns are characterized by the possession of a set of conserved sequence elements designated P1 and P3–P10. P4–P6 and P3–P9 helical domains constitute the catalytic core elements and P1 and P10 helices the substrate domain that contains the 5′ and 3′ splice sites (Burke et al. 1987; Cech 1988; Michel and Westhof 1990). Based on both conserved nucleotide sequences and secondary structure characteristics, group I introns are classified into five major groups (IA to IE), which can be subdivided further into subgroups (e.g., IA1, IC1) according to the presence/absence of peripheral paired elements (Michel and Westhof 1990; Suh et al. 1999).

The widespread and often sporadic phylogenetic distribution of group I introns may be explained through a dynamic equilibrium between gains, due to the mobility of these genetic elements capable of horizontal transfer within and across lineages, and losses due to drift and/or selection (Dujon 1989; Nikoh and Fukatsu 2001; Haugen et al. 2005). A cyclical model of invasion through homing, degeneration, and loss, followed by reinvasion has been proposed for group I intron-endonuclease elements and studied in detail (Goddard and Burt 1999). Alternatively, ribozyme-mediated intron transfer into target RNAs (the “reverse splicing” reaction [Woodson and Cech 1989; Roman and Woodson 1998]) provides another plausible model, although occurrence of this pathway has not been demonstrated in genetic crosses (Bhattacharya et al. 2005; Haugen et al. 2005). As a consequence of horizontal transfer events, the phylogenetic signal of such genetic elements may be misleading (Dujon 1989; Hibbet 1996; Nikoh and Fukatsu 2001; Goddard and Burt 1999) and the utility of group I introns of rDNA as phylogenetic characters rests on the assumption that sequence evolution of introns significantly exceeds the rate of horizontal transmission (Hibbet 1996).

In this study, the range and extent of group I introns in the nuclear SSU-rDNA gene have been examined among anamorphic fungal species linked to the teleomorphic genus Mycosphaerella. We report the structure organization of three group I intron lineages, each inserted within specific target sequences and after specific positions of the nuclear SSU-rDNA. To investigate in detail the potential horizontal transmission among species, comparative phylogenetic analyses were undertaken between group I introns and intron-host sequence phylogenies (e.g., the strikingly linked internal transcribed spacers 1 and 2 [ITS] sequences) using the maximum parsimony (MP) and the maximum likelihood (ML) phylogenetic reconstruction methods. Furthermore, in order to assess the statistical significance of these comparisons, both reconstruction methods were coupled with the Templeton (WSR-test; Templeton 1983) and the Shimodaira-Hasegawa (SH-test; Shimodaira and Hasegawa 1999) statistical tests of individual branch points, respectively.

Materials and Methods

Sampling and PCR Amplifications

Thirty-six of the 37 ingroup species described in a previous phylogenetic study of Septoria and related anamorphic species (Feau et al. 2006) were assayed for presence-absence of large sequence insertions in the SSU-rDNA. For the present study, only one Botryosphaeria quercuum outgroup isolate was examined. Culture conditions and total DNA extraction protocols were the same as those described by Feau et al. (2006). The nuclear SSU-rDNA repeats were amplified and sequenced with the primer combinations listed in Table 1 and described in Fig. 1 using previously published protocols (Feau et al. 2006).

Table 1. List of the nuclear SSU-rDNA group I intron sequences found in 14 morphospecies of Mycosphaerella a
Full size table
Fig. 1.
figure 1

Diagrammatic representation of primers used in this study and relative locations of the group I introns in the nuclear SSU-rDNA gene.

Full size image

Intron Secondary Structure Modeling and Sequence Alignments

The location of each intron insertion site was determined by comparison with the orthologous nuclear SSU-rDNA gene in Escherichia coli (accession no. J01695; Gutell 1993). Secondary structure models were predicted following the conventions for group I introns defined by Burke et al. (1987) and according to the models proposed by Cech (1988) and Michel and Westhof (1990). The intron structural elements (P1–P9 stem-loops) were individually identified by comparison with available group I intron sequences from the Comparative RNA web site (CRW at http://www.rna.icmb.utexas.edu/; Cannone et al. 2002) and then folded using the mfold web server at http://www.bioinfo.rpi.edu/applications/mfold/old/rna/form1.cgi (Zuker 2003). When appropriate, group I introns are denominated according to the nomenclature system defined by Johansen and Haugen (2001), (e.g., (1) three-letter abbreviation of host scientific name, (2) one-letter abbreviation of host gene, and (3) insertion site in the rDNA.

Phylogenetic Analyses

Phylogenies were reconstructed among each group I intron dataset obtained at one specific SSU-rDNA insertion site and the equivalent host sequence dataset. Given the low variability of SSU-rDNA intron host sequences observed at such closer taxonomic levels, we opted for comparisons with the polymorphic ITS regions of the nuclear rDNA gene. Group I intron sequences were manually aligned through juxtaposition of the secondary structural elements using BioEdit (Hall 1999). ITS sequences downloaded from GenBank (http://www.ncbi.nlm.nih.gov/) were aligned using the default parameters implemented in T-Coffee (Notredame et al. 2000) and then adjusted by eye with BioEdit (Hall 1999). All nucleotide sites were included in the phylogenetic analyses but gaps treated as missing characters. Heuristic maximum parsimony (MP) searches were performed with 1000 random addition of sequences (RAS), tree bisection-reconnection (TBR) swapping, MulTrees turned on, and “Max Trees” set to 1000. Non parametric bootstrap values were estimated with 1000 pseudo-replicates with settings as above. For maximum likelihood (ML) analyses, the best-fit model of DNA substitution and the parameter estimates used for tree construction were chosen by performing hierarchical likelihood-ratio tests in Modeltest (v. 3.06; Posada and Crandall 1998). Heuristic ML searches were performed with 100 replicates of random sequence addition and TBR branch swapping. Branch support was evaluated using 500 bootstrap (BS) replicates and 10 RAS per pseudo-replicate. All MP and ML analyses were conducted with PAUP ver. 4.0b10 (Swofford 2003).

Testing Phylogenetic Congruence Among Datasets

Topological incongruences between each group I intron phylogeny and the equivalent ITS phylogeny were evaluated by inspecting strength of bootstrap support for individual branches. Each incongruent node with bootstrap value >50% was used as a constraint for the alternative dataset and new MP and ML searches were conducted. The Wilcoxon sign-rank test (WSR-test) (Templeton 1983) was used to compare the number of steps required by each character on the constrained and unconstrained MP analyses. A one-tailed test using single comparisons was applied for correcting the statistical bias due to the comparison of the a posteriori tree (the unconstrained tree, e.g., the optimal tree for the data) with a specified a priori hypothesis (the constrained tree) (Templeton 1983). In addition, the Shimodaira-Hasegawa multiple comparison test (SH-test; Shimodaira and Hasegawa 1999; Goldman et al. 2000) implemented in PAUP, and employing re-estimated log likelihood (RELL) approximation with 1000 replicates, was used to compare ML trees that resulted from the constrained and unconstrained analyses. The set of topologies that was compared included the ML tree from the unconstrained search (a priori tree), the most likely tree from each constrained ML search (a posteriori trees), and a set of unconstrained parsimony trees (up to two steps longer than the best MP tree) of the dataset studied following the recommendations formulated by Goldman et al. (2000) and applied by Hufford et al. (2003).

Results and Discussion

Evolutionary studies of nuclear rDNA group I introns and detailed tree comparisons have already been made across distant intron host lineages and suggested the occurrence of horizontal transfers between distinct host lineages (Hibbet 1996; Nishida et al. 1998; Perotto et al. 2000; Haugen et al. 2005). When studies focused on lower taxonomic levels (intrageneric level), other aspects of the evolutionary dynamics of group I introns often emerged such as vertical transmission and occasional losses (Bhattacharya et al. 1996; Goddard and Burt 1999; Nikoh and Fukatsu 2001) and few evolutionary events such as horizontal transfers were expected (Nikoh and Fukatsu 2001; Simon et al. 2005). Hence, there are relatively few examples in which horizontal transfers of group I introns have been studied and statistically demonstrated among closely related species (Holst-Jensen et al. 1999; Simon et al. 2005). Our study offered here an opportunity to gain a realistic and detailed picture of the occurrence and significance of such events in group I intron evolution since we have considered one of the lowest taxonomic levels found in ascomycetous fungi (the Septoria and Cercospora anamorphic genera are nested within one single teleomorphic genus, i.e., Mycosphaerella) in which three group I intron lineages exhibited a variety of insertion patterns.

Characterization and Variability of Identified Group I Introns

Of the 36 morphospecies surveyed, 14 showed inserts by comparison of the SSU-rDNA sequences (Table 1). Sequence alignments with the insert-less regions revealed three insertion sites named SSU-943, SSU-1199, and SSU-1506, relative to their location with the corresponding nucleotide sites in E. coli rDNA (acc. no. J01695) (Fig. 1). All these insertions exhibited the characteristic features of group I introns, i.e., (1) similar position with other group I introns, (2) paired elements P1–P10, and (3) the last exon base U, immediately upstream of the 5′ intron splice site and the last intron base G, preceding the 3′ intron splice site (Cech 1988; Michel and Westhof 1990). Based on secondary structure predictions, SSU-943 and SSU-1506 introns have been placed in subgroup IC1 (Michel and Westhof 1990; Costa and Michel 1995; Lehnert et al. 1996) (Figs. 2A and B) and SSU-1199 introns in subgroup IE (Machouart-Dubach et al. 2001; Suh et al. 1999) (Figs. 2C and D). Sequence examination of the central core of IE introns revealed that all introns, except Bqu.S1199, belonged either to the subgroups IE1 (Smu.S1199, Sri.S1199, Sal.S1199, Sos.S1199, and Sac.S1199 found in S. aceris mol sp. 1, 2, and 4 and unknown species Ston1) or IE3 (Squ.S1199, Peu.S1199, Pfi.S1199) according to the classification developed by Li and Zhang (2005) (Figs. 2C and D). Here, we showed, conversely to what Li and Zhang (2005) reported, that the SSU-1199 site is not restricted to one unique intron type, suggesting, finally, less restriction of the intron type by this insertion site than expected. Generally, changes observed in the primary structure in the central core regions of an intron require compensatory changes in another part of the RNA structure (Haugen et al. 2005; Holst-Jensen et al. 1999; Li and Zhang 2005). In this study, the compensatory changes observed in the peripheral regions of IE1 introns were identical to those observed by Li and Zhang (2005) for members of the IE3 subgroup (GNRA tetraloop in P9 and branched P2.1 with a large bulge within P2.1a [P13 tertiary interaction]). Therefore, rather than represent two clearly distinct intron lineages, the IE1 and IE3 introns obtained at the SSU-1199 insertion site in this study shared characteristic features common to the different IE introns subgroup described until now. In addition, based on the central core primary sequence, the Bqu.S1199 intron remained unclassified but exhibited the GNRA tetraloop in P9 and the P13 tertiary interaction (Supplementary Fig. S1). This intron chosen for outgroup rooting remains (1) phylogenetically unresolved between both IE1 and IE3 intron lineages (Supplementary Fig. S2) and, thus, (2) structurally unrelated to one of the IE1, IE2, or IE3 subgroups described by Li and Zhang (2005).

Fig. 2.
figure 2

Predicted structures of group I introns. Secondary structures of the 381-nt SSU-943 (A) and 500-nt SSU-1506 (B) IC1 introns from S. musiva isolate m01.01d. The main figures in C and D represent secondary structures of the 381-nt SSU-1199 IE1 intron from S. musiva (isolate m01.01d) and the 411-nt SSU-1199 (D) IE3 intron from P. eumusae (isolate 487). The angled boxes indicate P9–P9.3 structure variants found (C1) in the IE1 intron obtained from S. aceris mol. sp. 2 (isolate CBS655.97) and (D1) in the IE3 intron of S. quercicola isolate CBS456.91. Intron sequences are in uppercase letters and 5′ and 3′ exons in lowercase letters. Motifs putatively involved in tertiary base-pair interactions are italicizied. Base triple interactions according to Guo et al. (2004) are shown as dashed lines.

Full size image

Striking length variation was observed among the P9.2 segments (absent in the Squ.S1199 [Fig. 2, D1] and Bqu.S1199 introns, 26–53 nt for others) and long secondary insertions were detected in the P9.3 loops of some SSU-1199 introns (Fig. 2, C1). These insertions ranged in size from 625 nt in the intron retrieved in S. aceris mol. sp. 2 to 447 nt in S. aceris mol. sp. 1/ and 64 nt in the Bqu.S1199 to 53 nt in the Mfi.S1199 introns. Both long secondary insertions more than 400 nt (S. aceris mol. sp. 1 and 2) appeared too divergent to be easily aligned across their entire length using the T-Coffee software (data not shown). Blastx searches with these long insertions were performed on the GenBank protein database and sequences retrieved were homologous to miscellaneous short sequences and unidentified hypothetical proteins with e-values varying from 0.43 to 8.1. In addition, no ORFs of more than 50 nt were found in the sense or antisense sequences of these insertions. Group I introns are considered as mosaic genetic elements since optional ORFs encoding homing endonuclease gene (HEG) are associated in noncritical regions such as terminal loops (Goddard and Burt 1999; Haugen et al. 2005). For example, endonucleases encoded by nuclear SSU-rDNA introns located in position 1199 have previously been characterized in the fungal pathogen Neonectria galligena and in one ericoid-associated fungal strain (Johansen and Haugen 1999; Perotto et al. 2000). Endonuclease genes that have been characterized in nuclear group I introns appeared exclusively to be members of a distinct family defined by a conserved 29- to 30-amino acid segment including His-Cys motifs (Johansen et al. 1993; Haugen et al. 2004). Insertion of introns can be mediated through homing which results, after recognition and cleavage of the double-stranded insertion site, in transfer of the intron-HEG unit and the flanking exon sequences (coconversion or co-inheritance) (Dujon 1989; Goddard and Burt 1999; Haugen et al. 2004). Coconversion of flanking exon sequences appears to account for preservation of the P1 pairing in order to maintain the splicing ability of the newly located intron. Furthermore, Bell-Pedersen et al. (1989) demonstrated that exon regions of about 50 nt adjacent to the site of intron insertion may be coconverted during intron mobility. We did not, however, obtain sequence evidence for such coconversion phenomena within the ITS1 region neighboring the SSU-1506 intron (located 21 nt just before the ITS1 region; Fig.  1).

Intron Phylogeny and Distribution

Reconstruction of the nuclear SSU-rDNA group I intron phylogeny resulted in (1) a substantial phylogenetic resolution at the group and subgroup levels and (2) intron clustering on the basis of their site of insertion in SSU-rDNA (Supplementary Fig. S2). This result has been reported in a number of previous studies that incorporated a broader sampling of introns (for a review, see Haugen et al. 2005). When mapped on a phylogeny of Septoria and closely related anamorphic species based on ITS, β-tubulin, and mtSSU-rDNA sequences (modified from Feau et al. 2006), the intron distribution was not restricted to a particular lineage but rather widely and sporadically distributed among distinct lineages within the Septoria and Pseudocercospora anamorphic genera (Fig. 3). Presence/absence patterns of the introns were quite complicated, even in well-supported monophyletic groups of closely related Septoria species such as those within the Populus clade (S. musiva, S. populicola, unknown species Ston1 and S. populi) or the S. aceris/M. latebrosa complex defined by Feau et al. (2006). Notably, the 02-92D isolate appeared to lack the SSU-1506 intron (based on ITS5/NS8 primer pair PCR detection), suggesting here intraspecific variation in the intron presence-absence pattern within S. alnifolia. However, concerning the conventional PCR approach used in this study for detecting SSU-rDNA intron, we cannot unambiguously identify taxa as being intron-less. All PCR reactions (35 cycles) yielded a single PCR product, suggesting that gene conversion mechanisms have successfully converted all rDNA copies to either possess or lack introns. However, because the PCR technique preferentially amplifies small DNA fragments, it should be noted that the presence of introns may go undetected if the rDNA copies containing these sequences occur in low numbers. Consequently, intron loss among lineages, even in large well-supported and monophyletic clades, needs to be carefully interpreted or reconsidered.

Fig. 3.
figure 3

Phylogenetic relationships among Septoria spp. and related anamorphic species all linked to the teleomorphic genus Mycosphaerella. This is a strict consensus tree from six MP trees of 3053 steps generated with 2417 aligned nucleotide sites obtained from ITS, β-tubulin, and mtSSU-rDNA (Feau et al. 2006). Bootstrap values ≥50% (500 pseudoreplicates) are presented above each node. The distribution of SSU-943 (■), SSU-1199 (●), and SSU-1506 (▲) introns is indicated. SSU-1199 intron subgroups are indicated in circles with white letters. The asterisk indicates intraspecific polymorphism of the SSU-1506 intron among the S. alnifolia isolates. The occurrence of group I introns in the nuclear SSU-rDNA gene of S. pini-thunbergii remains unknown.

Full size image

Phylogenetic Congruence Between ITS and Group I Introns

Horizontal transfer of group I introns among taxa has been proposed to explain the scattered distribution and presence/absence of introns in related taxa (Haugen et al. 2005; Simon et al. 2005). In order to address directly this question using our intron dataset, comparisons were made between the intron and ITS MP and ML trees for each SSU-rDNA intron position with the assumption that nuclear rDNA regions are strictly linked and vertically inherited. MP analyses conducted on each intron and ITS partition resulted in apparent good levels of resolution with the one tree obtained for each dataset (Table 2). Despite minor discrepancies in bootstrap support sensitivity, both phylogenetic methods (e.g., MP and ML) resulted in topologically congruent trees within each dataset (Fig. 4) and main divergences remained in the SSU-943 tree and the equivalent ITS tree within deep nodes among closely related species or introns. Comparison of the intron trees with their equivalent ITS phylogenies revealed substantial topological differences. Following this initial screen based on bootstrap support values, a combination of MP- and ML-based tests of individual branch points (WSR- and SH-tests, respectively) was used to help determine the statistical significance of the putative incongruent nodes. Both topological differences detected in the SSU-943 comparison met a reciprocal 70% bootstrap support criterion, but only the placement of the P. eumusae and M. musae taxa was significantly rejected by the SSU-943 dataset (WSR, p = 0.04; SH, p = 0.05) and, reciprocally, by the equivalent ITS dataset (WSR, p < 0.001; SH, p = 0.05). The differences in placement of the S. quercicola taxon explained mainly the disagreement found between the SSU-1199 and the equivalent ITS datasets. On one hand, the P. eumusae/P. fijiensis/S. quercicola node found in the SSU-1199 tree was rejected at p < 0.001 by the ITS dataset, and reciprocally, the SSU-1199 dataset rejected the ITS node S. quercicola/S. aceris mol. sp. 1, 2, and 4/S. quercicola/S. musiva (WSR, p = 0.002; SH, p < 0.001). On the other hand, although placement of the S. musiva taxon represented one additional point of disagreement, the alternate topology (S. musiva, S. ribes, S. alnifolia, and S. ostryae monophyletic) forced in the ITS partition was not significantly rejected by this dataset (WSR, p = 0.12; SH, p = 0.23). Interestingly, the incongruence observed in the SSU-1506 trees comparison was supported by high MP and ML bootstrap values in both the intron and the equivalent ITS tree but only the ITS tree topology was significantly rejected by the SSU-1506 dataset (WSR-test, p < 0.001; SH, p = 0.002). Furthermore, three of the five putative topological incongruences tested were rejected with a significant p-value (at the 5% threshold) by only one of the two reciprocal datasets (intron or equivalent ITS dataset). Thus, the topological tests applied here did appear useful and complementary to bootstrap support values to help determine in a statistical framework which node contributed to incongruence between data partitions. Finally, based on the remaining node comparisons, our detailed analysis suggests that horizontal transfer may have been implicated in at least two of the three intron positions.

Table 2. MP tree statistics, ML best-fitting substitution models, and parameter values for the group I introns datasets and the equivalent ITS datasets
Full size table

SSU-rDNA Group I Introns as Mobile Genetic Elements Among Septoria and Pseudocercospora spp.

Our detailed analysis supports the idea that horizontal transfer to the homologous site represents a prominent important force in group I intron evolution in the SSU-rDNA of the Mycosphaerella species sampled. Two large insertions (>400 nt) have been found in SSU-1199 introns (S. aceris mol. sp. 1 and 2) that could be remnants of inactivated endonuclease ORFs. HEGs are assumed to be central to the mobility of group I introns (for a review, see Haugen et al. 2005). The model of group I intron evolution postulated by Goddard and Burt (1999) involves recurrent gain through homing, degeneration and loss of the intron-HEG unit. Once the intron becomes fixed in the population, the HEG no longer has a biological function and does not confer benefit to the host; it accumulates mutations and eventually becomes inactivated. Over time, the intron loses its self-splicing ability and at last, unable to spread, is likely eliminated because of strong selection against nonfunctional gene products located in functionally important regions of rDNA (Dujon 1989). Lastly, the intact intron with the full-length HEG can remain active if it is reintroduced through gene flow or horizontal transfer to an intron-less population of the same or closely related species containing the appropriate recognition site to restart the homing cycle (Goddard and Burt 1999; Haugen et al. 2004, 2005). On the other hand, large insertions encoding HEGs remain generally rare in group I introns, as reflected in our study by introns located at positions SSU-943 and SSU-1506. If only endonucleases did mediate the lateral transfer of these introns, then one would have to suggest complete loss of these coding regions after the introns had attained their present distribution. In contrast, HEGs found in nuclear group I introns generally do not exhibit extensive deletions but rather frame-shift mutations or short truncations that result in their inactivation (Haugen et al. 2004; Bhattacharya et al. 2005). Thus, alternatively, the reverse splicing reaction of group I introns may represent another possible hypothesis to explain the distribution of self-splicing introns. This model has been supported by phylogenetic data to explain intron spread in the Physciaceae (Bhattacharya et al. 2002; Simon et al. 2005), and the Pezizomycotina nuclear rDNA (Bhattacharya et al. 2005) and in light of our data, the absence of HEGs or long inserts in the SSU-943 introns is consistent with reverse splicing being the likely mechanism of intron spread. Finally, if homing occurred at site SSU-1199, one of the most parsimonious hypotheses to explain the observed intron distribution at this insertion site (Fig. 3) would be, first, the spread and vertical evolution with loss of IE3 intron within the Mycosphaerella lineage and, second IE1 invasion through homing followed by vertical inheritance and HEG degeneration or loss.

Fig. 4.
figure 4

Phylogenetic comparisons between the group I intron tree from each insertion site and the equivalent ITS tree. Given the low variability observed within the species studied for (1) the intron nucleotide sites and (2) the ITS sequences, only one sequence per species was considered to facilitate tree reconstructions (see alignment in Supplementary Fig. S2). Intron and equivalent ITS alignment lengths, summary tree statistics issued from MP analyses, and ML models are given in Table 2. Bootstrap values ≥50% are presented above each node as follows: MP bootstrap value/ML bootstrap value. The italicized numbers under each branch correspond to branch lengths estimated under the ML models specified in Table 2. Boxes with arrows indicate the constraint nodes applied to the different datasets and results of both WSR- and SH-tests following constrained and unconstrained tree comparisons. Diagonal dotted lines indicate a significant difference between the intron and the host trees.

Full size image

Remarkably, one of the two horizontal transfers suggested by our data involved two anamorphic genera (Septoria and Pseudocercospora) within the teleomorphic lineage Mycosphaerella. Physical conditions and biological properties responsible for bringing both taxa close enough to make horizontal transfer possible remain generally speculative. In some cases, the close spatial and ecological proximities of multiple strains of fungi have been hypothesized to increase the likelihood of physical transfer. The parasitic fungal species we have sampled in this study may occur either in natural habitats on native tree species or within human mediated tree plantations. Previous studies have demonstrated that Mycosphaerella pathogens have adapted from native plants and could jump from completely unrelated hosts to infect plantation trees (Crous et al. 2004; Wingfield et al. 2001). This point was reinforced by the absence of the co-evolutionary pattern usually expected in parasites and host species phylogenies and the occurrence of unrelated Septoria species on the same host families (Feau et al. 2006). It is thus plausible that different members related to the teleomorphic genus Mycosphaerella occasionally come into physical contact in nature through species-jumps from different hosts. Furthermore, it has previously been hypothesized that mycoviruses may act as the external vectors in such contact cases (Gibb and Hausner 2003; Holst-Jensen et al. 1999). However, although many ascomycetous fungi harbor viruses, such elements have never been reported among Septoria and other anamorphs of Mycosphaerella. Other processes such as mycoparasitism, host-parasite transfers, and interspecific hybridization may also be advocated to mediate horizontal transfer across fungi.