• Species of arbuscular mycorrhizal fungi (AMF) differ markedly in their improvement of plant nutrition and health. However, it is not yet possible to relate the diversity of an AMF community to its functional properties due to the lack of information on the functional diversity at each taxonomic level. This study investigates the inter- and intraspecific functional diversity of four Glomus species in relation to a phylogenetic analysis of large ribosomal subunit (LSU) sequences.
• Growth and P nutrition of cucumber (Cucumis sativus) associated with 24 different isolates of AMF were measured in a two-compartment system with a 33P-labelled root-free soil compartment.
• Intraspecific differences were found in plant growth response and the extension of the fungal mycelium into the root-free soil patch whereas length-specific P uptake of the hyphae remained rather constant within each AMF species. Hence, the length-specific P uptake differed according to species, whereas lower phylogenetic levels were required to match functional characteristics such as fungal growth pattern and plant growth promotion.
• The large intraspecific diversity observed for mycelium growth and improvement of P uptake means that AMF communities of low species diversity may still contain considerable functional heterogeneity.
Arbuscular mycorrhizal fungi (AMF) are globally distributed and one of the most abundant inhabitants of below-ground ecosystems. This widespread mutualistic relationship between Glomalean fungi and the majority of vascular plants is believed to date back to 600 Ma and was probably an important element in the establishment of plants on land (Redecker et al., 2000). Arbuscular mycorrhizas are an intimate association in which obligate biotrophic fungi supply inorganic nutrients to plants in return for photosynthates. Despite their abundance, only 154 species have been described within the Glomales (http://invam.caf.wvu.edu/fungi/taxonomy/nomenclature.htm).
Recent development of molecular tools have confirmed the spore morphology-based taxonomy, but have also revealed intraspecific variations in, e.g. the ribosomal subunit sequences (e.g. Vandenkoornhuyse & Leyval, 1998; Kjøller & Rosendahl, 2000; Clapp et al., 2001). The meaning of this genetic variation in an ecosystem perspective remains unclear, since surprisingly little is known about the functional diversity within these species.
Ecosystem studies using these molecular tools have provided information on the composition of natural AMF communities (Helgason et al., 1998), but tools are required for the translation of this diversity information into functional properties of the communities. Progress in this respect may only be expected from the ability to link functionally uniform AMF groups to taxonomic groups of AMF.
Laboratory experiments with single pairs of AMF and host plants generally show a mycorrhiza-induced enhancement of plant nutrient uptake and growth (Smith & Read, 1997), but it is well documented that the outcome strongly depends on both fungal and plant genotypes. Performance of a single host species depends on the particular AMF strain associated (Jakobsen et al., 1992a; Streitwolf-Engel et al., 2001) and similarly, performance of a single strain of AMF depends on the host plant (Helgason et al., 2002). Despite growing evidence of functional variation between fungi in terms of plant growth promotion and P uptake, no clear functional grouping has emerged.
The present study aimed to link taxonomy and function by investigating the inter- and intraspecific functional diversity of AMF in terms of fungal growth and promotion of plant growth and P uptake. Direct measurements of growth and P uptake by the AMF mycelium were facilitated by a two-compartment model system with a root-free compartment containing 33P-labelled soil. The functional diversity was compared to variation in the large ribosomal subunit (LSU).
Materials and Methods
Biological materials and growth medium
Cucumber (Cucumis sativus L.) cv. Aminex (F1 hybrid) plants were grown in an irradiated 1 : 1 mixture of moraine clay loam and sand (w/w) supplied with basal nutrients minus P (Pearson & Jakobsen, 1993). The mixture, hereafter referred to as ‘soil’, contained 11 µg g−1 extractable P (Olsen et al., 1954). The addition of P to a +P control treatment resulted in 22 µg g−1 extractable P. The study comprised 13 Glomus mosseae (Nicol. & Gerd.) Gerdemann & Trappe, five G. claroideum (Schenck & Smith) emend. Walker & Vestberg, four G. caledonium (Nicol. & Gerd.) Trappe & Gerd and two G. geosporum (Nicol. & Gerd.) Walker isolates, all originating from arable soils (Table 1). Fungal inoculum was dried soil containing fungal spores and colonized root pieces produced with Trifolium subterraneum L. as host plant in similar soil and growth conditions as used in the experiment. The inoculum strength was determined in a 3-wk bioassay with T. subterraneum.
Table 1. Isolates of Glomus mosseae, G. claroideum, G. caledonium and G. geosporum used in the experiment
A, annual crops; G, grassland; F, fallow; R, garden soil. Isolates from the same origin and vegetation type were isolated from the same field. Accession numbers ‘AF’ (Kjøller & Rosendahl, 2000) were submitted in GenBank, ‘Y’ (van Tuinen et al., 1998) and ‘AJ’ numbers were submitted to EMBL.
Plants were grown in 50 mm PVC tubing cross-pots with a single sidearm (Fig. 1). The root-free compartment (RFC) (sidearm) was separated from the root compartment (RC) by 25 µm mesh, which prevented root entry. In the RFC, a 1-cm unlabelled buffer layer (20 g) adjacent to the mesh prevented root hair access to the 60 g soil labelled with 5.6 kBq g−1 carrier-free PO4 as well as 50 µg P g−1. The RC had three layers, RC1-3. A bottom layer of 375 g soil (RC3), a middle layer adjacent to the RFC of 150 g of a 1 : 1 mixture of inoculum and soil (w/w, M-plants) or 150 g soil (non-mycorrhizal controls) (RC2), and a top layer of 170 g soil (RC1) (Fig. 1). Hence, 75 g inoculum per pot was used in all mycorrhizal treatments except for BEG84, BEG90 and BEG86 where, respectively, 47, 72 (limited inoculum availability) and 100 g (increased to obtain equal inoculum strength) inoculum was added.
Of 78 cross-pots included, 72 contained inoculum while six pots were left uninoculated to serve as non-mycorrhizal (NM) control pots. Three of these pots received additional P. Two pre-germinated seeds were planted per pot, and thinned to one plant per pot after emergence. Each treatment had three replicates. Pots were watered to weight, initially to 55% of the soil water holding capacity (WHC) and later to 65–75% of WHC according to plant needs.
Plants were grown in a growth chamber with a cycle of 16 h light (500 µmol m−2 s−1) at 21°C and 8 h dark at 16°C. At the beginning of the experiment pots were completely randomized and subsequently rearranged at each watering. A total of 100 mg additional NH4NO3-N in aqueous solution was provided as five supplies per pot during the 4-wk growth period.
Harvest and analyses
F. wts and d. wts of shoots were determined at harvest. Roots were washed thoroughly, excess moisture was removed and f. wt was determined. Two subsamples were weighed, one of which was used for determination of root d. wt while the other was cleared in 10% KOH and stained in Trypan Blue by a modification of the method of Phillips & Hayman (1970), omitting phenol from the reagents and HCl from the rinse. Total root length was calculated according to Newman (1966) and the proportion of the root length that had been colonized by the AMF isolates was assessed simultaneously. Total P content of shoots and roots was measured using the molybdate blue method on a Technicon Autoanalyser II (Technicon Autoanalyzers, Analytical Instrument Recycle, Inc., Golden CO, USA) after acid digestion (Murphy & Riley, 1962). 33P contents were measured on the same extracts on a Packard TR 1900 liquid scintillation counter (Packard Bioscience Company, Meriden, CT, USA). Hyphal lengths in dried 2 g samples of radiolabelled soil as well as soil from RC2 were measured by a grid-line-intersect method (Jakobsen et al., 1992b). All soil samples for hyphal measurements were stored frozen until extraction of hyphae.
DNA from two or three single spores from each isolate was extracted and the D1 and D2 domain of the large ribosomal subunit rDNA was amplified by PCR using the primer combination LSU0061 (Kjøller & Rosendahl, 2000) and NDL22 (van Tuinen et al., 1998). Subsequently the PCR products were sequenced using the same primers and used for phylogenetic analysis. The aligned dataset was analysed using Bayesian analysis (MrBayes v.3.0, Huelsenbeck & Ronquist, 2001) with one cold and three heated chains. The analysis included 1 000 000 generations of which every 50th tree was sampled. The trees from the first 150 000 generations were excluded to ensure that a stable likelihood had been reached. The resulting 17 000 trees were imported into paup* and a majority rule consensus tree was constructed (Fig. 2).
Statistics and data analysis
One replicate plant of both the BEG160 and BOL3 treatments failed to establish and were excluded from the dataset. Data were analysed in S-PLUS 6.2 for Windows, http://www.insightful.com. Differences between species were analysed using a linear mixed model with isolates as a random factor nested within the species. Differences between species were tested using multiple comparisons by the Bonferroni method. Variance components calculated from the nested model were used to calculate the relative contribution of the variation between species and between isolates. Species-specific differences in slopes for the length specific hyphal P uptake were determined by testing the best fit of two linear models with or without the effect of the species. The overall effect of mycorrhiza was compared to the two non-mycorrhizal controls by treating all mycorrhizal treatments as one group in a one-way anova.
Results and Discussion
This study demonstrates that AMF can vary considerably with respect to a wide range of functional characteristics, even within the species. However, the degree of variation depends strongly on the functional variable in question. This is exemplified by our detection of large intraspecific differences in the fungal mycelium growth pattern and promotion of plant P uptake while the hyphal length-specific P uptake was much more constant. The results are presented and discussed in detail below.
Phylogenetic relationship of the 24 AMF isolates
The phylogenetic relationship of the 24 AMF isolates as derived from ribosomal LSU sequences is shown in Fig. 2. The appearance of four well-defined species groups is consistent with our selection of isolates of Glomus mosseae (13), G. claroideum (5), G. caledonium (4) and G. geosporum (2). Within G. mosseae, one distinct isolate group consisting of BOL3 and V296 sq1 was identified. In this case however, only one of the two sequences representing V296 is located within this group. Further validation of this group would therefore require analysis of other molecular markers.
Growth of plants and fungi
Overall, plant growth was improved significantly by the formation of mycorrhiza, but did not vary at the level of fungal species (P = 0.09) (Fig. 3a). Differences in shoot growth were also reflected in root growth. However, there were prominent AM fungal intraspecific differences in plant growth promotion which was also reflected by the fact that variance within the species accounted for 70% of the total variance. The presence of intraspecific differences in mycorrhizal plant growth response has, to our knowledge, only been reported a few times in studies which included two G. mosseae and two G. caledonium isolates (Carling & Brown, 1980) or three G. mosseae isolates (Stahl et al., 1990). Other studies (Jakobsen et al., 1992a; Smith et al., 2000, 2004; Drew et al., 2003) have discussed interspecific differences, but the present data clearly shows that it is not possible to generalize from a single isolate to the entire species.
All fungal isolates established well in the roots and colonized more than 60% of the root length for all except the two G. geosporum isolates and the two G. caledonium isolates, BEG86 and V162 (Fig. 3b). In spite of this relatively uniform intraspecific root colonization, the AMF isolates differed markedly in the mycelium growth into a root-free patch containing 33P-labelled soil (Fig. 4c). These intraspecific differences accounted for 70% of the total variance. Koch et al. (2004) reported similar intraspecific differences in an experiment in which several G. intraradices isolates grown in root organ cultures differed in hyphal length and growth rate.
Plant growth was not related to either root colonization (R2 < 0.13 for all fungi) or mycelium proliferation in the root-free compartment (R2 < 0.03 except for G. geosporum which however, showed no significant linear correlation (P = 0.13). Hyphal lengths in the root compartment, which on average were lower and more uniform (data not shown), did not explain these differences among isolates in shoot growth (R2 < 0.18 for all fungi) or hyphal lengths in the root-free compartment (R2 < 0.24) for all except G. claroideum (R2 = 0.49, P < 0.01). Hyphal lengths of G. claroideum increased in the root compartment with increasing hyphal length in the root-free compartment. Unfortunately, the measurements from the root compartment can serve as guidelines only, since it was not possible to distinguish between hyphae supplied with the inoculum and hyphae developed during the experiment.
Plant P content and AM fungus-mediated uptake of 33P from root-free soil
Shoot P content differed markedly between isolates (Fig. 4a) but in close correspondence with shoot growth (Fig. 5). This confirms that soil P was the primary growth-limiting factor and that additional P channelled from soil to plant via the fungus resulted in improved growth. Only shoot P data are discussed since fungal and plant P in roots cannot be separated with the current methodology. Similarly to shoot d. wt, shoot P content was not correlated to the proportion of colonized root for any of the fungi. Shoot P content however, correlated to the absolute colonized root length (R2 = 0.54, P << 0.0001; R2 = 0.69, P < 0.0001; R2 = 0.68, P < 0.001) as well as to the total root length (R2 = 0.64, P < 0.0001; R2 = 0.59, P < 0.001; R2 = 0.46, P < 0.02) for the three fungi represented by most isolates G. mosseae, G. claroideum and G. caledonium, respectively. Unfortunately, the relative importance of the two variables cannot be determined due to their internal correlation. Similar relationships were not observed for G. geosporum, which was only represented by two isolates.
Uptake of 33P into the plants from the root-free soil patches also varied markedly between isolates, but in due accordance with the hyphal length density of the mycelium (Fig. 4c). This correlation, reflecting the length specific hyphal P uptake, was particularly strong and species-specific for G. mosseae and G. caledonium (Fig. 6). A similar correlation was previously observed in a study with 16 isolates of G. claroideum (R2 = 0.62; P << 0.0001 (I. Jakobsen, unpublished data). The length specific hyphal P uptake therefore appears to be a rather robust variable within the fungal species. To our knowledge, this is the first study based on more than one isolate per species, to report a direct relationship between the fungal P uptake and the hyphal length in the soil.
Although fungal P transport from the 33P-labelled soil corresponded well to the hyphal spread, the total shoot P content was not related to fungal 33P transport from the root-free soil for G. mosseae, the fungus represented by most isolates. Furthermore, when total P and 33P uptake is compared (Fig. 4a,b) for each isolate, shoot P uptake may be quite high even if 33P transport is very low, e.g. for both isolates of G. geosporum or the G. mosseae isolates BEG29 and BEG84. This implies that a spatial component is required to adequately explain P uptake for symbiotic associations with seemingly similar uptake efficiencies. These isolates grew less than 1 cm away from the root, but still provided much P to the plant, presumably from within this 1 cm zone. The high P transport of these fungi could not be explained by increased hyphal length in the root compartment adjacent to the root-free compartment (RC2), which did not differ from fungi with a more extensive exploitation of the root-free compartment. Such differences in hyphal growth pattern were previously suggested for two species, one of which took up much P from the root-free soil compartment, whilst the other gained its P predominantly from soil layers near the root (Smith et al., 2000). The present experiment suggests an overriding impact of the isolate, the variance of which could explain 70% of the total variance in hyphal length. In the case of shoot growth response and shoot P content the isolate explained 70% and 62% of the variation, respectively.
Linking phylogenetic relationships to functional characteristics
Although the species groups were well defined by phylogenetic analyses of LSU rDNA sequences, they matched the observed functional differences only in some cases. Three of the important functional characteristics; hyphal length, shoot growth response and shoot P content varied more within the species than between species. Thus, depending on the function in question, subsets of phylogenetical levels based on simultaneous analysis of several molecular markers are required to explain these differences.
The intraspecific differences in hyphal growth patterns observed in this experiment suggest that the hypothesized functional complementarity of AMF (Read, 1998; Koide, 2000) could exist not only between, but also within AMF species. This would seem to strengthen the robustness in function of AMF communities in ecosystems of otherwise low species diversity. Our observed functional diversity, even at the intraspecific level, suggests that the AMF of individual communities could be functionally very different, even in arable soils where the species diversity of AMF is lower than in roots from undisturbed ecosystems (Daniell et al., 2001; Husband et al., 2002; Vandenkoornhuyse et al., 2002). We hypothesize that a high intraspecific functional diversity will compensate low species diversity in terms of efficient nutrient capture by the AMF. This hypothesis would of course need to be confirmed in an experiment using isolates sampled from the same community. Our results call for a direct comparison of disturbed and undisturbed ecosystems with respect to the level of intraspecific functional diversity of the commonly found species.
Our results have important implications for future studies on the mechanisms involved in functioning of the AMF–plant symbioses. While it has been widely agreed that the choice of species is important for the outcome of such studies (Jakobsen et al., 1992a; Helgason et al., 2002; van der Heijden et al., 2003), our results, in concert with Koch et al. (2004), have elucidated that the choice of AMF isolate within the species can be just as important for the outcome of a study. It would seem that the linkage between level of phylogeny and AMF functioning needs to be defined in accordance with the function in question. Hence, the length specific hyphal P uptake seems to be constant within the species as presently defined, whereas subsets of phylogenetical levels are required to match functional characteristics such as fungal growth pattern, fungal provision of P to the plant and plant growth promotion by the AMF.
We thank Anette Olsen and Anne Olsen for their excellent technical assistance, Tobias G. Frøslev for help with the phylogenetic analysis and Hanne Østergård and Sven Jesper Knudsen for advice and assistance with the statistical analyses.