Information Box 1. Species definitions and phylogenetics
In order for a hybrid to be considered interspecific, its parents must be recognized as distinct species according to established criteria (Claridge et al. 1997; Taylor et al. 2000). Any of three criteria are applied variously by different researchers to different groups of fungi and oomycetes: morphological, biological and phylogenetic species concepts. The morphological species concept is the traditional Linnaean approach based on phenotypes. However, genetic polymorphisms within species manifest phenotypic polymorphisms, which are also influenced by environmental variation and vary with age and development. To help resolve the limitations of the morphological species concept, a biological species concept is often adopted, whereby members of the same species should be capable of mating with each other when they have compatible mating types and possess the requisite female and male fertilities. This is an attractive concept because it applies neatly to humans and most plants and animals familiar to us. However, because it applies only to sexual organisms, it is a useless criterion for the myriad prokaryotes on our planet and of limited utility for other eukaryote groups, including fungi and oomycetes, which can include truly asexual species.
The third criterion defines a phylogenetic species as a population of organisms sharing suites of evolutionarily relevant characters that identify them as derived from a uniquely shared common ancestor. The growing popularity of this concept among mycologists arises from intractable problems for morphological and biological species. Of course, the phylogenetic species concept presents its own problems, such as the difficulty in determining which characters are evolutionarily relevant and to what degree. And, because monophyletic groups can be identified at any taxonomic level — not just the species rank — it is necessary to incorporate morphological or biological species criteria into the phylogenetic species concept. Furthermore, the common assumption of cladistic speciation, which predicts species monophyly, is probably violated in some instances (Avise & Wollenberg 1997; Craven et al. 2001b). Finally, hybridization between species can obscure species boundaries in phylogenetic analyses. These caveats notwithstanding, the phylogenetic species concept is currently the most effective at establishing an evolutionary basis for fungal systematics.
DNA sequence polymorphisms are attractive characters for phylogenetic species recognition because they are the most direct records of genetic evolution and not generally influenced by environment or development. A common criticism, however, is that most genes used for phylogenetic inference do not actually affect the morphological variation or interfertility/intersterility relationships relevant to other species concepts. This criticism arises from a misunderstanding of the aim of DNA-based phylogeny, which is essentially to read the historical record of gene evolution as a proxy for the history of species and higher taxon origins. However, the question remains as to how gene trees should translate to species trees; that is, how sequence-based phylogenetic data can be used to identify speciation events. At the very least, multiple unlinked genes or loci should be analysed.
A method to use multilocus phylogenies for estimating the last common ancestor of an interbreeding population is elaborated by Taylor et al. (2000) as ‘phylogenetic species recognition’. This approach is not as limited as the biological species concept because it can include asexual lineages, provided that they can be related to existing or historically recent interbreeding populations. It would be particularly interesting to apply this method to populations that are currently thought to constitute distinct species, but for which various levels of interspecies fertility have been observed; for example, for Melampsora species or Phytophthora species, which have been described mainly on the grounds of morphological distinctions and host range.
Shown here are phylogenies of the β-tubulin genes (tub2) and the translation elongation factor 1-α genes (tef1) for Epichloë species. Each dot or letter (for E. typhina) on a terminal branch represents the sequence from each isolate. Species have been described mainly on the basis of interfertility or intersterility, so different species names indicate different biological species. Shaded boxes surround clades representing phylogenetic species. In most instances all members of a biological species group in a distinct clade, so there is considerable concordance between biological and phylogenetic species. Both gene loci give similar indications of phylogenetic species. Although there are some topological differences between the gene trees (notice the position of E. brachyelytri, in particular), those differences were not statistically significant (Craven et al. 2001b). In contrast, the different gene trees gave significantly different branching orders within phylogenetic species, as expected. Note especially the different topologies associated with E. typhina isolates a–g, indicating that these isolates are members of the same interbreeding population.
These trees also indicate a conflict between phylogenetic and other species criteria, in that the former fails to distinguish E. typhina from E. sylvatica, despite indications that these two species are not interfertile, occupy different hosts and are morphologically distinct (Leuchtmann & Schardl 1998).
Information Box 2. Phylogenetic evidence for hybrids
Where phylogenetic species are otherwise discernible, interspecific hybrids would be identified as strains with relationships differing markedly from those of nonhybrids. The nature of the evidence for interspecific hybrids depends on the nature of the hybrids. This diagram shows relationships of β-tubulin (tub2) genes from Epichloë species and related hybrid and nonhybrid Neotyphodium species (note that Neotyphodium species, as defined currently, are asexual descendants of Epichloë species). None of the Epichloë species are hybrids, whereas all Neotyphodium species indicated except N. lolii and N. aotearoae are apparent hybrids. In most cases, the hybrids have two or (in the case of N. ×coenophialum) three tub2 alleles, a situation that can arise by any of three processes: (1) generation of multiple copies by gene duplication, or by aneuploidy or polyploidy arising in a clonal lineage; (2) ancient paralogy (gene duplication in the common ancestor of the whole group) with subsequent gene losses in many lineages; or (3) acquisition of multiple alleles from multiple species. In the Epichloë/Neotyphodium system it is possible to reject the first two possibilities as highly unlikely. The first would give a close grouping of the pairs or sets of alleles in each endophyte, not inclusion of each allele in a different clade, as shown here. The second would require an extremely large number of gene losses, and such losses would have to be contrived to ensure that members of each species tended to retain precisely the same paralogs to give the pattern shown here. The third possibility, interspecific hybridization, remains likely, particularly as a distinct ancestral Epichloë species is identifiable for each allele. Similar situations were observed for tef1 (translation elongation factor 1-α) and act1 (γ-actin) genes for several of these endophytes (Moon et al. 2000, 2002; Craven et al. 2001a), suggesting that the hybrid endophytes had considerably greater genetic material than nonhybrids. This prediction has been upheld for N. × coenophialum, which has c. 57 Mb of DNA, compared to c. 29 Mb for E. typhina, E. festucae, and N. lolii (Kuldau et al. 1999).
Some hybrids may lose their redundant gene copies. In such cases, phylogenetic analysis of each gene might place it within or near a different ancestral species. Such was the case for N. × uncinatum (Craven et al. 2001a), whose single tub2 allele grouped with E. typhina (as shown here), but whose single tef1 allele (not shown) grouped with E. bromicola. Other genetic analyses indicated that gene losses following hybridization have nearly brought N. × uncinatum to a haploid state.
Note that, in the absence of other biological information, it would be difficult to identify cases of hybridization, particularly if the process were very common in a group of fungi. In essence, the genotype of a hybrid such as N. × uncinatum, for example, might be taken as evidence that E. typhina and E. bromicola belong to a single interbreeding species. In this system, sexual and asexual endophytes are distinguished readily based on whether they produce presexual stromata on hosts, and by experimental mating tests (Schardl 2001). The sexual species give (for the most part) clear indications of distinct phylogenetic species, and the major deviations from this pattern are only associated with asexual species. In many other fungal systems, the sexual stages are so difficult to find or produce that distinctions between sexual and asexual species are very tentative (Taylor et al. 2000). Furthermore, if sexual hybridization occurs even rarely, but is nevertheless of evolutionary significance in the fungal group, phylogenetic species may appear much more broadly inclusive than experimentally delimited intersterility groups.
Another characteristic that makes many hybrids discernible in the Epichloë/Neotyphodium system is their tendency to retain the alleles from their different ancestors. However, the number of gene copies might also be unreliable as an indicator of hybrids in other fungal groups. Note, for example, that in Glomerella cingulata (Colletotrichum gloeosporioides), some strains possess two β-tubulin genes designated TUB1 and TUB2 (Panaccione & Hanau 1990), yet others possess only one or the other of these (Buhr & Dickman 1993). Given the extreme divergence of TUB1 and TUB2, these are almost certainly ancient paralogues, and one or the other has simply been lost in some lineages. However, even this is not a certainty, and it is conceivable that the two β-tubulin genes in G. glomerata were derived from an ancient hybridization event.
Given these difficulties identifying fungal hybrids, it seems likely that interspecific hybridization is a much more important evolutionary process in many fungal groups than has yet been demonstrated.