Plant and arbuscular mycorrhizal fungal diversity – are we looking at the relevant levels of diversity and are we using the right techniques?

A strong emphasis in modern ecology is placed on an experimental approach to understanding mechanisms that explain patterns actually observed in nature. Testing the relevant hypotheses experimentally can often draw on a wealth of descriptive ecological information that exists for a large number of plants and animals, much of which was probably collected before descriptive ecology became less fashionable. Thus the observation of patterns in nature typically precedes the experimental tests of which mechanisms might explain their existence. Experimental ecologists may often take such descriptive data for granted. However, for those of us working on organisms where we do not have the necessary information at hand about these patterns in nature, then testing the relevant hypotheses can be a bit like going fishing: before we start, we don't know whether we will get anything and even if we are successful, perhaps there is always the elusive ‘one that got away’ (the more important or the more ecologically relevant hypothesis that we did not think to test). Working on the ecology of arbuscular mycorrhizal fungi (AMF; Glomeromycota) falls into this latter category of organisms, where descriptive data on their community structure in different environments and in relation to the community structure of their host plants is sadly lacking. In this issue (pp. 493–504), Landis et al. present descriptive data showing the correlation in Wisconsin oak savannas between AMF species richness and plant community composition. This relationship should exist in natural communities if a number of experimental studies testing the relationship between these two factors were correct and were, indeed, testing ecologically relevant hypotheses. The study by Landis et al. also raises two issues that deserve some further discussion. The first of these is raised by the authors themselves and concerns the reconstruction of AMF species richness by using, controversially, the morphology of AMF spores, rather than molecular techniques to measure AMF species richness. The second concerns the ecologically relevant measure of AMF diversity: is it interspecific or intraspecific AMF diversity that is ecologically important? Few would doubt the ecological importance of arbuscular mycorrhizal fungi, given that they improve plant nutrition and growth and are found in symbiosis with the roots of the majority of land plants (Smith van der Heijden Bever, 2003), which also predicts that a change in the diversity of one symbiont should affect the diversity of the other. Experimental investigations also support this model (Bever, 2002). However, if the experimental ecologists have got it right, and the relationships between the diversity of plants and AMF are not obscured by other factors that were not considered, then we should see some correlations between the diversity of these two groups of organisms in natural ecosystems. It is therefore encouraging that this is exactly what Landis et al. show in their study. As the authors are fully aware, the study is based on correlations, meaning that one cannot say whether plant diversity directly affects fungal diversity or vice versa and it must be pointed out that both are also co-correlated with soil texture and soil nitrogen content. One aspect that is controversial regarding the study is the use of morphological techniques to measure AMF species richness rather than PCR-based molecular identification methods. AMF spores can be sieved from the soil and then classified into different morphological types. These can then either be counted in order to obtain their abundance and then used to reconstruct the AMF community structure or they can simply be recorded as present/absent in order to obtain a measure of AMF species richness (as favoured by Landis et al.). The problems with this approach are numerous. An AMF species could be present in low frequency but produce many spores. Another could be very abundant but hardly produce any spores. This is why Landis et al. do not favour this measurement for AMF abundance. However, what Landis et al. do not point out is that there are a number of other reasons why the method may be seriously flawed. First, the environment may affect sporulation. Thus, the comparison of AMF species richness in different environments may reflect differences in which, and how many, AMF species sporulate in the different environments, even though the same fungi could potentially be present in all environments. In this case, the existence of a correlation between these two factors would be a red herring in the search for patterns that match the experimental predictions. Second, AMF spores are notoriously difficult to identify when sieved directly from the soil and some AMF taxonomists only consider identification to be reliable when the fungi have first been established in pot culture, from which many replicate spores of the same fungus can be observed. Even then, it can take a skilled researcher many hours to correctly mount samples of one fungus and observe the spore wall structure for correct identification. Third, we now know from molecular based studies of AMF communities in plant roots that many AMF are present in roots, although spores of these fungi have never been found using the techniques proposed by Landis et al. One of the clearest examples of this comes from a study that was also published in New Phytologist. Clapp et al. (1995) used molecular techniques to show that the roots of one of the common understorey plants in a woodland in north Yorkshire, UK consistently contained AMF for which no spores had been found. There are several other studies that document similar findings. However, a more recent study provides even more worrying data for the proponents of the spore morphology-based methods. Rosendahl and Stukenbrock (2004) have sequenced AMF ribosomal DNA (rDNA) from plants in a Danish grassland community on a sand dune. Using a phylogenetic analysis, they found that the sequences fitted into 11 clades within the Glomeromycota. Of these 11 clades, only four matched to known AMF for which species have been described on the basis of their spore morphology. Furthermore, within the four clades of known AMF, very few of the spore morphotypes were actually found at the Danish grassland site that match to those sequences, although they are known from other locations. This study indicates that AMF species richness, based on spore counts, is probably only scratching the surface of what really occurs in plant roots. Landis et al. partly justify their chosen methodology because they claim that sporulation is an important part of the AMF life-cycle. The molecular data, however, indicate that some AMF species are obviously not sporulating very often, if at all, and the importance of sporulation may be much less than has previously been assumed. As Landis et al. point out, the molecular methods available for AMF identification are by no means problem-free. The most precise of the methods, to date, involves amplifying parts of the AMF ribosomal DNA from roots and then sequencing the DNA. This can be both costly and time consuming, given the diversity of sequences that can be found in the root system of one plant. But the real problem with all of the molecular methods is the amount of genetic variation within an AMF species and even within an AMF individual. It is well known that high sequence variation for a given region of DNA occurs in AMF, even within single spores (Sanders, 2002). For ecologists this is problematic because when two or more different AMF sequences are obtained from a root system we don't know whether those sequences originate from one, two or several different AMF individuals. Furthermore, without knowing the true extent of the genetic variation within an individual, among individuals of the same species and among different AMF species, it is very difficult to know exactly what this sequence diversity actually represents. The amount of genetic variation in an AMF spore or in an AMF species is in itself a controversial issue. The reconstruction of AMF phylogeny using rDNA sequences relies on the underlying assumption that the variation in rDNA is so low that distinguishing among AMF morphotypes with rDNA sequences is not problematic. However, the failure of so many attempts to make true AMF morphotype-specific or species-specific primers suggests that it is not quite so simple. Other studies, however, indicate that the diversity of rDNA sequences within a morphotype is so large that such identification on the basis of one or a few sequences could be very misleading (Rodriguez et al., 2004b). In fact, Rodriguez et al. (2004b) found that several isolates of two different morphotypes of AMF harboured extremely diverse rDNA sequences. However, they also found that the two morphotypes also shared some identical sequences. They claim that low sequence variation suggested by some researchers is simply due to the fact that they have not sequenced intensively enough to see some of the low frequency variants (Rodriguez et al., 2004a). Furthermore, the first quantification of genome-wide molecular variation among isolates of one species of an AMF shows extremely high genetic variation, even in a very small field (Koch et al., 2004). The study was not based on rDNA variation but on 250 different polymorphic markers. The issues about genetic variation in AMF are clearly not yet sufficiently resolved such that we know exactly what we are dealing with when we pull a few AMF rDNA sequences out of the roots of a plant. The study of within species genetic variation in AMF (Koch et al., 2004) highlights a final concern about studies that measure AMF diversity. Many of these studies, whichever method they employ, justify the ecological relevance of their study because of the experimentally demonstrated relationship between AMF species richness and plant diversity (van der Heijden et al., 1998). However, most of such experiments have used one AMF isolate (often originating from one spore) to represent an AMF species and have ignored the variation in plant growth that could be due to within-AMF species variation. The study of Koch et al. (2004) shows that there is very large genetic variation within an AMF species, even within a small field, and that variation in how the fungi grow also has a genetic basis. A recent study by Munkvold et al. (2004) has shown that different isolates of one AMF species can supply very different levels of phosphorus to the plant. Perhaps the relationship might exist between the number of genetically different AMF in the field and the diversity of plant species; the distinction as to whether this is due to variation among or within AMF species or morphotypes may be largely irrelevant. I would certainly want to know the answer to this before embarking on an extensive and probably very costly survey of AMF species or morphotype variation in the field. The methods employed by Landis et al. will not be able to tackle this question, but molecular methods based on quantification of molecular variation within the AMF community, when further refined, along with a better understanding of AMF genetics and levels of genetic variation, may provide the answers. The methods that are currently available for AMF identification in ecological studies are problematic. I remain sceptical regarding the validity of the occurrence of spores in the soil as a measure of AMF species richness. Despite my criticisms I was, however, struck by the following sentence of Landis et al. ‘However, we find it extremely hard to believe that the striking correlations among AMF community composition and species richness, plant community composition and richness, and environmental factors that are demonstrated in this paper by our use of spore morphology could be artefacts’. We may well find in the future that the correlations observed by Landis et al. between plant diversity and AMF species richness may not exist when we measure the true richness that occurs in roots. It may turn out that the results reflect changes in the numbers of AMF that sporulate with changes in plant diversity, and that the sporulating AMF mask completely different underlying levels of AMF diversity/richness of nonsporulating AMF. Indeed, the correlations observed by Landis et al. are unlikely to be artefacts and it will be exciting to find out in the future what is the underlying mechanism for their existence.