Arbuscular mycorrhizal fungi in the phylum Glomeromycota are found globally in most vegetation types, where they form a mutualistic symbiosis with plant roots. Despite their wide distribution, only relatively few species are described. The taxonomy is based on morphological characters of the asexual resting spores, but molecular approaches to community ecology have revealed a considerable unknown diversity from colonized roots. Although the lack of genetic recombination is not unique in the fungal kingdom, arbuscular mycorrhizal fungi are probably ancient asexuals. The long asexual evolution of the fungi has resulted in considerable genetic diversity within morphologically recognizable species, and challenges our concepts of individuals and populations. This review critically examines the concepts of species, communities, populations and individuals of arbuscular mycorrhizal fungi.
The arbuscular mycorrhizal fungi represent some of the most abundant organisms on this planet, where they form symbiotic associations with most land plants and are found in almost all vegetation systems from the sub-polar regions to the tropical rain forests, and even in some aquatic ecosystems. (Read, 1991; Nielsen et al., 2004). They form a unique symbiosis with the roots of most land plants, and their evolutionary success despite being asexual challenges our view on fungal evolutionary biology. Molecular analyses and studies of fossilized material suggest that the fungi have their origin with the first land plants, and the symbiosis may have been essential for the adaptation of plants to the terrestrial environment in the Ordovician (Berbee & Taylor, 1993; Remy et al., 1994; Redecker et al., 2000). Plant–soil interactions cannot be studied without considering mycorrhizal fungi, as they constitute a considerable and important part of the microbial biomass in many terrestrial ecosystems (Fig. 1) (Jastrow & Miller, 1997). The importance of mycorrhizal fungi in nutrient uptake of plants has received most attention, and has been the focus of several research programmes (Smith & Read, 1997). The effects on plants are not limited to the nutritional status of individual plants, but may also affect the composition and development of plant communities (Grime et al., 1987; van der Heijden et al., 1998). Although several earlier studies are entitled: ‘The effect of vesicular–arbuscular mycorrhiza on ...’, it is now generally accepted that although many arbuscular mycorrhizal fungi show low host specificity, species may differ in their effects on plant growth (Ravnskov & Jakobsen, 1995). This could be because of differences among species in their ability to acquire nutrients (Smith et al., 2004) and to provide the host plant with an improved stress tolerance toward drought (Michelsen & Rosendahl, 1990) and pathogens (Rosendahl, 1985).
The acknowledgement of physiological and ecological differences between arbuscular mycorrhizal fungi has necessitated an assignment of species names, or even isolate or strain identification, for the fungi of interest. Without such identification, it is not possible to compare studies and explain discrepancies between results obtained from apparently similar studies. More recent studies have focused on the distribution of arbuscular mycorrhizal fungi along environmental gradients and their potential influence on plant communities. The fungal species are often identified from their DNA sequences, and our interpretation of the studies needs not only a consensus on taxonomy and species recognition, but also on the concepts of abundance and diversity.
Diversity has been used and misused as a concept in both ecology and politics. The basic idea is to describe the multiplicity of nature, and there is no doubt that it is an important measure in ecology and conservation biology. Studies of microorganisms challenge the concept for several reasons. First, it is not possible to use the same criteria for species recognition in animals, plants, fungi and bacteria; second, individuality may not have the same meaning in these groups. The clonal structure of Glomeromycota populations may further challenge the concept of diversity as a measurable size, as the distinction between populations and individuals becomes less apparent. Problems in standardization of concepts create unnecessary problems and misunderstandings when different experiments are compared. This review will attempt to discuss critically the concepts of species, communities, populations and individuals of arbuscular mycorrhizal fungi.
II. Taxonomy and species recognition
The arbuscular mycorrhizal fungi belong to the Glomeromycota. They were formerly placed in the Zygomycota (Thaxter, 1922; Gerdemann & Trappe, 1974), but molecular analyses suggest that they should have their own phylum (Schüßler et al., 2001). The affinity to Zygomycota was based on the interpretation of the spores as azygospores, which are zygospores with only one gametangium. Later, the spores were considered to be a sporangium in which the spores do not develop, or a merosporangium with only one spore. The homology of the spores to other known fungal structures has never been established, and recent studies suggest that the Zygomycota are polyphyletic. At present the Glomeromycota has no obvious affinity to other major phylogenetic groups in the kingdom Fungi (James et al., 2006). The interpretation of homologies between Glomeromycota spores and other fungal structures is further complicated by the fact that Glomeromycota spores may, among themselves, represent different morphological and functional structures. In some Glomus species such as G. intraradices, the structure of the spore wall is simple, whereas the spores in Scutellospora spp. can have several inner membranous walls and develop complex germination shields. The finding of dimorphic species with both Glomus and Acaulospora morphs further suggests that all spores in the Glomeromycota may not represent homologous structures (Morton & Redecker, 2001).
The first classification systems were not always based on the Botanical Code, but used descriptive names such as yellow vacuolated (YV) for Glomus mosseae, honey-coloured sessile for Acaulospora laevis, etc. (Mosse & Bowen, 1968); or E1, E2, E3, etc. (Gilmore, 1968). These descriptive names were often mistaken for isolate or strain numbers. For example, the famous E3 isolate from Rothamsted Experimental Station only resembled Gilmore's E3 from California, but was actually isolated in the UK. The first taxonomic monograph was published in 1974 by Gerdemann & Trappe (1974), but no attempts to monograph the Glomeromycota have been made since that time. As an alternative, N. Schenck and Y. Perez at the University of Florida collected the species descriptions in a manual that was used at the taxonomic workshops in the 1980s and 1990s. A major breakthrough in species identification has been the establishment of culture collections and reference strains: Banque Européenne des Glomales (BEG) in Europe and the International Culture Collection of Vesicular (Arbuscular) Mycorrhizal Fungi (INVAM) in North America. The descriptions found on the INVAM web page (http://invam.caf.wvu.edu), provided by Joe Morton, have served as a reference for species designation by researchers worldwide. The descriptions are based on INVAM cultures, and do not attempt to cover the variation in morphological characters. Although there are problems with correct identification of species in various culture collections, these collections have greatly facilitated comparisons among experiments where such bona fide cultures are used.
The taxonomic system was based on interpretation of morphological features of the resting spores, and later studies have included characters related to the ontogeny of the spores (Franke & Morton, 1994). Interpretation of these characters has laid the ground for several controversies among taxonomists, a phenomenon not specific to mycorrhizal research. The main discussion has focused on the spore wall characters, where the relevance of the number and position of walls and wall layers has been a key issue. A so-called murograph was included in the description of some species, and new species were described containing up to seven walls. Several users of the classification found it very hard to identify species based on the descriptions (Koske, 1986), and some claimed that the number of walls depended more on the age of the spore and how hard you pressed the cover slip, than on the actual taxonomic identity (J. C. Dodd, personal communication). The discussion on whether arbuscular mycorrhizal fungi had several walls (sensu Walker, 1983) or one wall with several layers (Berch, 1987) may have been regarded as a purely academic debate, but the discussion reflected the question of whether or not the diversity had a common origin, or if the wall characters were independent characters. Later studies of wall ontogeny of Scutellospora pellucida showed that walls or wall layers cannot be seen as independent characters, as the inner walls differentiate during germination of the spore (Franke & Morton, 1994). On several occasions Morton has made a plea for a hierarchical interpretation of taxonomic characters (Morton et al., 1992; Morton 1995), but species are still described without such interpretation of the characters.
The number of recognizable species in the Glomeromycota is not apparent. Despite the worldwide distribution of the fungi, relatively few species have been described. Some authors recognize c. 200 species, but some additional descriptions exist and some of the descriptions could be synonyms. Recently, A. Schüßler in Munich listed all described species in Glomeromycota. The list is updated, and available from: http://www.lrz-muenchen.de/~schuessler/amphylo/amphylo_species.html. There are 104 described species in the genus Glomus alone, but interpretation of some of these descriptions can be problematic. For some species the herbarium specimens are unavailable and it is not possible to verify the descriptions; in other cases the material is in a very bad condition and the taxonomic features are no longer recognizable. This is often combined with old, rather vague descriptions of the material, leaving interpretation of key characters open for discussion.
1. Species concepts and species recognition
The number of species depends to a great extent on the concept used to recognize the species. Most biologists agree on an evolutionary species concept based on Darwinian evolution that defines species as: ‘a single lineage of ancestor-descendent populations which maintains its identity from other such lineages and which has its own evolutionary tendencies and historical fate’ (Wiley, 1978). However, the evolutionary species concept is a theoretical concept, and cannot be used to recognize or diagnose species. Instead, morphological, biological and phylogenetic species concepts can be regarded as operational concepts, as they all intend to recognize evolutionary species (Mayden, 1997). Based on this, Taylor et al. (2000) used the term ‘species concept’ for the theoretical concept and the term ‘species recognition’ for the operational concepts.
Taylor et al. (2000) discussed various species recognition concepts in fungi and proposed the Genealogical Concordance Phylogenetic Species Recognition (GCPSR). The basic idea of GCPSR is that in a phylogenetic tree, concordant branches represent species, whereas incongruity caused by recombination is found within species. They state that a drawback of the concept is that it will not be useful for strictly clonal organisms, but claim that such organisms may not exist, and that the studies of primarily clonal fungi have shown that recombination takes place. They conclude that if true clonal fungi exist, they are very rare, but they do not consider the Glomeromycota. Recombination has not yet been proved in the Glomeromycota, and most evidence suggests that the fungi have evolved asexually. The lack of incongruity to delimit species means that GCPRS cannot be used to recognize arbuscular mycorrhizal fungal species.
Presently, there is no formalized operational species concept in Glomeromycota. Species are described based on morphology and information from sequences of ribosomal RNA genes (rDNA). The access to sequence information has not changed the criteria for species recognition in the Glomeromycota, and studies based on rDNA have often confirmed the morphologically defined species (Fig. 2). Some new descriptions of Glomeromycota taxa include sequence information, which is particularly valuable for delimiting taxa with few available morphological characters (Walker et al., 2007). The molecular data have substantially changed the systematic, and several new genera and families have been erected (Walker & Schüßler, 2004; Walker et al., 2007).
III. Communities of arbuscular mycorrhizal fungi
Previously, community studies of arbuscular mycorrhizal fungi relied on the presence of resting spores in the soil. These spore surveys are often criticized, as it cannot be determined if species are truly absent or they are present but not sporulating. A comparison of communities is thus not possible if the communities are dominated by nonsporulating fungi. For example, Sparling & Tinker (1975) observed fewer spores in permanent vegetation and suggested that sporulation is related to disturbed systems. Quantitative studies of arbuscular mycorrhizal fungal communities based on the presence of spore numbers are also complicated as some species produce few spores on the mycelium, whereas cluster-forming species such as G. intraradices, G. versiforme or G. fasciculatum produce hundreds of spores on the same hypha. Although the spore surveys may not provide sufficient information on communities of arbuscular mycorrhizal fungi, they have revealed interesting patterns (Walker et al., 1982; Bever et al., 2001). A problem with the previous spore-based community studies is that they are sometimes difficult to interpret and relate to more recent studies, as the species identification is uncertain and may have changed over time. In several studies from the 1970s and 1980s, Glomus fasciculatum appears to be a very common species, but after the species was redescribed (Walker & Koske, 1987) it was shown to be uncommon; reports of G. fasciculatum in older surveys and experiments must have been of other species. Application of molecular techniques has made such comparisons more reliable as sequences are deposited in GenBank and can be compared with other known species.
1. Species known only from environmental samples
The molecular approach to community ecology of arbuscular mycorrhizal fungi has revealed a large unknown diversity, and has shown that known species sometimes can account for only a minor proportion of the diversity. In a community study from a perennial grassland in Denmark, four out of 11 phylogenetic clusters in Glomus belonged to species where only sequence information is available (Rosendahl & Stukenbrock, 2004), and only one sequence type of 10 was identified from another site in a study from Scotland (Gollotte et al., 2004). These studies are based on large subunit (LSU) rDNA, whereas most studies are based on small subunit (SSU) rDNA. The resolution of these genes is different, and a direct comparison of phylogenetically defined taxa is not possible. Closely related taxa such as G. caledonium and G. geosporum have only 2- or 3-bp substitutions in SSU, but can be easily separated by their morphology and their LSU rDNA sequence.
In phylogenetic lineages with no known morphologically defined species, it is not possible to make this evaluation. Often lineages are defined as taxa, but the taxa defined solely from trees based on single genes should be referred to as phylogenetic clusters rather than phylogenetic species, as the later implies species defined based on gene concordances (Taylor et al., 2000). Divergence in one locus may give a history only of the locus, not of the species, as different genes or gene regions may be subjected to different mutation rate, selection pressure, etc. To detect cryptic species, phylogenies inferred from more than one genomic region (multilocus DNA sequence phylogenies) must be used. Such phylogenies should detect speciation before the whole genome has diverged (Kohn, 2005).
The gene tree approach to taxon recognition is sensitive to sampling bias. Often only sequences from a particular location are included in the phylogenetic analysis. If some taxa are poorly represented in an area, or intermediate phylogenetic lineages are missing, artificial high similarity between distant taxa may occur. In a community study from Denmark (Rosendahl & Stukenbrock, 2004), where samples were taken as individual root fragments, some groups were clearly better represented than others (Fig. 3). In the phylogenetic lineage with G. intraradices (often referred to as Glomus Ab), several sequences were found only once. If the level of sequence divergence observed between the morphologically defined species G. caledonium, G. mosseae and G. geosporum is used to delimit species, at least 15 phylogenetic taxa can be defined in this study. All sequences fall within five main lineages, but interestingly two of the main lineages (I and II) have no known morphologically defined species.
2. Species abundance and distribution
From the community studies, it is obvious that communities of arbuscular mycorrhizal fungi follow the same distribution pattern as many other organisms, with few very common species, and a number of more and more rare species (Fig. 4). Which fungi are common and which are rare depends on the habitat. We still know little about the autecology of arbuscular mycorrhizal fungi, but from several observations it can be concluded that fungi such as G. mosseae, G. caledonium, G. geosporum and Scutellospora pellucida are common in arable soils, whereas the patterns in woodlands and uncultivated grasslands are less apparent. This is mainly because very few such systems have been studied. In their overview of studies of global arbuscular mycorrhizal fungal communities, Öpik et al. (2006) conclude that some fungi could be specific to forests. However, this comparison was based on very few studies, and there was an obvious correlation between the number of studies at different sites and the number of phylogenetic groups found.
The development of molecular tools to characterize microbial communities has resulted in a considerable increase in the number of community studies of arbuscular mycorrhizal fungi. The fungi have been identified using either sequencing (Van Tuinen et al., 1998) or denaturing gradient gel electrophoresis (DGGE) and terminal-restriction fragment length polymorphism (T-RFLP) approaches (de Souza et al., 2004; Lekberg et al., 2007) following the isolation and PCR of fungal DNA. In studies where Glomeromycota DNA from roots has been sequenced, the ribosomal genes have been targeted: either the SSU, first used by the group at York University (Helgason et al., 1998, 2002); or the LSU, used in Dijon (Van Tuinen et al., 1998) and Copenhagen (Kjøller & Rosendahl, 2000). In many studies, only one pair of primers (NS31 and AM1) for SSU was used to cover the diversity in Glomeromycota (Helgason et al., 1998). This primer pair will amplify most but not all groups (Redecker et al., 2003), and several nonglomalean sequences may be amplified (Douhan et al., 2005), making subsequent use of DGGE and T-RFLP problematic. The LSU provides a better resolution, but several primers are necessary for amplifying all genera of Glomeromycota. Another important disadvantage of targeting the ribosomal genes is that copies of the multi-copy gene may vary within a single spore, making subsequent cloning necessary (Scheublin et al., 2004). If the amplified PCR products are cloned, the quantitative aspect is lost, which can be important when host preferences or responses to environmental gradients are studied. To retain the quantitative aspect, LSU rDNA was amplified from 1-cm root fragments and sequenced directly, and the frequency of sequence types was quantified as the number of colonized root fragments (Rosendahl & Stukenbrock, 2004; Stukenbrock & Rosendahl, 2005b). Although both approaches have drawbacks, the results have been remarkably similar, showing a limited diversity of arbuscular mycorrhizal fungi in most vegetation systems.
The species richness in arbuscular mycorrhizal fungi communities is not easy to estimate because of problems in identifying nonsporulating species. A study of spore diversity in a field in North Carolina revealed 37 species recognized as spores either sampled directly or trapped on plants (Bever et al., 2001). This relatively high species richness probably reflects the sampling intensity, rather than characteristics of the area. The authors made no attempt to extrapolate from the study, but noted that the richness in the field was similar to what has been reported from continents. Öpik et al. (2006) compared most community studies and found that the average number of species (taxa) ranged from 1 to 29. However, the criterion for species recognition is crucial for interpreting and comparing community studies as well as the different genetic markers used to identify species.
IV. Populations of arbuscular mycorrhizal fungi
The concept of a population is used widely in ecology, population genetics and evolutionary biology. Several definitions can be found, but the central theme is that a population is used to describe a group of individuals of a certain species. Two main paradigms exist: the ecological and the evolutionary. The ecological paradigm identifies a population as: ‘a group of organisms of the same species occupying a particular space at a particular time’ (Krebs, 1994). Although this definition may sound precise, the problem is that different researchers may look at different space and time scales, and thus not draw the same conclusions. To overcome this problem, the ecological population concept can be amended by adding: ‘and have the opportunity to interact with each other’ (Waples & Gaggiotti, 2006). The evolutionary paradigm is centred on the definition by Dobzhansky (1937): ‘a community of individuals of a sexually reproducing species within which mating takes place’. This definition emphasizes the reproductive interaction between the individuals, and has some obvious limitations when considering fungal populations. The broader definition by Waples & Gaggiotti (2006), which identifies a population as ‘a group of individuals of the same species living in close enough proximity that any member of the group can potentially mate with any other member’, is more appropriate to mycology. The main idea in the evolutionary paradigm is that populations can be characterized by their allele diversity. Differences in allele frequencies reflect subdivisions between populations. Gene flow is the main introgression of new alleles into a population and the main mechanism to reduce genetic differentiation between populations. However, in strictly asexual organisms the genes are linked and can only move as a block. This type of gene flow is better termed ‘genotype flow’. In arbuscular mycorrhizal fungi without recombination, it makes no sense to talk about gene flow, as gene flow considers single alleles. With complete linkage of alleles, the individuals will migrate and will keep their genetic integrity.
1. Global migration or endemism?
Populations of microbial eukaryotes are believed to cover continents because of the effective dispersal of airborne spores. This is often referred to as the ‘ubiquitous dispersal hypothesis’ (Finlay, 2002; Fenchel & Finlay, 2004), and is based on the famous citation ‘everything is everywhere, but the environment selects’ (Baas Becking, 1934). The hypothesis suggests that fungi have large effective population size, and that genetic drift plays a minor role in evolution of the fungi. The hypothesis is controversial and has been challenged by population genetic studies of cosmopolitan fungi, demonstrating that what may look as cosmopolitan is actually hidden endemism (Taylor et al., 2006). Several species in the Glomeromycota are known to occur on different continents supporting the ubiquitous dispersal hypothesis. However, genetic differentiation between geographical isolates has not been studied, and it is not known if the morphologically recognized species hide endemic cryptic species. Alternatively, the apparent global distribution of some arbuscular mycorrhizal fungi species could be a recent phenomenon caused by human activity related to agriculture. Several studies have demonstrated global migration among fungal pathogen populations (Goodwin et al., 1994; Zhan et al., 2003).
Recombination and gene flow are important factors in shaping the genetic structure of populations. Asexual fungi evolve without the homogenizing effect of recombination and gene flow. The individual genotypes within a population can evolve differently by the accumulation of mutations in loci that are not under selection. In time, this genetic differentiation of clonal individuals can lead to a phenotypic or functional differentiation. Dispersal becomes important in fungi where individuals that differ from the most frequent genotype have a selective advantage and can establish or even become dominating in an area. Among pathogens, such asexual dispersal may initiate epidemics and even pandemics. The human pathogen Penicillium manefii is an example of an asexual fungus in which, despite widespread aerial dispersal, isolates of the species showed extensive spatial genetic structure at local and country-wide scales (Fisher et al., 2005). The authors suggested that this endemism is a consequence of the lack of sexual reproduction that has led to evolution of niche-adapted genotypes despite the extensive aerial dispersal of the pathogen. Their results are interesting when considering arbuscular mycorrhizal fungi. If asexuality has led to the evolution of niche-adapted genotypes, it is reducing the potential of the fungus to diversify. In a heterogeneous environment, variance in fitness between clonal lineages is expected to lead to populations that are adapted to local conditions (Goddard et al., 2005). In this case, limited dispersal becomes an inevitable consequence of asexuality, as invading clones are outcompeted by better-adapted local competitors because of the decreased ability of invaders to diversify within novel environments (Buckling et al., 2003).
2. Ancient origin without recombination?
Despite an early observation of zygospore formation in Gigaspora (Tommerup & Sivasithamparam, 1990), arbuscular mycorrhizal fungi are generally believed to be asexual. Recombination can be shown by analyses of linkage of multiple alleles (index of association) (Maynard Smith et al., 1993). This has been used to demonstrate recombination in Coccidium immitis, where no sexual state has been observed (Burt et al., 1996). In arbuscular mycorrhizal fungi, analysis has not so far revealed signs of recombination in data obtained from pot cultures or field-collected spores (Rosendahl & Taylor, 1997; Stukenbrock & Rosendahl, 2005a). A single study found recombination in a data set based on inter simple sequence repeat (ISSR) markers obtained from field-collected spores (Vandenkoornhuyse et al., 2001). However, dominant markers such as ISSR are problematic for estimating linkage, as the homology between the alleles is difficult to verify. If the alleles are not homologous, analyses such as index of association (Maynard Smith et al., 1993) may suggest recombination erroneously. A study by Gandolfi et al. (2003) used haplotype networks to demonstrate that DNA sequences from the same Glomus species may show signs of recombination. Unfortunately, the data were based on sequences from GenBank, and some of the data sets contained sequences from more than one species. Some of these species are only distantly related, and the reticulate structure of the haplotype networks is more easily explained by a saturation of mutations rather than recombination.
Linkage of alleles indicates lack of recombination, but will not prove that the organisms are ancient asexuals. Normark et al. (2003) hypothesized that a clonal population structure would appear after few asexual generations if the effective population size (Ne) is low. This was based on the assumption that the coalescent time since the most common recent ancestor is approximately 2Ne generations (Birky, 1996). The coalescence model assumes a neutral model, and the actual coalescent time would be less if the loci are linked. We know little about the population size and generation time of arbuscular mycorrhizal fungi, but if the population size is low, the recombining structure of populations should be lost in few thousand generations (Normark et al., 2003).
A genomic signature that would advocate for an ancient asexual evolution would be strong divergence between alleles in the same locus, as has been shown for diploid bdeloid rotifers (Mark Welch & Meselson, 2000). However this would only be the case with diploid organisms, as the pattern of divergence is equally consistent with the organism being sexual and haploid. Strong divergence has been observed in genes that may have a common origin and may suggest an ancient asexual evolution of arbuscular mycorrhizal fungi (Kuhn et al., 2001; Pawlowska & Taylor, 2004).
Evidence against an ancient asexual evolution would be the existence of genes involved in sexual functions. Preliminary studies have found possible candidate genes, and this may change our view of arbuscular mycorrhizal fungi as ancient asexuals (Colard et al., 2007). Further evidence of recent meiotic events would be the presence of selfish genetic elements or retrotransposons, which are believed to disappear in the absence of meiosis (Hickey, 1982; Arkhipova & Meselson, 2000). Retrotransposons have been detected in Glomeromycota genomes, but contained stop codons and were not expressed in either hyphae or germinating spores (Gollotte et al., 2006).
3. Genetic structure of arbuscular mycorrhizal fungi populations
The genetic structure of a population refers to the distribution of genetic variation, and can be studied by hierarchical sampling and analysis of genetic variation within and between fields, plots, subplots, etc. Nei's (1987) analysis of genetic variation and analysis of molecular variance (amova) (Excoffier et al., 1992) are two methods for estimating population differentiation directly from molecular data and for testing hypotheses regarding factors that may cause such differentiation.
AFLP data analysed by amova (Koch et al., 2004) and multilocus genotyping of arbuscular mycorrhizal spore populations in agricultural systems, analysed using Nei's approach (Stukenbrock & Rosendahl, 2005a), showed a patchy distribution of genotypes within populations of common mycorrhizal species. A patchy distribution of arbuscular mycorrhizal populations will arise when genotypes in the field are structured as distinct mycelia that are able to establish and maintain their genetic integrity. In an agricultural system with annual crops, the turnover of root systems requires germination of new mycorrhizal spores every year. In a model that will explain the observed pattern, the spores germinate in the spring and hyphae of the same genotype fuse into a common mycelial network (Fig. 5). This will result in genetically uniform hyphal networks producing new spores of the same genotype. Vegetative incompatibility between Glomus isolates from geographically different areas has been reported by Giovannetti et al. (2003), also demonstrating a high frequency of self-fusion between hyphae of the same isolates. If self-recognition in arbuscular mycorrhizal fungi ensures hyphal fusion only between hyphae of the same genotype, this mechanism may be regarded as a genetic bottleneck, excluding rare or mutated genotypes from the mycelial networks. It is, however, important to note that the genetic background for incompatibility in arbuscular mycorrhizal fungi is not known, and could be different from what has been studied in Ascomycota (Glass et al., 2004).
V. Individuals, genets or clones?
We know very little about what constitutes the individual mycelium of an arbuscular mycorrhizal fungus. Several authors have discussed individuality in fungi, but no conceptual agreement exists. This is partly because individuals can be defined in both genetic and physiological terms. The mycelium that grows from a spore (sexual or asexual) or a hyphal fragment will, in physical or physiological terms, constitute an individual. However, the genetic or evolutionary unit will comprise genetically identical fragments or spores derived from mating between two nuclei. Such units can be defined as genets. In plants, genets are defined as all vegetative derivatives of a seed, including vegetative propagated plants from stolons or rhizomes. In mycology, the same definition can be used for many basidiomycetes with dikaryotic or diploid mycelia. In Basidiomycota, fusion between two haploid gametes will form a new dikaryotic mycelium. Even if the fungus may have subsequent asexual propagation, the genets are similar to plant genets in both functional and genetic aspects. This definition is not directly applicable for fungi with haploid mycelia such as the Ascomycota, Zygomycota and Glomeromycota (Hijri & Sanders, 2004). In Ascomycota, fusion of gametes or gametangia will result in dikaryotic ascogenic hyphae that will form the asci where meiosis takes place. This structure is often part of a fruiting body (ascoma), which is neither a functional nor an evolutionary unit. Anderson & Kohn (1995) proposed that genets of haploid fungi (e.g. Ascomycota) are defined as all vegetative derivatives of the fusion of two genetically unlike gametic nuclei and the following meiosis. For fungi with asexual reproduction, genets can be composed of several functional units, but they will all have the same origin. If the origin is not known, the term ‘clone’ is used to cover organisms with the same multilocus genotype. The term clone often refers to genetically identical individuals, but clones of ancient origin might accumulate mutations during mitotic cycles. Such nearly identical clones can be defined as belonging to the same clonal lineage. Identification of such lineages can be difficult if the organisms are ancient asexuals and/or distributed over wide areas, as the organisms may develop new phenotypes and genetic differences that could be interpreted as representing more than one genet.
1. Recognizing genets and clones of arbuscular mycorrhizal fungi
Very few authors have addressed the issue of individuals and genets of arbuscular mycorrhizal fungi. Again, it is important to distinguish between functional or physical coherent mycelia and genetic evolutionary units. Studies have been made of the extension and growth rate of individual hyphae (Mosse et al., 1982; Warner & Mosse, 1983) suggesting that arbuscular mycorrhizal fungi could be capable of forming extensive coherent mycelia. However, it is not known if the mycelia are coherent or fragmented into smaller ramets. Anastomosis in arbuscular mycorrhizal mycelia can be observed in roots as H-connection, and between hyphae of germinating spores (Mosse, 1959). These anastomoses may result in a reticulate growth of the mycelium, and may also allow possible nuclear migrations between individual mycelia. The experimental system set up by Manuela Giovannetti and coworkers in Pisa has made it possible to observe hyphal anastomosis in vivo, and revealed that the germinating hyphae of G. mosseae and other arbuscular mycorrhizal fungi anastomose shortly after germination (Giovannetti et al., 2001). In a later study they demonstrated that mycelia colonizing different host plants form anstomoses, which makes it possible to have an infinite hyphal network that connects different plant species (Giovannetti et al., 2004). Nuclear migration through the anastomoses was also observed, but a study using isolates from different geographical locations showed that anastomoses occurred only within the same isolate of G. mosseae (Giovannetti et al., 2003). The studies were conducted under laboratory conditions and therefore cannot be translated to field conditions, but the potential of forming hyphal networks has led several authors to discuss their role in nutrient acquisition of the plant (Jakobsen et al., 1992; Jansa et al., 2003).
If arbuscular mycorrhizal fungi grow as fragmented mycelia all belonging to the same clone, such mycelia may not differ phenotypically. This raises the question of functional redundancy – a subject that has been discussed by several authors. Fitter (2005), in his presidential address, discusses functional redundancy and comes to the conclusion that the concept of functional redundancy may itself be redundant, as the question can be reduced to: what determines the ability of numerous species with similar ecological function to coexist? The question is highly relevant for Glomeromycota, as the coexistence of clonal lineages of the same phylogenetic type or species is a fact within fields (Stukenbrock & Rosendahl, 2005a). Functional differences between arbuscular mycorrhizal fungi may exist, but so far it has not been possible to assign specific functions to species or isolates. A large variation was seen in phosphorus uptake and the amount of external mycelium within isolates of both G. mosseae and G. caledonium, but isolates within each species were strikingly similar in their length-specific hyphal phosphorus uptake. (Munkvold et al., 2004). Furthermore, isolates of G. intraradices from one field differed in their effect on plant growth (Koch et al., 2006). This indicates that the functional diversity could be structured in the field, but in order to study this a more detailed sampling of species and isolates from various spatial scales is required.
2. Homokaryons, heterokaryons or polyploids?
Some authors have claimed that the arbuscular mycorrhizal fungi represent multigenomic organisms (Kuhn et al., 2001; Sanders, 2002), and it has been hypothesized that this heterokaryotic structure has arisen by hyphal anastomosis between genetically different mycelia and by accumulation of mutations (Bever & Wang, 2005). The issue is controversial, and raises several questions regarding the ecology and evolution of such multigenomic organisms. Studies of arbuscular mycorrhizal fungi in situ have not confirmed the multigenomic status of the fungi. Instead, the fungi seem to form discrete and genotypic uniform mycelia. (Stukenbrock & Rosendahl, 2005a).
The existence of high genetic variability within spores cannot be questioned, and multiple variants of ribosomal genes are known to occur within single spores (Sanders et al., 1995; Corradi et al., 2007). However, whether the observed polymorphism is structured between genetically different nuclei (Kuhn et al., 2001), between different sets of chromosomes (Pawlowska & Taylor, 2004), or within each nucleus as duplicated genes (Hosny et al., 1999; Rosendahl & Stukenbrock, 2004) is a matter of debate. In sexual organisms, tandemly repeated copies of rDNA genes are homogenized by concerted evolution, a process where unequal crossing-over and gene conversion will homogenize sequences during meiosis. The lack of meiosis excludes concerted evolution and leads to different gene variants in genomes, as observed in other asexual organisms (Gandolfi et al., 2001).
The hypothesis of multigenomic mycorrhizal mycelia raises several questions regarding the biology and evolution of such organisms. Hyphal fusion and heterokaryon formation is a known phenomenon in filamentous Ascomycota (Glass et al., 2004). Although there are apparent benefits associated with heterokaryon formation, heterokaryosis by hyphal fusion is believed to be virtually excluded in nature by genetic differences at heterokaryon incompatibility loci. This nonself recognition mechanism prevents the establishment of a compatible heterokaryon between genetically different individuals. The patchy distribution of fungal genotypes in nature may reflect this nonself recognition mechanism, which could ensure genetic integrity of the mycelia.
The possible existence of heterokaryosis in the mycelia of arbuscular mycorrhizal fungi raises another important question. How would gene regulation be coordinated between the populations of genetically different nuclei in one continuous mycelium? In studies regarding patterns of gene expression, heterokaryosis has not been reported. It is striking that when the gene diversity is studied from markers developed from cDNA, no sign of heterokaryosis is seen, whereas markers developed from degenerated primers result in several gene variants within each spore. Future research on arbuscular mycorrhizal fungi should focus not only on the genetic background of hyphal anastomosis, but also on variation in cDNA libraries within and between populations of different mycorrhizal genotypes.
VI. Speciation in Glomeromycota
Species are regarded as a fundamental unit in biology, yet the true existence of such units can be questioned. Speciation has been reviewed by several authors, but few reviews focus on speciation in fungi (Kohn, 2005). As stated earlier, reproductive isolation is believed to be an important factor for speciation in sexual species. Within the species, interbreeding will unify the individuals, and the restricted gene flow between species will allow the species to diverge. In asexual organisms, this unifying effect of interbreeding does not exist, and it has been postulated that asexual species do not exist as the individual is the only evolutionary unit (Coyne & Orr, 1998). Whether speciation occurs in asexual organisms is still unclear, but if reproductive isolation cannot explain the existence of species, then adaptation to specific niches must be a crucial factor in speciation (Coyne & Orr, 1998). This suggests that speciation can occur sympatrically in different niches, and that asexual species can be revealed as phylogenetically distinct clades. Most asexual organisms are probably of recent origin, and it is difficult to study the processes of speciation. Asexual ascomycetes such as Penicillium and Aspergillus have closely related sexual relatives, and their origins are probably due to loss of a sexual cycle (LoBuglio et al., 1993). Other results indicate that asexuality in the rice blast pathogen Magnaporthe oryzae could be the result of loss of a mating type during adaptation to rice, and thereby also a recent event related to the onset of agriculture in Asia (Couch et al., 2005). An exception is the previously mentioned asexual bdelloid rotifer that may be more than 100 million yr old. Interestingly, recent results show that these organisms diversify into entities similar to sexual organisms, and raise the question as to whether sex is necessary for speciation (Fontaneto et al., 2007). The Glomeromycota have no known closely related sexual species and, as discussed earlier, their asexuality could also be of ancient origin. Similar to the bdelloid rotifers, morphological species do exist, and the species are supported by phylogenetic analyses of molecular data. The closely related species G. mosseae, G. geosporum and G. caledonium clearly represent entities with a morphological and genetic integrity (Fig. 2). The age of the divergence of these species is not known, but it could be between 10 and 100 million yr. If speciation occurs without recombination and with complete linkage of genes, selective sweeps must have occurred at various times. Such selective sweeps may make intermediate lineages extinct, leaving lineages that are genetically distant (Fig. 6). If the linkage of genes has resulted in the evolution of genetically distant individuals that can be regarded as physiological species, the phylogenetic and morphological species that we recognize today may represent a much higher level in the genealogical hierarchy compared with other fungi.
The genetic differentiation in Glomus spp. populations observed within fields can be the result of niche differentiation in a heterogeneous soil environment (Figs 2, 3). Both experimental and retrospective studies support this hypothesis. An arbuscular mycorrhizal community assembly was shown experimentally to depend on the phylogenetic relatedness of species, suggesting that reduced competition between distinct evolutionary lineages promotes their coexistence (Maherali & Klironomos, 2007). This experimental study was conducted with phylogenetically distant arbuscular mycorrhizal fungi, but a similar pattern was observed in an analysis of 400 Glomus sensu stricto sequences obtained from perennial grassland in Denmark (Rosendahl & Stukenbrock, 2004), where the sequences were organized in a distinct pattern with five major lineages above species level. However, the study also revealed that there were several sequence types coexisting within these clades. Similarly, a study from an agricultural field showed that several genotypes of three Glomus species were able to coexist (Stukenbrock & Rosendahl, 2005a). Whether these genotypes compete is not known, but other studies have shown that isolates within the same species can be phenotypically different, which may allow them to coexist in the same habitat (Koch et al., 2004; Munkvold et al., 2004). Although it may be tempting to suggest that the haplotypes have evolved as a result of niche differentiation, the number of mutation suggests that the diversification is ancient, at least exceeding the age of Denmark (approx. 10 000 yr). The diversification must therefore have occurred in a different environment, and the haplotypes colonized the field later.
Arbuscular mycorrhizal fungi are often referred to as mysterious, and very different from other fungi. This review has attempted to explain Glomeromycota diversity within the paradigms of population genetics and evolutionary biology, and there are no apparent features of arbuscular mycorrhizal fungi that contradict our understanding of speciation and population differentiation. The most important feature of the Glomeromycota is the possible existence of ancient asexual lineages. Coyne & Orr (2004) suggest that asexual species are not biological species but a collection of ‘microspecies’, with each individual propagating its own genetic lineage. This certainly seems to be the case with the ‘clones’ of Glomus spp. and the G. intraradices isolates from Switzerland (Koch et al., 2006), but could also be the case further in the genealogical hierarchy if the morphological species have evolved asexually. The Glomeromycota species may be recognized as microspecies sensu Coyne & Orr (2004), but a significant difference from other such clonally propagated entities is the considerable age of the lineages. The speciation events can be ancient, and even genera within the families may have evolved asexually. If the last meiotic events are ancient, Glomeromycota may consist of very few clones or clonal lineages with several morphological species originating from the same clone. Future studies should use coalescence-based models to determine the age of the speciation events. This is, however, complicated by the ancient origin and asexual evolution with many selective sweeps that have eliminated intermediate lineages. Before any conclusion can be drawn on speciation and population divergence, it is important to note that very few fungi have been studied at the population level: only two studies use population genetics, and these are restricted to agricultural fields in Denmark and Switzerland.
I thank Eva Stukenbrock and Ylva Lekberg for valuable comments on the manuscript and the Danish Natural Science Research Council for financial support.