Correspondence: Paul Widden, Concordia University, Department of Biology, Groupe de recherche en écologie forestière interuniversitaire (GREFi), 7100 Sherbrooke Street West, Montreal, QC, Canada, H3G 1M8. Tel.: +1514 848 2424 ext: 3413; fax: +1514 848 2880; e-mail: email@example.com
Arbuscular mycorrhizal fungi (AMF) communities in Clintonia borealis roots from a boreal mixed forests in northwestern Québec were investigated. Roots were sampled from 100 m2 plots whose overstory was dominated by either trembling aspen (Populus tremuloides Michx.), white birch (Betula papyrifera Marsh.), or mixed white spruce (Picea glauca (Moench) Voss) and balsam fir (Abies balsamea (L.) Mill.).Part of the 18S ribosomal gene of the AMF was amplified and the resulting PCR products were cloned. Restriction analysis of the 576 resulting clones yielded 92 different restriction patterns which were then sequenced. Fifty-two sequences closely matched other Glomus sequences from Genbank. Phylogenetic analysis revealed 10 different AMF sequence types, most of which clustered with other uncultured AM sequences from plant roots from various field sites. Compared with other AMF communities from comparable studies, richness and diversity were higher than observed in an arable field, but lower than seen in a tropical forest and a temperate wetland. The AMF communities from Clintonia roots under the different canopy types did not differ significantly and the dominant sequence type, which clustered with AM sequences from a variety of environments and hosts at distant geographical locations, represented 66.9% of all the clones analyzed.
The arbuscular mycorrhizal fungi (AMF) form an ancient and widespread type of endomycorrhizal association found on the roots of up to two-thirds of the world's vascular plant species (Newsham et al., 1995). This mycorrhizal symbiosis is one of the most important plant-microbe associations. The AMF aid their hosts by improving nutrient and water uptake, increasing pathogen resistance and enhancing stress tolerance (Smith & Read, 1997). AMF also play an integral role in determining plant community structure (van der Heijden et al., 1998). Plants differ in their responses to different AMF types and AMF differ in their ability to distribute nutrients between coexisting plants. This high variation in the responses of plant growth to different AMF species is thought to be a controlling factor in the overall structure of plant communities (Klironomos, 2003; van der Heijden et al., 2003). Despite their global distribution and ecological importance, knowledge of the community structure of AMF fungi is, however, scarce.
AMF are obligate symbionts and cannot be grown in culture without their host. Due to the difficulties of culturing the AMF from plant roots, the identification and quantification of AMF within an ecosystem has usually been based on studies of the asexual spores collected from the soil. Such studies have shown that host plant communities can influence the fungal community composition (Johnson et al., 1992; Eom et al., 2000). However, there are many problems associated with using spores to analyze AMF communities. For example, recent advances in our understanding of the phylogeny of the Glomeromycota based on DNA sequences have demonstrated that some highly divergent taxa are not distinguishable by their morphological characteristics (Morton & Redecker, 2001). Hence, studies of the AMF based on morphological characters of spores alone may not only leave many species unresolved, but, even when they can be identified, basing our understanding of AMF communities on spores in the soil is much like basing studies of plant communities only on the seed bank available. It is, therefore, not surprising that molecular studies have shown that populations of spores in soil do not reflect the fungi present in roots (Clapp et al., 1995; Rosendahl & Stukenbrock, 2004). Using the AM1 primer (Helgason et al., 1998), which targets the small subunit ribosomal RNA (SS rRNA) gene of AMF, we are now able to detect the DNA of the fungi actually colonizing plant roots.
Using SS rRNA gene sequences to understand AMF community structure also has its limitations, as it is known that different nuclei within a single isolate can contain different copies of the SS rRNA gene (Sanders, 2002; Hijri & Sanders, 2005). However, Schüßler (Schüßler, 1999; Schüßler et al., 2001) reports that the within-isolate variation of the 18S SS gene is relatively small, and that AMF phylogeny has been based on this gene (Schüßler et al., 2001). Moreover, this method allows for the comparison of the genetic variation of the AMF found in this study with that from other ecosystems based on the same gene. To date, data on the genetic variation of the AMF using the AM1 primer is probably the largest data set available on the genetic diversity of AMF collected from diverse natural environments.
Using SS RNA, Helgason et al. (1999) showed that the AMF of bluebell [Hyacinthoides non-scripta (L.) Chouard ex. Rothm] roots collected from under oak were dominated by Acaulospora spp. whereas those collected from under sycamore canopies were dominated by Glomus spp. This was one of the first studies to show that the frequency of AMF sequences collected from a single plant host differed under differing canopy types. Molecular studies have also shown that different AM taxa are present in the roots of plants growing in different environments. Helgason et al. (1998) and Daniell et al. (2001), have shown that the AMF diversity in plant roots collected from an arable field was lower than that from a near-by woodland, and Helgason et al. (1998), also showed that 92% of the sequences from crop plants were from Glomus mosseae, a species rarely found in the woodland system. Another common finding from molecular studies of AMF is that many of the sequences collected from the field do not match sequences from known, pot-cultured AMF. Thus, currently, sequences are partitioned into groups based on their similarities and cannot be assigned to definite species.
The mixed boreal forest of eastern Canada is an ecosystem in which the ecological processes are controlled to a large extent by disturbances such as fire and pest outbreaks, creating a landscape with a high spatial heterogeneity in canopy composition. In the Abitibi region of Québec, this results in a mosaic of patches dominated by either angiosperms such as birch and poplar, or by conifers such as pine, spruce or fir (Bergeron, 2000), all of which are ectomycorrhizal (ECM) hosts. Legaréet al. (2001) have shown that the understory plants, most of which are AM hosts, differ in community composition among these different canopy types, though some species, such as Clintonia borealis (Ait.) Raf. and Aralia nudicaulis, can be found throughout.
From the results of previous studies on AMF communities (Helgason et al., 1999) we predicted that different AMF sequence types would be found in the roots of Clintonia plants collected from under the different canopy types. With the large genetic variation reported in the AMF, we also expected to find many novel sequence types in the Clintonia roots collected in this previously unexplored ecosystem.
We therefore undertook this study in order to obtain basic knowledge about the community composition of AMF in Clintonia borealis (Ait.) Raf. growing in a natural boreal mixed wood forest. Our specific objectives were (i) to examine the AMF SS rRNA sequence diversity from Clintonia borealis roots collected from a boreal mixed-wood forest and to compare the AMF sequence diversity and richness with those of other studies using the same primer pair, and (ii) to test whether the AMF community in the Clintonia roots would differ in plants collected under birch, aspen or conifer dominated canopies, and whether the AMF community would differ in sites of different post-fire ages.
Materials and methods
The study area is located around Lake Duparquet, in northwestern Québec (48°30′N, 79°20′W). This area is part of the western balsam fir–paper birch (Abies balsamea–Betula papyrifera) bioclimatic domain (Grondin, 1996), which extends over the Clay Belt region of Québec and Ontario. The closest weather station is located at La Sarre, 35 km north of Lake Duparquet. The average annual temperature is 0.8°C, daily mean temperature is −17.9°C for January and 16.8°C for July, and the average annual precipitation totals 856.8 mm (Environment Canada, 1993). By dendrochronological analysis, Bergeron (1991) and Dansereau & Bergeron (1993) determined that the stands originated from fires that took place 34 to 281 years ago.
A total of 18 sample plots, each measuring 10 × 10 m, were selected on sites with similar clay deposits from an already existing design in the Lake Duparquet Research and Teaching Forest (Legaréet al., 2001). Plots were located in stands of two different age classes, originating either from fires that took place in 1870, or in 1916/1923. Within each stand age, three replicate plots of three different forest canopy types were selected. Plots were categorized as trembling aspen (Populus tremuloïdes Michx.), paper birch (Betula papyrifera Marsh.), or white spruce–balsam fir (Picea glauca (Moench) Voss–Abies balsamea (L.) Mill.) if the dominant species occupied at least 75% of the total basal area. In all stands, dominant trees originated after fire, except for the aspen stand of 1870, which is a second cohort of aspen (Bergeron, 2000). At each stand a total of nine plots were sampled (three × three canopy types).
Overstory, shrub layer and understory vegetation analysis
The overstory composition within the 100 m2 plots was originally recorded in 1994 (S. Legaré, pers. commun.). In August 2002, we re-analyzed the overstory composition of these plots to ensure that the data still reflected the overstory composition within the plots. At the time of sampling, a shrub layer of ∼1.5–2.5 m was present in most of the plots. By visual observation, the percentage cover of each tree in this shrub layer was also recorded for each plot. In order to assess the understory herbaceous layer, a 1 × 1 m subplot was set up in each of the 18 100 m2 plots, and the percentage cover of each species in each subplot was estimated.
Equal portions of dried root samples of each of the three plants collected in each plot were combined and DNA was extracted from this root mixture. In a mortar, liquid nitrogen was poured over the root samples and the roots were crushed with a pestle. 300 μL of CTAB lysis buffer (Gardes & Bruns, 1993) containing 0.02% b-mercaptoethanol were added to the mortar and this solution containing the crushed root sample was aspirated into a 1.5 mL eppendorf tube. 100 μL of a 200 μg mL−1 solution of proteinase K in TE buffer with 5% lauryl sulfate sodium salt was added to the sample and incubated for 1 h at 64°C. One volume of Tris-saturated phenol:chloroform:isoamyl alcohol (24 : 24 : 2) was then added and the tubes were mixed gently and centrifuged at 13 000 g for 15 min. The aqueous phase was placed into a clean eppendorf tube and the chloroform:isoamyl alcohol step was repeated. After centrifugation, the aqueous phase was placed into a clean tube and 100 μL of a 50 μg mL−1 solution of RNAse in TE buffer (Sambrook et al., 1989) were added and the samples were incubated for 1 h at 37°C. The phenol:chloroform:isoamyl alcohol and chloroform:isoamyl alcohol separation steps were repeated. The aqueous phase was placed into a clean tube and 0.8 volume of cold isopropanol was added. To this solution 0.1 volume of sodium acetate (3M, pH 5.2) was added and the samples were kept at −20°C for 1 h. The samples were then centrifuged at 13 000 g for 15 min and the supernatant was discarded. Eight hundred microliters of cold 70% ethanol were added to wash the pellet and the samples were centrifuged at max speed for 15 min. The supernatant was removed and samples were dried at room temperature. DNA was re-suspended in 50 μL of sterile ultra-pure water and incubated at 37°C for 1 h. After being re-suspended the samples were stored at 4°C. DNA was extracted from 18 different root mixtures. Partial SS rRNA gene fragments were amplified using a universal eukaryotic primer, NS31 (Simon et al., 1992), and AM1, a primer that has been shown to amplify 3 families of the AM fungi (Glomeraceae, Gigasporaceae, and Acaulosporaceae; Helgason et al., 1998). A few DNA dilutions were tested (undiluted, 1 : 10 and 1 : 100) to increase the likelihood of obtaining a strong amplification product. Controls with no DNA were run with every series of amplifications to test for the presence of contaminants. The reactions were carried out in a final volume of 50 μL in the presence of 0.2 mM dNTPs, 25 pmols of each primer, and 2.5 U of Taq DNA polymerase. The thermal parameters used were similar to those cited in Gardes & Bruns (1993). The PCR parameters were initially tested using DNA extracted from Glomus intraradices spores obtained from Premier Tech Ltd. (Rivière du Loup, Québec), and from DNA extracted from Clintonia borealis roots from a nearby forested area to ensure AMF DNA would be amplified. Based on these tests, the annealing temperature was raised to 58°C.
The resulting PCR products were cloned using the p-GEM-T easy cloning kit (Promega Inc., Madison, WI). 32 positive clones of the desired length (∼550 bp) were selected and re-amplified with the same primer pair as in the initial PCR for each DNA sample. A total of 576 cloned products were then digested with HinfI, NdeII, and TaqI according to the manufacturer's instructions (Promega Inc., Madison WI).
If available, the remaining fine roots from each individual plant were stored in formalin-acetic acid for the morphological examination of the AM colonization. Roots were stained with 0.1% Chlorazol black E (De Bellis & Messier, 2002) and stained roots were mounted on microscope slides in glycerine jelly (Widden, 2001). The presence or absence of fungal material (AMF hyphae, vesicles or arbuscules) was recorded using a gridline intersect method (McGonigle et al., 1990) and the percentage of colonized intersections was calculated.
Restriction fragment length polymorphism analysis and taxonomic analysis
The 32 restriction fragment length polymorphism (RFLP) patterns obtained from each plot were initially separated on a per plot basis using GERM (Dickie et al., 2003). Once the analysis of each plot was complete, a data set containing all unique RFLP patterns from all 18 plots was re-analysed. From this analysis, 92 unique RFLP types were found, one of which was extremely dominant. Examples of each RFLP type and 39 samples of the dominant RFLP type were re-amplified, and PCR products were sequenced using an ABI PRISM® 3730XL DNA Analyzer system at the McGill University and Genome Québec Innovation Centre.
To identify the RFLP types found, sequences were initially aligned using nucleotide-nucleotide BLAST (blastn) in Genbank in order to locate similar AMF sequences to act as references. The neighbour-joining (NJ) algorithm and maximum parsimony (MP) procedures, using PAUP* 4.0b10 (Swofford, 2002), were used to infer the placement of our sequences relative to the reference sequences obtained from Genbank. Distances for the NJ tree were computed using the Kimura 2 parameter model. For the MP analysis, a heuristic tree search with tree bisection and reconnection (TBR) as the swapping algorithm using 500 replicates with a random addition order of sequences was used to find the shortest tree. Bootstrap values for branches were estimated from 500 replicates for the MP analysis and from 1000 replicates for the NJ tree. ClustalX 1.81 (Thompson et al., 1997) was used to align the AM fungal sequences from this study along with 43 reference sequences from GenBank. The alignments were then manually adjusted using Bioedit Ver. 5.0.6 (Hall, 1999). The tree was rooted with a nonflagellate protozoan Corallochytrium limacisporum sequence obtained from Genbank (Accession no. L42528).
We tested whether the abundance of AMF sequence types differed between the Clintonia roots collected under the three different canopy types and between sites of different postfire ages. For this analysis, data for each plot of similar canopy type of each fire year were pooled, which resulted in three replicates per group. The data was not normally distributed, therefore the Mann–Whitney U and the Kruskal–Wallis H tests were used (SPSS, version 10.1). Tests were performed only with groups which had a minimum of five clones per sequence group, therefore omitting comparisons among groups with rare sequence types.
Vegetation data for the overstory, shrub and understory layers were averaged for all plots of similar canopy type and year (Fig. 1). In all 18 plots, the upper canopy was still dominated (>75%) by either trembling aspen, white birch or spruce-fir, as was originally recorded in 1994 (Fig. 1a). The most common species in the shrub layer were Acer spicatum Lam., an arbuscular mycorrhizal species, followed by Abies balsamea (L.) Mil., Corylus cornuta Marsh. and Picea glauca (Moench) Voss., all of which are colonized by ECM fungi (Fig. 1b). The most common understory plants included: Aster macrophyllus L., Aralia naudicaulis L., Linnaea borealis L., Acer spicatum seedlings, Cornus canadensis L., Rubus pubescens Raf. and Viola renifolia A. Gray. The mean percentage cover of AM plants vs. ECM plants in the understory for each canopy type per fire year is shown in Fig. 1c.
AM colonization rates
The AM colonization rates for 31 of the 54 roots samples collected in this study were calculated. The morphology of the AMF colonizing the roots was similar to that published by Widden (1996). Fourteen plants from the 1870 site and 17 plants from 1916 site were analyzed. Colonization rates per plot ranged from 42% to 78% (Table 1).
Table 1. Arbuscular mycorrhizal colonization rates of some of the plants collected from the 18 plots
Mean % coloniz ation/plot
Mean % coloniz ation/canopy type/fire year
RFLP analysis, diversity, and sequence analysis
In total, 576 clones were screened. Six clones were eliminated from the analysis after the RFLP screening because band lengths did not add up to ∼550 bp. The remaining 570 clones were divided into 92 different RFLP types, all of which were sequenced. Fifty two of the sequences closely matched Glomus sequences. Of the 40 remaining sequences, one was closely related to a Penicillium sp., another closely matched a bacterium and another had a sequence that matched C. borealis. Seventeen other sequences did not match any other sequence in Genbank, and we did not obtain clear DNA sequences for the remaining 20. Thus all these sequences were excluded from the analysis. The 20 RFLP types that did not sequence very well only occurred once, thus we did not attempt to re-sequence them. Five hundred and twenty-two of the 576 clones screened belonged to one of the 52 RFLP types that had high sequence similarity (>90%) to Glomus spp.
Both the NJ and MP analyses of the 18S SS sequences collected from the Clintonia roots revealed 10 similar clusters. As both methods gave similar results, only the NJ tree is shown in Fig. 2. The sequence identity within the clusters ranges from 97.3% to 100%. All 52 sequences have been submitted to the Genbank database (Accession nos. DQ122622 to DQ122673, Supplementary Table 1). All sequences obtained from this study fall within the Glomeraceae. The QU-Glo1 group (Fig. 2) contains Glomus sequences collected from a variety of plants from different environments, as well as our most abundant sequence type. Sequences collected from legume and nonlegume plants collected from a dune grassland in Holland, from Pulsatilla spp. in Estonia, from Agrostis capillaris L. and Trifolium repens L. from a grassland in Scotland and from Hyancinthoides non-scripta (L.) Chouard ex Rothm., Gleochoma hederacea L. and Ajuga reptans L. roots in a temperate forest in England all cluster within this group. Group QU-Glo4 also contains sequences from the Estonian and Scottish studies mentioned above. Groups QU-Glo2 and QU-Glo9 contain sequences obtained from seedlings of indigenous woody plants in a warm-temperate deciduous broad-leaved forest in Japan. Group QU-Glo3 contains a sequence submitted by the University of York, and another submitted from the University of Montana from undisclosed host plants. Group QU-Glo5 also contains sequences collected from studies of dune grassland in Holland, from Estonia and from a grassland in Scotland (all of which were mentioned in relation to group QU-Glo1). This group also contains sequences from Thymus polytrichus A. Kerner ex Borbás ssp. britannicus (Ronn.) Kerguelen from metal contaminated soil in northern England and from Phragmites australis (Cav.) in a temperate wetland in Germany. Group QU-Glo6 contains sequences from the dune grassland in Holland and sequences submitted by the University of Kansas and the University of Montana from unknown host plants. Sequences in the QU-Glo7 group include 2 from Phragmites australis (Cav.) collected from a temperate wetland in Germany. Group QU-Glo8 contains sequences from Thymus polytrichus in metal contaminated soils in northern England. QU-Glo3 contains sequences from temperate grassland in the UK and one deposited in Genbank from the University of Missouri (US) and forms a clade with G. intraradices. QU-Glo10 included our sequence AMR31 and a known Glomus species, Glomus viscosum.
Richness and diversity indices based on the 10 AM sequence groups were calculated for each canopy type with the data from the six plots from the two fire years grouped together. The highest diversity was found in the birch plots, with a Shannon's diversity index of 1.41, followed by the aspen and conifer plots with a diversity index of 1.03, and 0.95 (respectively). AM richness for the pooled plots of similar canopy type are 8, 6, and 8 for the birch, conifer, and aspen plots respectively.
The dominant sequence group, QU-Glo 1, represented 66.9% of the clones. This sequence was found in all canopy types, and it dominated 13 (>50% of clones analyzed) of the 18 plots. Sequence groups QU-Glo 2, 4, and 5 comprised 9.2%, 7.9%, and 6.5% of the clones, respectively. Sequence groups QU-Glo 3, 8, and 9 had frequencies between 2–4%, and all other sequence groups constituted less than >1% of the clones (Fig. 3). Pairwise comparisons of the 39 sequenced clones of the most common RFLP type showed that all sequences were ≥99% similar. Only one of those sequences, therefore, was used in the construction of the neighbor joining tree, sequence AMR1, which clusters in the QU-Glo1 group (Fig. 2).
Differences under the varying canopy types and between sites of different post-fire ages
Most of the plots were dominated by the QU-Glo1 sequence group (Fig. 3). This sequence group was dominant in all plots except the three 1870 birch plots and the two 1916 aspen plots (Fig. 3). The statistical analyses did reveal that the abundance of this sequence group between the birch plots of the two post-fire ages was significantly different (P=0.05), and the abundance of QU-Glo1 between the 1870 birch plots was also significantly different from that of the 1870 conifer and aspen plots (P=0.05), however the abundance of this sequence type did not differ significantly between the 1870 conifer and aspen plots. There were no significant differences in the abundance of sequence groups among the plots of different canopy type from the 1916 site.
The three 1870 birch plots have a high percentage of AM plants in their understory and have an approximately 60% cover of mountain maple (an AM plant) in their shrub layer, which is higher than that found in any other plots (Figs 1b and c). The abundance of mountain maple and AM herbs may be contributing to the lack of dominance by the QU-Glo1 sequence group in these three plots. However, two of the 1916 aspen plots also were not dominated by this group; although they did have a relatively high percentage of AM plants in the understory, they did not have a dense AM cover in the shrub layer (Figs 1b and c).
Although the remaining AM groups were not found in high numbers, we saw that certain AM groups were found exclusively in certain plot types. QU-Glo 6 and 7 were exclusively found in the aspen plots and QU-Glo 10 was only present in a birch plot. QU-Glo 2 was only found in birch and conifer plots, and QU-Glo 9 was only found in birch and aspen plots.
The data presented in this paper is the first look at AM sequences in plant roots collected from a boreal mixed wood forest in eastern Canada. Our main findings are that there was a high dominance by one AM group and that there were no significant differences in the abundance of AM sequence groups among the different stand types. All sequences collected from the Clintonia roots belong to one family, the Glomeraceae. Also, most of the sequences present in the Clintonia roots do not match sequences of known cultures of AM fungi but they did match other uncultured AM sequences obtained from plant roots collected from various field/natural sites.
Dominance by Glomus spp. in plant roots has been reported from a number of different habitats ranging from stable forest environments (Husband et al., 2002a,b; Opik et al., 2003), wetlands (Wirsel, 2004) and dune grasslands (Scheublin et al., 2004), to highly disturbed agricultural fields (Daniell et al., 2001). The boreal mixed wood forest is a stable environment that is dominated by ECM trees. Thus, it may be very beneficial for an understory herbaceous plant growing under a tree canopy to tap into the already established AM mycelial network and quickly gain access to the underground nutrient absorbing system. As it is known that different AM fungi can affect plants in different ways (Klironomos, 2003; van der Heijden et al., 2003), a high diversity of AM fungi may be needed for seedlings at the critical establishment stage but once established in a fairly stable environment, AM fungi that are best suited to that particular stable environment may be selected for. Husband et al. (2002a,b) reported a decline in fungal diversity and evenness and noted a gradual replacement of fungal types over time in the AM fungi colonizing Tetragastris and Faramea seedlings collected from a tropical forest. In C. borealis, new ramets arise from rhizomes, rhizome sections can persist for 10 or more years and the establishment of new patches is rare (Pitelka et al., 1985). In this situation, AMF that can quickly colonize available rootlets via hyphae may be selected for and these selection pressures may result in an environment with low AMF diversity due to the dominance of the fungi with a high preference for Clintonia borealis. Species within the Glomeraceae may be best suited to such an environment as studies have shown that they can colonize new roots from hyphal fragments (Tommerup & Abbott, 1981) and can do so at a much faster rate than members of the Acaulosporaceae and Gigasporaceae (Hart & Reader, 2002). A new C. borealis plant developing along an already existing rhizome does not experience the same establishment pressures that a seedling may encounter and may simply benefit the most by quickly tapping into the already established mycelium.
The dominant sequence group, QU-Glo 1 (Fig. 2) contains AM sequences from a temperate deciduous woodland in England (AF485876), from Pulsatilla plants from Estonia (AJ496046), and from a dune grassland in Holland (AY512354). This group seems to represent a generalist fungus with a broad host and geographical range. In fact, most of the sequences found in this study cluster with other Glomus sequences from a variety of environments. The primer pair AM1 and NS31, used in all of these studies, may be contributing to the marked similarities in sequences obtained from studies from distant geographical locations. The AM1 primer was designed on the basis of only 12 AMF sequences (Helgason et al., 1999) and it is known that it can amplify AM sequences from only three families within the Glomeromycota [the Glomeraceae, Gigasporaceae and Acaulosporacae) Daniell et al., 2001]. Furthermore, de Souza et al. (2004) have discovered that the V3-V4 region of the 18S rRNA gene, the region amplified by AM1 and NS31, did not contain enough variation to discriminate between different Gigaspora species. However, based on studies of spores found in the soil, species of Glomus are by far the most abundant AM fungi in the soils of the Eastern Canadian Shield, with some Acaulospora spp. also being reported (Klironomos, 1995; Moutoglis & Widden, 1996). Nevertheless, as sampling was done just once and the primers may exhibit amplification bias, the 10 different sequences obtained may not be an entirely accurate representation of the AMF community in this ecosystem.
The AM richness and diversity from this boreal forest was higher than that observed in an arable field (Daniell et al., 2001), but was lower than that seen in many other studies from a range of habitats (see Table 2). However in most of the studies listed in Table 2, more than one plant host was analysed and plants were sampled at more than one time, both of which can have a great impact on AMF richness and diversity (Helgason et al., 2002). In order for our data to be more comparable with the data from these other studies, diversity and richness based on single plant host from a single sampling period or site were calculated and are presented in the last two columns of Table 2. When the data is shown in such a manner, we see that the diversity observed from the boreal mixed wood forest is lower than that observed in a tropical forest (Husband et al., 2002a,b) and a temperate wetland in Germany (Wirsel, 2004), but similar to that seen in a dune grassland in Holland (Scheublin et al., 2004). The diversity in this study is similar to that of a woodland in North Yorkshire, UK (Helgason et al., 2002), but the number of clones screened in that study was much lower than those screened in the present study (see Table 2), therefore the values from the 2 studies are not very comparable.
Table 2. Comparison of AM richness and Shannon's diversity (H′) based on the SS rRNA gene from nine other studies of plant roots collected in natural ecosystems
P. pratensis, Pulsatilla pratensis;P. patens, Pulsatilla patens. The number following the plant name represents the number of root samples that were successfully amplified. The following letter and number code in brackets represents the sampling sites: G1 and G2=dry meadow, F1 and F2=boreal pine forest, B1=sandy area bordered by pine forest, B2=roadside bordered by pine forest.
Another objective of this study was to see whether there would be any shifts in the AMF community in plants collected under the different canopy types. The most common sequence group QU-Glo1 was present in significantly lower amounts in the 1870 birch plots compared to the other canopy types of similar age, but there were no differences in the abundance of QU-Glo1 between plots of different canopy type from the 1916 sites, hence this finding could not be attributed to difference in tree composition. Although Clintonia plants were collected under an ECM canopy, in each plot a lush AM herbaceous layer was present and in many plots a shrub layer consisting mainly of Acer was present (Fig. 1). In the Helgason et al. (1999) study, differences in the types of AM sequences found in the Hyancinthoides roots collected under either a sycamore- or oak-dominated canopy were seen. Differences in the types of fungi colonizing the roots of Clintonia in the different plot types were not observed, but all plot types in this study were dominated by ECM tree species while the Hyancinthoides were collected under an ECM species (oak) and an AM plant (sycamore). The only other study where an AM herbaceous plant was collected from under an ECM canopy is that of Opik et al. (2003) who looked at the AM community in the roots of 2 Pulsatilla species that were collected from a Scots Pine forest Estonia. In that study however, only six root samples collected from the forest or roadsides bordered by pine forest were successfully amplified. The resulting diversity and richness values were quite low, probably due to the small sample size.
Most of the main findings of this study are in agreement with those of other studies using similar methods. We have seen that most of the AM sequence groups found do not match sequences from known cultures but closely match other sequences obtained from roots collected in the field. This further emphasizes the lack of knowledge of these organisms in natural environments. We have shown that Glomus spp. dominate in this forest system, as was seen in several other studies from a variety of environments. Contrary to Helgason et al. (1999), we did not see any differences in the distribution of the different AM taxa in Clintonia plants collected from under the different canopy types. What is of particular interest is that the sequences found in this study are very similar to those found in different plant hosts from a variety of ecosystems from many parts of the world. The fact that the most common sequence type found in this boreal forest is similar to AM sequences found from a wide variety of plants from various geographic locations may indicate that the AM fungi have some sort of evolutionary mechanism that may be quite particular to this group of symbiotic organisms or these similarities may be due to primer biases and we are amplifying only a small portion of the actual AMF colonizing plant roots. More field-based studies targeting several AM genes are needed to obtain a better understanding of AMF. We conclude by agreeing with Rodriguez's et al. (2004) plea that future work on the diversity of the AMF should include data on a few genes, in order to further expand or knowledge on the ecology and genetics of these organisms.
This study was supported by a grant from the Natural Sciences and Engineering Research Council of Canada (NSERC) to P.W. and a strategic grant by NSERC to Robert Bradley (P.I), Y. Bergeron and P.W. Further support of this work came from Le Fonds québécois de la recherche sur la nature et les technologies, J.W. McConnell, Groupe de recherche en écologie forestière interuniversitaire, Power Corporation and the Canadian-Italian Business Professional Association fellowships to T.D. We greatly thank Goldie Marmor for her laboratory assistance, Sarah McNair for her field assistance, and Dr J. Klironomos for his comments on an earlier version of this manuscript.