SEARCH

SEARCH BY CITATION

Keywords:

  • amplified fragment length polymorphism (AFLP);
  • arbuscular mycorrhizal fungi (AMF);
  • environmental heterogeneity;
  • genotype-by-environment interaction;
  • Glomus intraradices;
  • resource acquisition;
  • selection;
  • specificity of response

Summary

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • • 
    Arbuscular mycorrhizal fungi (AMF) are plant symbionts that improve floristic diversity and ecosystem productivity. Many AMF species are generalists with wide host ranges. Arbuscular mycorrhizal fungi individuals are heterokaryotic, and AMF populations are genetically diverse. Populations of AMF harbor two levels of genetic diversity on which selection can act, namely among individuals and within individuals. Whether environmental factors alter genetic diversity within populations is still unknown.
  • • 
    Here, we measured genetic changes and changes in fitness-related traits of genetically distinct AMF individuals from one field, grown with different concentrations of available phosphate or different host species.
  • • 
    We found significant genotype-by-environment interactions for AMF fitness traits in response to these treatments. Host identity had a strong effect on the fitness of different AMF, unearthing a specificity of response within Glomus intraradices. Arbuscular mycorrhizal fungi individuals grown in novel environments consistently showed a reduced presence of polymorphic genetic markers, providing some evidence for host or phosphate-induced genetic change in AMF.
  • • 
    Given that AMF individuals can form extensive hyphal networks colonizing different hosts simultaneously, contrasting habitats or soil properties may lead to evolution in the population. Local selection may alter the structure of AMF populations and maintain genetic diversity, potentially even within the hyphal network of one fungus.

Introduction

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Arbuscular mycorrhizal fungi (AMF) are obligate biotrophs and form symbioses with roots of the majority of plant species (Smith & Read, 1997). This symbiosis is of fundamental ecological and agricultural importance because the fungi improve plant nutrient acquisition, especially phosphate availability, and promote floristic diversity and ecosystem productivity (Grime et al., 1987; Lu & Koide, 1994; Harrison, 1997; van der Heijden et al., 1998b; Vogelsang et al., 2006). Arbuscular mycorrhizal fungi are coenocytic fungi, lacking any form of septation. This lack of compartmentalization results in the absence of distinction between soma and germ line, where any mutation is potentially transmitted to vegetative offspring and is therefore also potentially evolutionarily relevant (Pontecorvo, 1946; Buss, 1987). Arbuscular mycorrhizal fungi are also known to harbor considerable within-isolate genetic variation, where an isolate is defined as a culture obtained from a single spore (Pringle et al., 2000; Clapp et al., 2001; Kuhn et al., 2001; Pawlowska & Taylor, 2004; Rodriguez et al., 2004). Some of this variation is explained by genetic differences among co-occurring nuclei within isolates of AMF (Kuhn et al., 2001; Hijri & Sanders, 2005). However, the stability and dynamics of genetically different nuclei within AMF isolates is still unknown (Pawlowska, 2005). A portion of this within-isolate genetic diversity could be nonneutral (i.e. either detrimental or beneficial in a given environment) and therefore of ecological importance.

Much of the past work on AMF ecology and genetics has focused on functional diversity and differences among AMF morpho-species (Sanders & Fitter, 1992; Bever, 1994; Bever et al., 1996; van der Heijden et al., 1998b; Pringle & Bever, 2002). A recent study showed that the relatedness of AMF species alters AMF community assembly in a manner where species richness is highest among unrelated species, suggesting that competition is stronger among closely related species (Maherali & Klironomos, 2007). These results support the view that the phylogenetic distance among genotypes is related to their functional divergence and that more closely related species are increasingly likely to forage for the same resources and occupy similar niches (Comins et al., 1980; Taylor, 1992). However, while competition occurs both within and among species, selection is the strongest within populations at the species level, ultimately altering relative genotype frequencies. Analyses of AMF populations have revealed that they are diverse both phenotypically and genetically (Bever & Morton, 1999; Pringle et al., 2000; Koch et al., 2004; Pawlowska & Taylor, 2004; Stukenbrock & Rosendahl, 2005; Croll et al., 2008). Thus, within an AMF population, diversity occurs at two levels, namely between isolates and within isolates. Nonneutral heritable genetic variation at any of these two levels could alter the fitness of the given unit (isolates, nuclei) in a given environment and thus be selectively important. Between-isolate selection occurs if genetically distinct isolates have different fitness (i.e. spore or hyphal production) in contrasting environments (Pringle & Taylor, 2002). Selection could occur within AMF isolates if genetically different nuclei confer differing effects on fitness. Selection at either of these two levels could result in changes in the relative frequencies of genetically distinct isolates (or in different nuclei within an isolate) in different subpopulations, leading to a change in their relative frequency in selectively contrasting environments. Selection acting at either of these two levels could be ecologically important, because single AMF isolates form extensive interconnected hyphal networks that can simultaneously colonize different host plant species (Giovannetti et al., 2004) and potentially grow in contrasting abiotic conditions.

To date little is known about the mechanisms maintaining AMF diversity at the population level, and virtually nothing is known about the dynamics of nuclei within isolates. However, which factors affect AMF species diversity has been extensively addressed. Co-occurring AMF show differences in phenologies, growth traits and strong differential effects on plant growth (Streitwolf-Engel et al., 1997; van der Heijden et al., 1998a), and vice versa, host plant species identity affects the growth and spore production of different AMF (Helgason et al., 2002). However, it is unclear to what degree these findings actually demonstrate ecological differences among AMF species because intraspecific differences in symbiotic efficiency and growth traits have been documented not only among isolates from distant geographic locations (Hart & Reader, 2002; Munkvold et al., 2004), but also among isolates originating from the same field (Gamper et al., 2005; Koch et al., 2006). Complex among-species interactions and fitness feedbacks have been proposed as mechanisms for mediating the coexistence of plant and AMF species diversity in natural ecosystems (Bever, 2002; Klironomos, 2002; Hart et al., 2003), and similar mechanisms could also be important in mediating the co-existence of different genotypes within AMF populations, within species.

Despite the increasing awareness of the different levels at which genetic variation in AMF occurs, the question of which mechanisms might enhance or maintain genetic diversity within AMF populations and isolates has not been experimentally addressed. The main reason for this is the difficulty in sampling, isolating and maintaining cultures of an AMF population from the field. Using several AMF isolates of one morphospecies that all originate from the same field and that were maintained in clean and standardized conditions for several years (Koch et al., 2004, 2006), we performed two experiments to test whether phosphate availability and host species identity result in significant isolate-by-environment interactions for AMF fitness traits and detectable genetic change within AMF isolates. We chose to change phosphate availability and host species identity because they are both important factors in the AMF symbiosis (Koide & Elliott, 1989; Eom et al., 2000; Bücking & Shachar-Hill, 2005; Cavagnaro et al., 2005). We found significant isolate-by-environment interactions in fungal fitness as well as some evidence for genotypic divergence within AMF isolates when grown in contrasting environments.

Materials and Methods

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Isolation and culture of an arbuscular mycorrhizal field population

Five single spore cultures of Glomus intraradices (Schenck & Smith; hence called isolates) from one agricultural field site in Tänikon, Switzerland, were used to establish axenic cultures (Anken et al., 2004; Koch et al., 2004). The cultures were maintained under standardized conditions in an axenic culture system on standard M growth medium with a clone of transformed Daucus carota L. roots (Becard & Fortin, 1988). The isolates used in this work were previously shown to differ phenotypically and genotypically (Koch et al., 2004, 2006). The nomenclature of the isolates used in both experiments follows Koch et al. (2004), where letters stand for different field plots and numbers indicate different single spore lines within a plot. Isolates A4, B3, C3 and D1 were used in Expt 1. Because of poor growth at the start of Expt 1, isolate C2 was included one generation later than the other four, and no phenotypic measurements were made on this isolate. However, this isolate was maintained, along with the other isolates, to extract DNA for molecular analysis. In Expt 2, isolates B3, C2 and C3 were used. In both experiments, two-compartment plates were used to grow clean fungal material that was extracted as described as in Koch et al. (2004) and subsequently used for DNA extraction.

Expt 1: Effect of decreased phosphate availability on the growth and genotypes of four fungal isolates and transformed D. carota host roots

This experiment aimed to test how the growth and genotype of the fungal isolates is affected when phosphate availability is reduced. Four isolates were vegetatively propagated with a clone of transformed D. carota roots on standard M medium (Becard & Fortin, 1988) as well as on two media where phosphate availability was reduced compared with M medium. The concentration of potassium phosphate in the growth medium was reduced to a final concentration of 1% and 0.1%, respectively, compared with that in standard M medium. The reduced amount of potassium was compensated for by adding the equivalent amount of potassium (in the form of KCl) to the medium (Koch et al., 2006). The treatments were abbreviated as P100 (standard M medium), P1 and P0.1 (for the two treatments with reduced phosphate availability). At the start of the experiment, for each isolate and for uncolonized, namely nonmycorrhizal, D. carota roots (controls), 20 parental plates (15 wk of age) were used to inoculate 15 new replicates (9-cm Petri dishes) of each of the three treatments. This gave a total of 45 plates for each isolate and for nonmycorrhizal roots (controls), all of which were subcultured on the three treatments for four clonal generations. One clonal generation consisted of a growth period of 15 wk, starting from a piece of inoculum of approx. 2 cm2 (containing medium, colonized roots, hyphae and spores) from a 15-wk-old plate. Root, hyphal and spore densities of the four isolates and controls were measured in all treatments at the end of the fourth clonal generation only. This was performed to reduce carry-over effects (from the P100 medium in which they were maintained for over 3 yr), and to allow for the occurrence of potential phenotypic and genotypic changes as a response to growth in the new medium. The growth traits of 15 replicate plates of each isolate and control roots were measured as previously described (Koch et al., 2004, 2006). Contaminated plates or plates with no AMF hyphal growth were discarded, reducing the total number of plates from 225 to 214 (42 control plates with uncolonized roots and 172 plates with AMF-colonized roots).

After isolates A4, B3, C3 and D1 had been subcultured for four (and isolate C2 for three) generations in P100 and P0.1 media, they were transferred to 12 two-compartment plates containing the corresponding P100 or P0.1 medium in order to produce hyphae and spores that could be harvested without carrot DNA. DNA from isolates C3 and D1 could not be used for molecular analyses because of poor DNA yields.

Expt 2: effect of host species identity on the growth and genotype of three fungal isolates

A second experiment was performed in parallel to Expt 1 to test how the growth and genotype of three fungal isolates (B3, C2 and C3) were affected by three different host species. We used G. intraradices isolates that had been maintained on D. carota roots for 3 yr before the experiment. We established new axenic cultures on D. carota roots as well as root clones of two other transformed hosts: Medicago truncatula Gaertn. (strong-spined medic) and Solanum tuberosum L. (potato). These clones were obtained from the research laboratories of Marcel Bucher (ETH Zürich) and G. Bécard (University of Toulouse). We established three lines for each isolate, on each host treatment. Each line was started with one randomly chosen spore cluster containing 20–50 spores (Fig. 1). Each of nine spore clusters per isolate was randomly chosen and harvested from one well-grown 15-wk-old two-compartment plate and individually transferred onto a new plate containing standard M medium (Becard & Fortin, 1988) and nonmycorrhizal roots of one of the three hosts. In that way, 27 lines were obtained, each of which was subsequently subcultured for four clonal generations with roots of the same host plant species (Fig. 1). Root, hyphal and spore densities were measured on each plate at the end of the fourth generation. Unlike Expt 1, root density was measured using the same technique as used for measuring hyphal density, at eight randomly chosen locations on each plate. Contaminated plates, or plates with no fungal growth, were discarded, leaving a total of 264 plates with a minimum of six (of the 12) replicate plates per line. The transformed D. carota roots were the same clone as used in Expt 1. All isolates and lines were maintained on standard M medium.

image

Figure 1.  Design of Expt 2 (For more details see the text). AFLP, amplified fragment length polymorphism.

Download figure to PowerPoint

Each of the three lines of the isolates B3, C2 and C3 growing on S. tuberosum and D. carota roots were used to inoculate nine two-compartment plates at the end of the seventh clonal generation. Isolate C2 grew only poorly on the two-compartment plates and there was not enough fungal material for DNA extraction. Because all three isolates grew poorly on M. truncatula, fungal material growing with this host species was not used for DNA extraction.

Molecular methods

Fungal material and DNA extraction  Hyphae and spores of each isolate were extracted from the root-free compartment as described by Koch et al. (2004). In both experiments, four to eight two-compartment plates of a treatment combination contained sufficient fungal material to be used for DNA extraction. The mycelium and hyphal contents of all plates were pooled; the mycelium was then extracted as described in Koch et al. (2004). In Expt 1, DNA was extracted from two treatments (P100 and P0.1) for each of three isolates. In Expt 2, DNA was extracted from all three lines from each of two host plants (carrot and potato) for isolate C3 and from two lines from each of the same two host species for isolate B3. Where there was sufficient fungal material, the bulk, consisting of extracted mycelium and spores, was divided into two equal parts to allow replication of the DNA extractions. There was only enough fungal material to allow replication of DNA extractions in two lines of isolate C3 growing on D. carota in Expt 2. There was only one DNA extraction for each treatment combination for all other samples in Expts 1 and 2. In both experiments, and for each sample, freshly extracted fungal material was separately dried overnight at 48°C and ground into a fine powder using a Retsch MM300 machine (Qiagen). The DNA was extracted using a modified version of the Cenis method for fungal DNA extraction (Cenis, 1992) with an additional step of a 1:1 dilution with a solution of chloroform : isoamyl alcohol (24:1, v/v) before the final precipitation, to remove any remaining impurities.

Amplified fragment length polymorphism  Amplified fragment length polymorphism (AFLP) fingerprinting was used to score genetic variation using 100–200 ng of DNA for each replicate digestion. Because of poor yield, replicate DNA extractions were only obtained for two samples in Expt 2 (lines D2 and D3 of isolate C3 grown on D. carota). However, in both experiments we replicated digestions from the same DNA extraction in all samples where it was only possible to obtain one DNA extraction. In both experiments, DNA samples were standardized to the same DNA concentration for each isolate and treatment using Picogreen reagents (Molecular Probes, Inc., Eugene, OR, USA). Amplified fragment length polymorphism analysis was performed using the method of Vos et al. (1995), except for the following: digestion was performed with 20 µl of DNA template (adjusted with nuclease-free water (Sigma) to 200 ng, or 100 ng where the yield was lower) after adding 5 µl of mastermix containing 5 U of MseI, 5 U of EcoRI, as well as 0.25 µl of BSA and 2.5 µl of EcoRI buffer (New England Biolabs, Ipswich, MA, USA), and an incubation time of 2 h at 37°C. The ligation was performed in 30.5-µl reaction volumes with an adaptor solution, as in Vos et al. (1995), using T4 ligase and T4 ligase buffer (New England Biolabs) and incubation overnight at 16°C, and then diluted 1:1 with 1× TE (10 mm Tris/HCl 0.1 mm EDTA). The pre-amplification reaction contained 2.5 µl of the diluted ligation product, 2.5 µl of 10× PCR buffer plus Mg (Qbiogene), 0.02 of each desoxynucleotide, 0.324 µm E-0 primer, 0.36 µm M-0 primer and 2.5 U Qbiogene Taq DNA polymerase in a total volume of 25 µl (Table 1). The reaction product was diluted 1:10 with TE. Selective amplification was performed in a 10 µl mix containing 2.5 µl of the diluted pre-amplification product, 1 µl of 10 × PCR buffer plus Mg (Qbiogene, Santa Ana, CA, USA), 0.025 of each desoxynucleotide, 0.3 µm EcoRI fluorescence-labeled selective primer, 0.1125 µm MseI selective primer and 0.5 U Qbiogene Taq DNA polymerase. Twelve different primer pairs were used for Expt 1 and 16 different primer pairs were used for Expt 2 (Table 1). All PCR protocols were performed as described in Vos et al. (1995). The selective PCR products of different fluorescent probes were multiplexed, and 13.3 µl of HiDi formamide and 0.2 µl of ROX-500 size-standard (Applied Biosystems, Inc., Foster City, CA, USA) were added to 1.5 µl of multiplexed products. After a quick heat denaturation of 1 min at 95°C, fragments were run on an automated ABI-3100 Genetic Analyzer™. Amplified fragment length polymorphism fragments were manually scored using GeneMapper 3.7™ software (Applied Biosystems, Inc.).

Table 1.  Adapters and primers used for amplified fragment length polymorphism (AFLP) analyses in Expts 1 and 2
  1. (a) Sequences of the adapters and of the pre-selective primer. (b) Primer combinations used to track potential changes in the genotype in Expt 1. (c) Primer combinations used to track potential changes in the genotype in Expt 2 with isolate C3. (d) Primer pairs used in Expt 2 with isolate B3. The polymorphic primer pairs are in bold.

(a)(b)
   EcoRI-GTG MseI-TA
EcoRI-ad1CTC GTA GAC TGC GTA CC EcoRI-TT MseI-TA
EcoRI-ad2AAT TGG TAC GCA GTC EcoRI-AAG MseI-TA
MseI-ad1GAC GAT GAG TCC TGA G EcoRI-AA MseI-TC
MseI-ad2TAC TCA GGA CTC AT EcoRI-AAC MseI-TC
   EcoRI-AA MseI-TG
EcoRI-0GAC TGC GTA CCA ATT C EcoRI-TT MseI-TG
MseI-0GAT GAG TCC TGA GTA A EcoRI-TT MseI-AT
   EcoRI-AA MseI-AT
   EcoRI-GTG MseI-TT
(c)  
EcoRI-TC MseI-TT  
EcoRI-AAC MseI-TT  
EcoRI-AA MseI-TT  
EcoRI-GTG MseI-TT   
EcoRI-AAG MseI-TG  
EcoRI-GTG MseI-TG   
EcoRI-AAC MseI-TC   
EcoRI-TA MseI-TC (d)
EcoRI-GTG MseI-GC EcoRI-AA MseI-TC
EcoRI-TT MseI-GC EcoRI-TT MseI-TC
EcoRI-TA MseI-GC EcoRI-AA MseI-TG
EcoRI-AA MseI-GC EcoRI-AAC MseI-TG
EcoRI-TT MseI-AT EcoRI-AAG MseI-TG
EcoRI-AAC MseI-AT EcoRI-GTG MseI-TA
EcoRI-TA MseI-AT EcoRI-GTG MseI-TT
EcoRI-AA MseI-CT EcoRI-AAG MseI-TT

Only AFLP fragments with lengths between 50 and 500 bp were analyzed. Loci were scored if at least one peak of all replicates was greater than 200 relative fluorescence units (RFU). Alleles were scored as present if they were above 100 RFU. Alleles with a peak height between 50 and 100 RFU were scored as unknown state. Under 50 RFU, the alleles were scored as absent. Unknown replicates were changed to the consensus score of 0 or 1 given that the second replicate was either 0 or 1, but were left as unknown if the second replicate score was also unknown.

Statistical analysis

ANOVA was performed with data from both experiments on the traits root density, hyphal density and spore density. To test if there was an overall mycorrhizal inoculation effect on root density in Expt 1, data of all isolates were pooled for each of the three phosphorous treatments and compared with the nonmycorrhizal controls. A crossed two-way ANOVA with the main fixed factors phosphorous treatment (three levels) and AMF colonization (two levels) was calculated, using the measurements on root density from a total of 214 replicate plates. To test whether there was a significant AMF isolate, phosphate treatment effect or isolate-by-phosphate treatment interaction (IxPT) on fungal growth, the nonmycorrhizal treatment was removed from the data set. Multivariate analysis of variance (MANOVA) was performed simultaneously on the two fungal growth traits, hyphal density and spore density, in a crossed model with the main factors AMF isolates (four levels) and phosphorous treatments (three levels) to test for overall isolate, phosphorous treatment and isolate-by-phosphate interaction effects. Subsequently, univariate ANOVA on the two fungal growth traits and root density were separately calculated using a crossed two-way mixed-model with AMF isolate as random factor and phosphorous treatment as fixed factor. To meet the requirements of the statistical tests, the following transformations of the fungal growth traits were used: Box-Cox (hyphal density) and (spore density + 1)^0.5 (Sokal & Rohlf, 2000). We performed a Z-test (Zar, 1984) to determine whether the Pearson's correlation coefficient between hyphal density and root density in the P0.1 treatment was significantly higher than in the P100 and in the P1 treatments.

In Expt 2, data analysis was similar to Expt 1 except for the analyses that included a nonmycorrhizal treatment. Each of the 27 lines was considered as a replicate. For each growth trait, the average value of the replicate plates of each of the 27 lines was calculated. This reduced the data set to three replicates (the three lines) for each isolate–host combination. Data (spore and hyphal densities) were analyzed by MANOVA and ANOVA (the same two fungal traits and root density) using a mixed model with the two crossed main factors: AMF isolate (random, three levels); and host species (fixed, three levels). To meet the requirements of the statistical tests, the means of the lines were transformed as: Log10(hyphal density), and (spore density)^0.5 (Sokal & Rohlf, 2000).

To test whether the observed AFLP profiles of AM isolates were affected by growth in contrasting environments, we analyzed polymorphic AFLP loci, which represent DNA fragments of any given size that are detected in some samples but are absent in other samples. Such differences in the AFLP fingerprint could arise by genetic changes in a given fragment (e.g. through mutations, deletions or insertions that occurred at a cutting site of the restriction enzyme or at the binding site of primers, or a change in the frequency of a given fragment owing to a multigenomic state) or to differences in DNA-extraction quality. We performed a permutation test to assess whether, over both experiments, the observed changes in AFLP fingerprints within isolates were possibly caused by genetic changes as a result of growth in contrasting environmental growth conditions. Isolate B3 did not show any AFLP polymorphism in Expt 1 and was therefore not included in this test. Thus, we considered the AFLP fingerprints from the isolates A4 and C2 in Expt 1, and from B3 and C3 in Expt 2. Under the null hypothesis that no genetic change within each isolate had occurred, all observed polymorphisms are expected to be a result of methodological (i.e. random) error and are expected to occur irrespective of the treatments. The alternative hypothesis is that polymorphisms did not occur by methodological errors and that the isolates had genetically diverged. In that case, contrasting environments may act selectively and could induce genetic changes within AMF individuals that consistently occur more often in certain treatments. In Expt 2, where we set up replicate lines for each isolate and host species, it is also possible that lines diverged from each other as a result of drift (i.e. stochastic frequency changes of genetically different nuclei that may co-exist within AMF isolates). However, because the goal of this study was to test for consistent treatment effects and not drift, we used spore clusters to minimize such stochastic effects at the start of Expt 2.

For each DNA extraction, we scored the number of polymorphic AFLP loci present. The only ‘unknown’ locus was scored as 0.5, and for the only AFLP error (polymorphism between two AFLP replicates of the same DNA extraction) one of the two profiles was randomly chosen for each permutation run. The data for all four isolates were then permutated 99 999 times in the following manner. In each permutation run, the observed data for each isolate were randomly permutated between the two treatments (old and new environment). We then assessed, for each isolate, whether the total number of scored loci in the new treatment was equal to or larger than that observed in the experiment. The number of runs in which all isolates of the permutated data set had a higher or equal number of polymorphic AFLP markers in the old environment, compared with the observed data, were counted. This number, divided by the number of permutation runs, estimates the overall probability (P-value) by which the apparent loss of loci in new environments could be explained by pure chance. We then re-performed similar analyses, with the difference that only one randomly chosen DNA extraction for isolate C3 in Expt 2 was used for the two lines, where we had AFLP fingerprints from replicate DNA extractions. This latter analysis avoids potential pseudo-replication at the line level because lines may have also diverged from each other within the same treatment.

All these analyses were performed using the programs JMP 5.0 (SAS Institute Inc., Cary, NC, USA) and R version 2.0.1 (R Development Core Team, 2003, http://www.R-project.org).

Results

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Effects of decreased phosphate availability and three different host species on the clonal growth of fungal isolates and transformed D. carota roots

At the end of the fourth growth period, treatments with decreased phosphate availability had significantly reduced root density (F2,208 = 9.39, P < 0.0001, Fig. 2a). Overall, the density of colonized roots did not differ significantly from the density of nonmycorrhizal control roots (F1,208 = 2.02, P = 0.16, Fig. 2a). However, there was a significant fungal colonization by phosphate treatment interaction (F2,208 = 3.98, P = 0.0201). In the treatment with the lowest phosphate availability, colonized roots grew to a significantly lower extent than nonmycorrhizal roots (Fig. 2a). Although there was no significant main-isolate effect on root density, there was a significant fungal isolate by phosphate treatment interaction (Fig. 2b, Table 2a). This means that roots forming a symbiosis with different AMF isolates were not affected in the same way by phosphorus availability. While in the P1 treatment the roots growing with D1 and B3 were the longest, they were the shortest in the P0.1 treatment (Fig. 2b). Root and hyphal density were negatively correlated in the P100 and P1 treatments (RPearson = −0.0131, P > 0.92 and RPearson = −0.3544, P < 0.006, respectively). In the P0.1 medium the correlation was positive (RPearson = 0.46, P < 0.0001) and was significantly higher than the correlation in the standard medium (Zcritical value = 3.343, P < 0.0001).

image

Figure 2.  Expt 1. (a) Mean root density of nonmycorrhizal control roots (open circles) and mean over isolates of AMF-infected roots, all isolates pooled (closed circles). (b) Mean root density of roots grown with four different isolates of Glomus intraradices (closed circles: B3; open circles: C3; closed squares: A4; open squares: D1; solid line: mean of isolates). (c) Mean hyphal density of four isolates of G. intraradices. (d) Mean spore density of four isolates of G. intraradices (closed circles: B3; open circles: C3; closed squares: A4; open squares: D1; solid line: mean of isolates).

Download figure to PowerPoint

Table 2.  Two-way crossed mixed-model ANOVA for three growth traits of (a) Expt 1 and (b) Expt 2
(a)
TraitExpt 1
Phosphate treatment (PT)Isolate (I)I × PT interaction
Spore density F 2,6 = 0.094(ns) F 2,160 = 78.76 *** F 6,160 = 2.74 *
Hyphal density F 2,6 = 1.1145(ns) F 2,160 = 123.05 *** F 6,160 = 1.86(ns)
Root density F 2,6 = 8.6 * F 2,160 = 2.29(ns) F 6,160 = 3.07 *
(b)
 Expt 2
Host species (H)Isolate (I)I × H interaction
  1. Significance levels: ns, P > 0.1; *, P < 0.1; *, P < 0.05; **, P < 0.001; ***, P < 0.001.

Spore density F 2,4 = 6.31(*) F 2,18 = 4.18 * F 4,18 = 9.16 ***
Hyphal density F 2,4 = 30.51 ** F 2,18 = 10.37 ** F 4,18 = 2.39(*)
Root density F 2,4 = 94.27 *** F 2,18 = 0.07(ns) F 4,18 = 0.17(ns)

MANOVA revealed that there was no significant overall phosphorous treatment effect on fungal growth (Pillai's Trace = 0.0402, approximated F4,320 = 1.6427, P = 0.16) in Expt 1. However, the isolates differed in their growth (Pillai's Trace = 0.9243, approximated F6,320 = 45.8293, P < 0.0001) and there was a significant isolate by phosphate treatment interaction (Pillai's Trace = 0.1453, approximated F12,320 = 2.0898, P = 0.0173). Univariate ANOVAs confirmed the absence of a significant phosphorus treatment effect on the two fungal traits (Fig. 2c,d, Table 2a). Spore density and hyphal density differed significantly among AMF isolates (Fig. 2c,d, Table 2a). The isolate by phosphate treatment interaction was significant for spore density, but not for hyphal density (Fig. 2c,d, Table 2a).

In Expt 2, different host species significantly altered fungal growth (Pillai's Trace = 1.618, approximated F4,36 = 38.07, P < 0.0001; Fig. 3). The isolates also differed in their growth (Pillai's Trace = 1.1391, approximated F4,36 = 11.909, P < 0.0001) and there was a significant isolate by host species interaction (Pillai's Trace = 0.9543, approximated F8,36 = 4.1066, P = 0.0015). Univariate ANOVAs showed that hyphal density and root density were significantly affected by host species (Fig. 3a,c, Table 2b). For all isolates, hyphal density on M. truncatula was reduced by > 50% compared with the other two host species. Similarly, spore density was decreased strongly on M. truncatula, and was increased by approx. 35% on S. tuberosum, relative to D. carota. Root density was not affected by AMF isolates (Fig. 3c, Table 2b). The isolate by host species interaction (IxH) was significant only for spore density (Fig. 3b, Table 2b). Isolate C3 produced almost twice as many spores as the other two isolates on D. carota, whereas on S. tuberosum this was the case for isolate C2. The observed differences in AMF growth cannot be explained by differences in root density of the three host species (Table 2b), because M. truncatula promoted the least fungal growth while producing the most roots (Fig. 3c).

image

Figure 3.  Fungal growth in Expt 2. (a) Mean hyphal density of three isolates of Glomus intraradices. (b) Mean spore density of three isolates of G. intraradices (black circles: B3; white circles: C3; grey circles: C2; solid line: mean of isolates). (c) Mean root density of roots of Medicago truncatula, Daucus carota and Solanum tuberosum, grown with three different isolates of G. intraradices.

Download figure to PowerPoint

Molecular variation within AMF isolates as a result of cultivation with differing phosphate concentration or different hosts

Over both experiments, we only scored one single AFLP polymorphism among AFLP replicates from the same DNA extraction. Considering that we scored over 25 000 AFLP peaks and that the AFLP procedure involves a restriction step and a ligation step, as well as two separate PCR reactions, the AFLP procedure worked very reliably for any given DNA sample.

In Expt 1, 10 different AFLP primer pairs gave a total of 1016, 1016 and 1077 differently sized fragments for isolates A4, B3 and C2, respectively. Four primer combinations revealed no polymorphism among the samples for all three isolates (Table 1b). The remaining six primer combinations showed a total of seven changes in isolate A4, and five changes in isolate C2, in the presence or absence of alleles (Fig. 4a). Isolate B3 did not show any changes. Overall, this represented a polymorphism of approx. 0.4%. Out of the 12 changes, all but one occurred as the absence of alleles in the P0.1 treatment compared with the P100 treatment.

image

Figure 4.  Minimum mutational steps generated from amplified fragment length polymorphism (AFLP) fingerprints in both experiments. (a) Changes in the AFLP fingerprints in Expt 1 for the two isolates A4 and C2 in the two treatments (P100, the standard M media; and P0.1, the media with the most reduced concentration of phosphate). (b) Changes in the AFLP fingerprints of Expt 2 of the lines (1–3) of isolate C3 growing with two different hosts: Solanum tuberosum (P) and Daucus carota (D); numbers represent the parental isolate (C3), or host (P, or D) lines (1–3), letters represent replicates (a,b); in bold are the two lines with replicate DNA extractions. Each step represents one change in the fingerprint. Numbers 0 and 1 adjacent to a given sample represent the presence (1) or absence (0) of the 13 polymorphic loci observed in isolate C3, and the 7 and 5 polymorphic loci in Expt 1.

Download figure to PowerPoint

In Expt 2, we used 16 primer pairs and detected 1255 differently sized fragments in the three lines of isolate C3 growing on D. carota (C3-D1, C3-D2, C3-D3), and the three lines growing on S. tuberosum (C3-P1, C3-P2, C3-P3). Nine primer combinations showed no polymorphism (Table 1c). The remaining seven primer pairs resulted in a total of 13 polymorphic loci, which represented a polymorphism of approx. 1%. Four of these polymorphisms occurred between the two different DNA extractions of the line C3-D2 on D. carota (Fig. 4b). The lines C3-D2 and C3-P1 were the most divergent lines. Depending on the DNA extraction, the AFLP profiles of line C3-D2 had 8 or 12 polymorphic loci, whereas line C3-P1 had had either no or only one polymorphic locus (Fig. 4b). This difference was the only AFLP error observed in our study. The other two lines growing on D. carota, representing three independent DNA extractions, appeared to be identical to each other, although they each contained six loci that were polymorphic in the complete data set (Fig. 4b). The other two lines on potato had seven (line C3-P3) and four (line C3-P2) polymorphic loci (Fig. 4b). The line C3-P2 also contained the only case where a locus could not be determined as present or absent. Overall, therefore, the change of host from carrot to potato resulted in a loss of polymorphic markers (Fig. 4b).

Of eight primer pairs used on isolate B3, we detected a total of 562 loci, and only one primer pair revealed any polymorphism (Table 1d). With that primer pair only one polymorphism was observed, where one line on D. carota clearly showed the presence of one allele, and the other lines growing on D. carota and S. tuberosum were either at the threshold of being absent or were absent for both replicates of each samples (data not shown).

We performed a permutation test to assess whether the consistent loss of polymorphic loci is caused either by random effects (possibly caused by methodological error or drift) or by actual, treatment-induced genetic changes within these AMF individuals. When considering AFLP profiles from different DNA extractions of the same treatment (irrespective of the line identity for isolate C3 in Expt 2), only in 1.25% of cases would such a consistent loss of loci occur by pure chance. When considering potential pseudo-replication among DNA extractions of the same line, the observed outcome is still similarly unlikely to be explained by extraction error only (P = 0.0219). These analyses suggest that when AMF isolates were forced to grow in a new environment, they overall changed genetically in such a way that their AFLP profile had a reduced presence of AFLP markers compared with their counterparts that grew in the old environment (i.e. normal M medium and carrot roots, Fig. 4).

Discussion

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Mycorrhizal research has a long tradition of studying the effects of altered environmental conditions in order to investigate AM fungal and host plant growth to understand their function and distribution in nature. Only in the last few years have studies documented genetic and phenotypic diversity in AM fungal populations at very local scales as well as at large scales (Koch et al., 2004, 2006; Munkvold et al., 2004; Scheublin et al., 2004; Gamper et al., 2005; Stukenbrock & Rosendahl, 2005; Croll et al., 2008). In this study we investigated the effect of varying the abiotic and biotic environment on the fitness and genotypic make-up of G. intraradices isolates that originate from the same field. Using an axenic system we showed that the biotic factor host species identity and the abiotic factor phosphate availability can alter fitness-related growth traits as well as the genotype of AMF isolates. Even though this culture system is artificial, it is a good way of preventing any form of contamination and eases the quantification of fungal fitness (number of spores and hyphae). The environmental conditions (e.g. identity of the host) are standardized and of a discrete nature and are therefore likely to influence fungi directly. While the present study suggests environment-induced genetic changes within single AMF isolates, the findings of Croll et al. (2008) suggest that plant species identity also affects AM population structure. Thus, it is quite likely that our findings are also valid outside this experimental set up.

Manipulating phosphate concentrations by three orders of magnitude changed the outcome of the interaction, as different isolates conferred the highest or the poorest host growth in different treatments. Our results showed that there is an existing and expressed variance in symbiotic efficiency under these growth conditions among isolates of G. intraradices originating from the same field. Such differences were exacerbated in phosphate-poor environments where the least efficient isolate reduced the growth of the host by one-third compared with the noncolonized controls. Overall, we found stronger isolate effects and stronger dependency of plant growth under conditions of reduced phosphate availability, as shown by a change in the sign of the phenotypic correlation between root and hyphal density in the different treatments. It has been shown at the species level that AMF can dominate phosphate supply, regardless of plant growth (Smith et al., 2003). Contrasting nutrient availability, and especially different phosphate concentrations in the soil, can alter the cost–benefit ratio of the different partners involved in the AMF symbiosis (Johnson et al., 1997). Our results suggest that this also occurs at the population level and that there is intraspecific variability in the degree of benefits that AMF individuals confer to their host plant.

Clonal growth of genetically different AMF isolates was also strongly affected by the host species identity. Interestingly, in our two experiments, different host plants seemed to have a more pronounced effect on changes in fungal growth than altered phosphate availability. The AMF isolates did not react equally to different host species and phosphate availability, which shows that there is genetic variation for fitness-related AMF traits in this population. The principle of competitive exclusion (Hutchinson, 1959) states that co-existing genotypes are expected to exhibit such genotype-by-environment interactions for fitness, and no single genotype is able to out-compete all others throughout a range of contrasting patches. In this niche-based, nonneutral view, co-existence is enhanced if different genotypes or species have fitness varying in space or time (Chesson, 2000; Barot & Gignoux, 2004). The observed isolate-by-host interaction on spore density also shows that different AMF individuals vary in the benefits they retrieve from their hosts. Arbuscular mycorrhizal fungi are generally considered as having a low specificity of association (Reynolds et al., 2003). Our results show a pattern of specificity of response within populations (at the isolate level), but no specificity of association because all AMF isolates were able to grow on all plant species. The view that ‘generalist species’, such as AMF, are actually made up of more specialized genotypes is also supported by a recent meta-analysis across a wide range of taxa of ‘generalist populations’ that tended to be made up of more specialized individuals (Bolnick et al., 2007). Our findings suggest that selection in AMF populations could occur at potentially very small spatial scales and that genetically different AMF isolates of the same species have evolved alternative growth strategies, which may facilitate their co-existence.

In both experiments a majority of the primer pairs used for genotyping AMF isolates gave highly reproducible results and did not allow us to detect any genetic changes (99% of the bands were identical and therefore reproducible). However, overall, our genetic analyses indeed revealed some changes in the genetic composition of AMF isolates. The observed 0.5–1% genetic changes in the G. intraradices individuals may seem to be low. However, these changes occurred only over a few generations of cultivation in contrasting conditions. Furthermore, using a permutation test, we showed that the genetic changes observed are unlikely to be of a random nature. As AFLP represent a superposition of the global diversity within one AMF individual, and given that they are coenocytic and multigenomic (Kuhn et al., 2001; Hijri & Sanders, 2005), genetic changes (or AFLP polymorphism) could predominately occur in two ways: nuclei bearing newly arisen mutations increasing in frequency; or changes in frequencies of genetically different nuclei (nucleotypes). The majority of the observed changes occurred as absences at loci and occurred significantly more often in the new environment (reduced phosphate or change in host species) than in the starting environment (M medium with D. carota as the host), This suggests that most of the detected polymorphisms were caused by frequency changes of nucleotypes rather than by the occurrence of new mutations. In this study it would imply a potential loss or decrease in frequency of nucleotypes bearing those loci, as a result of experiencing a new environment. The result of our permutation test also implies that differences in the quality of DNA extraction among the samples are unlikely to be the only cause of the observed loss of markers in the new treatments (new host, reduced phosphorus) compared with the old treatment (M media with transformed carrot). Large differences among DNA extraction quality would be expected to result in much greater overall polymorphism than we actually observed. Isolate B3 showed no polymorphism in Expt 1, and only one AFLP polymorphism in Expt 2, and thus differed qualitatively from the other isolates where more polymorphisms were detected. This could reflect differences in the genetic architecture (e.g. diversity among nucleotypes) among isolates, limitations of this method to detect genetic changes for certain AMF genotypes, or increased genetic stability of this particular isolate. Within-individual stochastic processes (i.e. drift) are unlikely to have played a major role in the divergence during the experiment, as we vegetatively propagated the cultures using a large inoculum (hyphae, spores and colonized roots, Fig. 1). However, drift could have occurred when we first established replicate lines at the beginning of Expt 2, using clusters of spores (20–50 spores), and could be responsible for some of the divergence between the lines within treatments. Overall, our findings suggest that the nuclear composition within an AMF individual is more dynamic than previously thought and not necessarily stable. However, we cannot establish from our experiments whether these new genetic polymorphisms are stable or transient. AFLP markers are nonspecific and it is difficult to draw a definitive conclusion about within-individual evolution solely on their basis; other studies addressing this topic would be needed using a greater number of lines, a greater number of environments and a greater number of generations.

Our results imply that levels below the morpho-species are fundamental for the evolutionary understanding of AMF. The AM symbiosis is a multilevel hierarchical system, where different genomes co-occur within individuals, different individuals are present within populations and populations make up species and communities. All these different levels directly or indirectly interact with each other as well as with the host plant community and with other biotic and abiotic factors. Fitness-feedbacks, divergent phenologies and heterogeneity in soil properties have been shown to affect AM communities and can enhance the maintenance of different species of AMF and plants (Bever, 2002). Our findings suggest that similar mechanisms (e.g. differential host affinities of AMF isolates) are potentially important in shaping and maintaining genetic diversity within populations of AMF. Our results are also in line with the recent results of Croll et al. (2008), where the genetic diversity of AMF isolates from the same population showed a nonrandom pattern of association driven by the host species identity. The results of Croll et al. (2008) were based on a population analysis of genetic diversity and are thus indirect, whereas our study provides direct experimental evidence for a strong pattern of fitness differences driven by host identity. As AMF isolates form hyphal networks simultaneously colonizing different hosts or encountering different abiotic environments at different locations of the network, the consequences of this on the maintenance of genetic diversity, among isolates within populations and within isolates are difficult to predict; it is even more difficult to predict the consequences on the genetic diversity knowing that closely related hyphal networks have the ability to anastomose and exchange genetic material (Croll et al., 2009). The different environments will affect local fitness of a single hyphal network, but the overall fitness of the network will probably depend on its size, on the number of encountered hosts and on other environmental factors (e.g. the occurrence of other AM taxa (Maherali & Klironomos, 2007) or other isolates of the same species). Estimating the relative effects of these factors on fitness of an individual fungal network is far from trivial and will require more investigations.

We conclude that there is genetic variation in AMF populations for individual fitness in response to contrasting abiotic and biotic factors. Furthermore, the genetic make-up of AMF individuals appears to be affected by environmental variables. Identifying and quantifying the relative importance of different factors that influence the dynamics and maintenance of genetic diversity within AMF individuals and populations needs further investigation. Because of the high intraspecific variability in AMF documented here and elsewhere, we caution the use of terms such as ‘effects of AMF species’, where findings are based on research that used single isolates as the only representatives of an AM species. Assessing the functional and ecological importance of genetic variability within and among AMF species and communities will be necessary to understand in greater detail the genetics of these important plant symbionts.

Acknowledgements

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

The authors thank the research groups of Marcel Bucher (ETH Zürich, Switzerland) and Guillaume Bécard (CNRS, University Paul Sabatier, Toulouse, France) for providing the roots of S. tuberosum and M. truncatula. We thank three anonymous referees for their comments that greatly improved the manuscript, Tad Kawecki for helpful discussion, Daniel Croll, Guillaume Evanno, Laurent Lehmann for helpful comments on the manuscript and Rui Candeias for skilful technical assistance. This research was funded by the Swiss National Science Foundation (projects 3100AO-105790/1 and PAOOA-119519) to whom support is gratefully acknowledged.

References

  1. Top of page
  2. Summary
  3. Introduction
  4. Materials and Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References