|Home | About | Journals | Submit | Contact Us | Français|
The relationships between plant carbon resources, soil carbon and nitrogen content, and ectomycorrhizal fungal (EMF) diversity in a monospecific, old-growth beech (Fagus sylvatica) forest were investigated by manipulating carbon flux by girdling. We hypothesized that disruption of the carbon supply would not affect diversity and EMF species numbers if EM fungi can be supplied by plant internal carbohydrate resources or would result in selective disappearance of EMF taxa because of differences in carbon demand of different fungi. Tree carbohydrate status, root demography, EMF colonization, and EMF taxon abundance were measured repeatedly during 1 year after girdling. Girdling did not affect root colonization but decreased EMF species richness of an estimated 79 to 90 taxa to about 40 taxa. Cenococcum geophilum, Lactarius blennius, and Tomentella lapida were dominant, colonizing about 70% of the root tips, and remained unaffected by girdling. Mainly cryptic EMF species disappeared. Therefore, the Shannon-Wiener index (H′) decreased but evenness was unaffected. H′ was positively correlated with glucose, fructose, and starch concentrations of fine roots and also with the ratio of dissolved organic carbon to dissolved organic nitrogen (DOC/DON), suggesting that both H′ and DOC/DON were governed by changes in belowground carbon allocation. Our results suggest that beech maintains numerous rare EMF species by recent photosynthate. These EM fungi may constitute biological insurance for adaptation to changing environmental conditions. The preservation of taxa previously not known to colonize beech may, thus, form an important reservoir for future forest development.
In temperate and boreal forest ecosystems, most tree species form ectomycorrhizal fungal (EMF) associations. EM fungi ensheathe the root tip, forming characteristic mantlelike structures (1). The presence and lengths of hyphae emanating from the mantle are characteristic of different EMF species and establish different soil exploration types (2). It has been assumed that EMF communities are adapted specifically to mobilize sparse soil nutrient resources in boreal and temperate forests (11, 50). Current estimates indicate that about 80% of all nitrogen and phosphorus present in plants has been taken up via mycorrhizas (30, 41, 63).
Unlike free-living soil microbes, EM fungi have direct access to reduced carbon from their host plants. More than 50 years ago, Melin and Nilsson (46) showed that 14C applied to leaves was recovered within one day in EM fungi, suggesting a strong dependence of fungal metabolism on host photosynthesis. Subsequent isotopic studies corroborated tight connections between current photosynthate and EM fungi (28, 42). EMF hyphae constitute the main path of plant-derived carbon into the soil (24, 29). Furthermore, EMF hyphae contribute substantially to soil respiration (25% from hyphae and 15% from roots) (27). As hyphal respiration decreases strongly in response to girdling of trees, a tight metabolic link between extramatrical mycelia and host photosynthetic activity must exist (5, 9, 32). In addition, fruiting body formation of EMF species was strongly dependent on host photosynthetic capacity (32, 40). In contrast, the significance of the current assimilate supply for EMF colonization of root tips and for community composition is not yet well understood. Since trees contain substantial stores of carbohydrates in the roots and stem (7), it may be expected that EM fungi can be maintained if this carbon resource is available. For example, defoliation experiments with conifers, which restricted but did not eliminate current photosynthate transfer to roots, showed no effects on root EMF colonization. Massive defoliation that negatively affected aboveground biomass production suppressed morphotypes with thick mantles compared to those with thin mantles, suggesting a shift to less-carbon-demanding EMF species (14, 40, 44, 54, 56). Earlier studies also reported decreased EMF colonization of root tips (21, 52).
In a common garden experiment with young beech trees, strong shading over several years, which severely limited plant growth, suppressed EMF colonization and resulted in low EMF diversity (20). EMF community composition was affected strongly by shading and slightly by short-term girdling, suggesting that EMF taxa are sensitive to changes in plant internal carbohydrate resources (20). However, the overall EMF diversity was low, probably because the young trees were grown in nutrient-rich compost soil (20). The significance of photoassimilates for EMF abundance, diversity, and community composition, therefore, remains to be shown for adult forest trees, which usually have high EMF diversity and low nitrogen availability (10, 26, 53, 61).
The aim of this work was to test the hypothesis that EMF abundance and diversity are independent of the current photoassimilate supply and can be maintained by internal resources. To investigate this concept, old-growth beech trees (Fagus sylvatica L.) were girdled to suppress carbon allocation to roots. Since disruption of the current assimilate flux affects the carbohydrate source strength, we hypothesized that changes in EMF taxon composition would occur if EMF species had different carbon demands. Tree carbohydrate status, root demography, EMF colonization, and EMF taxon abundance were measured repeatedly during 1 year after girdling. Since girdling also affects carbon release into and probably nutrient uptake from soil, the influence of possible feedback by changes in the ratio of dissolved organic carbon to dissolved organic nitrogen (DOC/DON) in the soil on EMF diversity was also assessed.
The experimental site is located in the Swabian Jura in southwest Germany, 800 m above sea level (47°59′N, 8°45′E). Mean annual air temperatures, measured at 1.5 m above ground level (humicap HMP45D; Vaisala, Helsinki, Finland), in the study years 2006 and 2007 were 7.7°C and 7.0°C, respectively. The sums of annual precipitation, measured with a tipping bucket (ARG 100; Vaisala, Helsinki, Finland), in 2006 and 2007 were 729 mm and 824 mm, respectively.
Six plots each consisting of 5 adjacent trees were established in an 80- to 90-year-old beech (Fagus sylvatica L.) forest. Soil profiles were characterized as Rendzic Leptosols derived from limestone with high fractions of stones and rocks. pH values were measured in 0.01 mM CaCl2 and ranged between 6.5 and 7.1. For determination of total carbon and total nitrogen concentrations, soil of the Ah horizon was dried, sieved (<2 mm), and subjected to dry combustion using an elemental analyzer (Vario EL; Elementar Analysensysteme GmbH, Hanau, Germany). Inorganic carbon concentrations of samples were determined after ignition of samples for 4 h at 550°C. The concentrations of organic carbon were calculated as the difference between total and inorganic carbon. The Ah horizon is characterized by a high organic C content (11.0% ± 0.4%) and a high N content (0.8% ± 0.1%).
Soil temperatures were measured continuously at depths of 10, 30, 50, 100, and 200 mm by use of PT100 probes. Mean temperature data from these profiles are shown in Table Table1.1. Soil moisture was determined using two probes according to the time domain reflectometry (TDR) method (CS615; Campbell Scientific, Shepshed, United Kingdom). The probes, which have a sensor length of 0.3 m, were buried vertically in the ground, thus covering the uppermost 0.2 m of the slope parallel ground. The TDR soil moisture recordings were compared with gravimetrically taken soil moisture measurements to allow a soil-specific calibration. Soil moisture is indicated as means from both probes (Table (Table1).1). Further details of the site, stand, and soil have been described elsewhere (15, 33).
On 9 August 2006, on three of the six plots, five trees grouped as a pentagon (see the scheme in supplement S1 in the supplemental material) were girdled approximately 1.2 m above ground by complete removal of an approximately 83-mm-wide bark strip. The other three plots, each containing five trees grouped in a pentagon, served as controls. At each sampling date, one soil core of a volume of 1,004 ml (0.08-m diameter and 0.2-m height, which corresponds to the Ah horizon) was harvested at a distance of about 1 m from each of the selected trees, resulting in 15 soil cores per date and treatment. Since about 80% of fine-root biomass is present within a distance of 2 m of the stem (45), it was assumed that the majority of fine roots belonged to the closest tree. Soil cores were collected in October 2006, May 2007, August 2007, and October 2007 on each plot, yielding a total of 30 samples per harvest date. An exception was October 2006, when mixed samples were collected from each of the three plots with girdled trees. The soil cores were kept in polyethylene bags at 4°C until analysis.
The roots were carefully washed and spread out randomly, and four to six roots corresponding to a mean weight of 1.71 g (fresh mass) were randomly removed from the mixture. All root tips from this sample were counted using a stereomicroscope (Stemi SV 11; Zeiss, Jena, Germany). The number of root tips per sample ranged between 700 and 900. Live and dead roots were separated according to the method of Allen et al. (3). Vital and dead root tips colonized by EM fungi (termed vital ectomycorrhizal root tips and dead ectomycorrhizal root tips, respectively) were sorted as described by Downes et al. (19).
Different types of ectomycorrhizal fungi were morphologically classified following the procedure and identification keys of Agerer (1). Each morphotype was described and photographed (Coolpix 4500; Nikon, Tokyo, Japan). To calculate mantle surface, the photographs of the morphotypes and of a scale bar were used to assess the diameter (d) of the mycorrhizal root tips and length (l) of coverage with the hyphal mantle. Surface was calculated as d × π × l. EMF taxa were classified according to the method of Agerer (2) as contact type (no hyphae), short- to intermediate-distance-exploration type (hyphae), and long-distance-exploration type (rhizomorphs present) based on the description of the morphotypes (see supplement MT in the supplemental material). Ten to 20 root tips containing each morphotype were collected and kept frozen at −80°C until molecular identification.
About five to 10 root tips containing each morphotype were used for total genomic DNA isolation carried out with a plantDNA-OLS kit (OLS OMNI Life Science, Hamburg, Germany) according to the manufacturer's instructions. Before extraction, samples were ground in liquid nitrogen with a ball mill (type MM2; Retsch, Haan, Germany). Isolated DNA was used as a template for PCR amplification of the internal transcribed spacer (ITS) region of the fungal ribosomal DNA. We used the primers ITS5 (5′GGAAGTAAAAGTCGTAACAAGG3′) and ITS4 (5′TCCTCCGCTTATTGATATGC3′) according to the method of White et al. (64). PCR was conducted as described previously (20). PCR products were separated by electrophoresis on a 1.5% agarose gel supplemented with ethidium bromide and visualized under UV light (Fluor-STM multi-imager; Bio-Rad, Munich, Germany) as described by Sambrook and Russell (55). When no PCR products were found, the procedure was repeated if frozen morphotypes were still available. Single-band PCR products were purified for DNA sequencing using a straightPCR-OLS kit (OLS OMNI Life Science, Hamburg, Germany). When the PCR resulted in more than one amplification product, the PCR products with the expected sizes were cloned into a pGEM-T vector (pGEM-T system I; Promega, Madison, WI). Selection was performed with a blue-white assay on LB agar plates supplemented with ampicillin (100 μg ml−1). The transformants were analyzed for the presence of plasmid DNA by ITS region reamplification. PCR products of the expected sizes were purified by precipitation with isopropanol for 1 h at room temperature, followed by a 30-min centrifugation (17,900 × g, room temperature, centrifuge 5417 R; Eppendorf, Hamburg, Germany), and sequenced. Sequencing was performed via the dideoxy chain termination method using a BigDye Terminator cycle sequencing kit (Applied Biosystems, Foster City, CA) and an automatic sequencer (ABI Prism 3100 genetic analyzer, 36-cm capillary, Matrix Pop 6; Applied Biosystems, Foster City, CA).
Sequences were assembled using Staden Package 4.10 (http://staden.sourceforge.net). For fungal identification, BLAST searches were carried out against the NCBI (http://www.ncbi.nlm.nih.gov/) and UNITE (http://unite.ut.ee) public sequence databases. Sequences were assigned matching species names when the BLAST matches showed identities higher than 95% and scores higher than 800 bits. If no appropriate match was found, the sequence was assigned a higher-level taxonomic name or was called an uncultured EM fungus and numbered. A phylogenetic tree of nucleotide sequence alignments for the ITS regions was computed with MEGA 4.1 software (60). Phylogenies were inferred by the neighbor-joining method and tested by bootstrapping using 1,000 replicates.
Bark, phloem exudates, wood, and coarse and fine roots sampled in October 2006 and October 2007 were used for carbohydrate analyses. Wood comprising about 5 annual rings and bark were collected 50 mm above and below the girdle. Control samples were taken from nongirdled trees at 1.2 m above ground. Phloem exudates were obtained by incubation of fresh bark strips in 2 ml 10 mM EDTA solution for 5 h (22, 57). The resulting supernatant was stored at −20°C. Other materials were dried at 60°C and ground with a ball mill (MM2000; Retsch). One ml double-deionized water was added to 45 to 55 mg of sample, and the mixture was agitated at 4°C for 1 h, heated at 95°C for 10 min, cooled down to room temperature, and centrifuged (10 min, 12,000 × g). The supernatant was stored for sugar analysis, and the remaining pellet was stored for starch measurements.
The concentrations of fructose, glucose, and sucrose in phloem exudation solutions and in plant extracts were determined by high-performance liquid chromatography (HPLC) (Dionex, Idstein, Germany). Aliquots (200 μl) of the sample were diluted with 500 μl of double-deionized water. From this mixture, 100 μl was injected into the HPLC system. Sugars were separated using a CarboPac PA1 separation column (4 by 250 mm; Dionex), with 200 mM NaOH as the eluent and a flow rate of 1 ml min−1. All sugars were identified and quantified with external standards and are expressed in glucose equivalents per g of tissue.
Starch concentrations were determined using a commercial kit for starch analysis (no. 10 207 748 035; Boehringer-Mannheim, Darmstadt, Germany). Starch in the pellets was solubilized in dimethyl sulfoxide and HCl and hydrolyzed with amyloglucosidase. The resulting glucose concentration was determined enzymatically using a spectrophotometer at an absorbance of 340 nm (8).
Statistical analysis was performed using Statgraphics Plus 3.0 (StatPoint, Inc., St. Louis, MO). Experimental factors were treatment (girdled/control) and time point (sampling date). The following diversity indices were calculated: species richness, Hmax = ln (number of all species); Shannon-Wiener index, H′ = −∑ pi ln pi, where p is the probability of the species i; and evenness, H′/Hmax (58). The sampling unit “soil core” served as the basis for calculating these indices.
The distribution of fungal species in samples was tested by a χ2 test. The probability of finding a species on y root tips was calculated as P = 1 − (1 − x)y, where x is the abundance of the species in the community (62). Rarefaction was necessary since roots of control trees contained higher numbers of mycorrhizal root tips than roots of girdled trees. Individual-based rarefied species richness and H′ were calculated using EcoSim software version 7.72 (25). Pearson product moment correlations were calculated between each pair of variables as explained in the tables or text. Where means are shown, normal distribution of the data was tested by calculation of standardized skewness and standardized kurtosis. Since the values of these parameters ranged between −2 and +2, analysis of variance (ANOVA) was followed by a multiple-range test. Means were considered to be significantly different from each other when the P value was ≤0.05. Significant differences are indicated in the tables and figures by different letters. The effect of girdling and time on root demography and diversity indices was analyzed by repeated-measures analysis of variance by the General Linear Models (GLM) procedure of Statgraphics Plus 3.0 software using trees as the subject (StatPoint, Inc., St. Louis, MO). Regression analyses were plotted with the program Origin 7G (OriginLab Corporation, Northampton, MA). Detrended correspondence analysis (DCA) was used to visualize the similarity of EMF communities between treatments and sampling time points. The analysis was performed with R software version 2.9.2 (R Project for Statistical Computing) with the package “vegan,” using the relative abundance of fungal species per time point and treatment as the input matrix. For analysis of the β diversity of the EMF community composition, the Sørensen index was calculated using EstimateS software version 8.2.0 (12).
Nucleotide sequences have been deposited in the NCBI GenBank database under accession numbers FJ403484 to FJ403518, FJ823388 to FJ823391, FJ823393, FJ823394, FJ823396, and FJ823397.
Root tips of mature, healthy beech trees investigated during different seasons displayed the following four fractions: 55% ± 4% vital mycorrhizal root tips, 1.6% ± 1.1% vital nonmycorrhizal root tips (vital root tips not colonized by EM fungi), 31% ± 1% dry mycorrhizal root tips, and 15% ± 3% dry nonmycorrhizal root tips (dry root tips not colonized by EM fungi). These means showed some seasonal fluctuations (Fig. (Fig.1),1), which were apparently not related to the acute weather and soil conditions (Table (Table11).
Repeated-measures ANOVA revealed significant demographic changes in root tips of girdled trees compared with those of nongirdled control trees (Fig. (Fig.1).1). The fraction of dead mycorrhizal tips was about 20% higher at roots of girdled trees than at roots of control trees; the fraction of vital EM root tips was correspondingly decreased (Fig. 1A and C). The shift toward increased fractions of dead EM root tips was observed about 6 weeks after girdling and did not increase further in the following year. In contrast to the results for dead EM root tips (whose fraction was relatively stable), a time-dependent, approximately 3-fold decrease in dead nonmycorrhizal root tips was observed (Fig. (Fig.1D,1D, significant interaction of girdling × time). This result suggests differences in longevity of mycorrhizal and nonmycorrhizal fine roots and may indicate that the formation of fine roots became slower than the degradation.
When only vital root tips were considered, EMF colonization of roots of nongirdled trees was 97% ± 1% across all sampling dates (n = 28,060). Girdling had no significant influence on the fraction of vital EM root tips per total vital root tips (97% ± 2%, P = 0.939); however, fewer vital root tips were present (n = 13,435). The fraction of nonmycorrhizal vital root tips was always very small (Fig. (Fig.1B1B).
To investigate whether the persistence of EM fungi and high colonization rates for more than 1 year after girdling were related to the consumption of plant internal carbon storage pools, we determined concentrations of starch and soluble sugars in belowground and aboveground tissues (Fig. (Fig.2;2; Table Table2).2). Starch concentrations in coarse and fine roots of girdled trees decreased compared with levels found in controls (Fig. (Fig.2A).2A). After about 1 year, 25% of the maximum starch concentrations found in control roots was still present in roots of girdled trees. The concentrations of soluble carbohydrates (glucose, fructose, and sucrose) in coarse roots did not decrease significantly (Fig. (Fig.2B).2B). In contrast, in fine roots of girdled trees, reductions in soluble carbohydrates occurred at both sampling dates (Fig. (Fig.2B).2B). None of the carbohydrate pools became completely depleted.
The carbohydrate concentrations were also lower in phloem exudates of girdled trees than in those of control trees (Table (Table2).2). The decrease was stronger below than above the girdle, supporting the fact that basipetal transport of carbon was disturbed. Since the stem above and below the girdling site still contained considerable concentrations of soluble sugars and starch, continued transport to roots in the stem at low rates cannot be excluded. However, the transfer rate is very low (20).
To investigate whether girdling affected EMF community structures and diversity, we conducted morphotyping and ITS sequencing (GenBank accession numbers are listed in Table Table3;3; for morphotypes, see supplement MT in the supplemental material). Morphotypes for which ITS sequence information was not available because PCR products were not obtained are called MTs. ITS sequence information was obtained for 55 morphotypes that corresponded to 41 different taxa, indicating that morphotyping overestimated species numbers by a factor of 1.3 (Table (Table3).3). ITS sequences not belonging to known taxa were called uncultured ectomycorrhizal fungi (UEM) and numbered consecutively (Table (Table3).3). We constructed a phylogenetic tree to assign clade names to the UEM and to ascertain the annotation of species names obtained by BLAST (see supplement S2 in the supplemental material). The relative abundance of the sequenced taxa summed up to 90% root colonization for both control and girdled trees.
Assuming that each of the MTs represented one new species, we found a maximum of 89 EMF species at roots of control trees and 39 at roots of girdled trees. Rarefied species accumulation curves confirmed this massive species loss (see supplement S3 in the supplemental material). The use of 1.3 as the correction factor for species overestimation for nonsequenced morphotypes resulted in estimated numbers of 79 and 36 EMF taxa at roots of control and girdled trees, respectively. Since we considered these errors (which were in the range of 11% to 16%) relatively low, all unidentified MTs were considered different species for the following calculations and assessments.
The EMF species were sorted according to descending rank order for girdled trees (Fig. (Fig.3).3). Girdled trees hosted one new species (Cortinarius sp., which had the highest similarity with Cortinarius solis-occasus [rank 90 in Fig. Fig.3]).3]). The 50 species that had disappeared from roots of girdled trees together contributed 7.8% to the root colonization of control trees. We wondered whether the lost species, most of which occurred only with very low relative abundance at roots of control trees, had a reasonable chance to be observed, since roots of girdled trees contained fewer vital root tips than those of nongirdled trees (Fig. (Fig.1;1; also see number for root tips, y, in Fig. Fig.3).3). We estimated the theoretical probability of detection of each lost species according to the method of Taylor (62). As a prerequisite for this analysis, dispersion indices were calculated; these showed that the species were not clustered (see supplement S4a in the supplemental material). Among the 50 species that had disappeared, only nine species (ranks 80 to 89) had a detection probability of less than 99%. Furthermore, rare species showed random dispersion (see supplement S4b in the supplemental material). This indicates that, for the majority of the undetected species, disappearance was not a sampling effect.
It should be noted that our experimental design does not completely exclude introgression of roots from distant nongirdled trees. However, the observed species loss, together with the decrease in soluble carbohydrates (Fig. (Fig.2),2), the reduced number of vital root tips, the decrease in vital EM root tips, and the increase in dead EM root tips (Fig. (Fig.1),1), supports that the majority of the roots were from girdled trees.
In addition to species loss, pronounced changes in the relative abundances of several EMF species occurred at roots of girdled trees compared with the level for controls (Fig. (Fig.3).3). We considered increased or decreased all species that were outside the boundaries of a 99% confidence interval after linear regression of species abundance of EM fungi at roots of controls versus species abundance of EM fungi at roots of girdled trees (see supplement S5 in the supplemental material). Diminished abundances at roots of girdled trees were found for nine putative species (Tomentella pilosa, UEM [Russulaceae], UEM [Pezizales], UEM 11 [Basidiomycetes], UEM 9 [Ascomycetes], and MT 48, 62, 89, and 126 [rank numbers 28, 29, 30, 31, 33, 34, 35, 36, and 38 in Fig. Fig.3]).3]). Ten putative species occurred with higher abundances at roots of girdled trees than at those of control trees (Hysterangium nephriticum, UEM [Sebacinaceae], Tuber puberulum, UEM 5 [Sebacinaceae], UEM 12 [Ascomycetes], Russula mairei, UEM 10 [Sebacinaceae], UEM [Tricholoma], and MT 44 and 67 [corresponding to ranks 4, 9, 10, 12, 14, 15, 16, 17, 23, and 24 in Fig. Fig.3]).3]). To find out whether disappearance or increased abundance was related to the mantle thickness of the morphotypes, we estimated the surface of EMF species and related this to the change in abundance. However, no clear pattern was observed (see supplement S6 in the supplemental material). We also categorized EM fungi according to contact type (no hyphae), short- to medium-exploration type (hyphae present), and long-distance-exploration type (rhizomorphs present). Each category lost species, but there was no shift between the categories (see supplement S6 in the supplemental material).
To find out whether species were cooccurring or might have excluded each other, Pearson product moment correlations were calculated (see supplement S7 in the supplemental material). Since there was a large fluctuation in species between sampling dates (see below), we included only species that were present in at least half of the samples. We found negative correlations between Cenococcum geophilum and Tomentella lilacinogrisea and an unknown morphotype (UEM 2 [Phallales]). UEM 2 was positively correlated with Tomentella lilacinogrisea and with Tomentella pilosa (see supplement S7 in the supplemental material). Lactarius blennius showed a negative correlation with UEM 2. Negative correlations were also found for Tomentella lapida and Russula mairei, as well as for UEM (Hebeloma) and Tomentella subclavigera (see supplement S7 in the supplemental material).
We analyzed the influence of time after girdling on the Shannon-Wiener index (H′) and on species richness (S) to find out whether species loss was gradually progressing. Since girdling resulted in lower numbers of vital root tips (Fig. (Fig.1),1), H′ and S were analyzed using rarefied data, which showed that both indices were generally lower for girdled than for control trees (Fig. (Fig.4).4). A pronounced decline over time was not found. Instead, H′ and S of controls showed significant fluctuations, suggesting that the formation of a more diverse EMF flora was favored by intact trees in autumn than in other seasons. These fluctuations were not related to mean soil humidity, which was relatively stable at the different sampling dates (Table (Table11).
Evenness, a measure for the dominance structures of species, remained unaffected by girdling (Fig. (Fig.4).4). The relative stability of this parameter was caused by a group of seven major EMF species (Fig. (Fig.3)3) that together contributed 80% of ectomycorrhizal root tips of control and girdled trees. Among these species, Cenococcum geophilum was by far the most dominant fungus, with a relative abundance of about 40% (Fig. (Fig.3).3). Six additional species together contributed an additional 40% of colonization (Tomentella lapida, Lactarius blennius, Hysterangium nephriticum, UEM 2 [Phallales], UEM [Hebeloma], and Tomentella lilacinogrisea).
Detrended correspondence analysis showed that EMF communities were clearly separated according to seasonal fluctuations and by girdling along the first and second axes (DCA1 and DCA2), with Eigen values of 0.2465 and 0.196, which explained 39.6% and 31.6% of the proportions of variation, respectively (Fig. (Fig.5).5). The analysis indicates that the influence of seasonal fluctuations (DCA2) on community composition was larger than that of girdling (Fig. (Fig.5).5). Analysis of β diversity, measured as the Sørensen index, confirmed that the similarity of EMF communities between girdled and healthy trees (0.52 to 0.78) was greater than that between sampling dates (0.26 to 0.50) (see supplement S8 in the supplemental material).
We investigated whether the decrease in H′ was correlated with carbohydrate concentration in fine roots. Regression analysis showed a positive significant relationship between H′ and the concentrations of starch (P = 0.038), fructose (P = 0.003), glucose (P = 0.001), and the sum of the latter two carbohydrates in fine roots (Fig. (Fig.6A).6A). No correlation was found between sucrose levels and H′ (P = 0.156). The same was true for Hmax (not shown). Evenness showed no significant regression with any of the carbohydrates (not shown).
We also tested whether individual EMF species were significantly correlated with root carbohydrates. For this analysis, only data sets which contained a given species in at least half of the samples were used. Cenococcum geophilum decreased with increasing glucose concentrations (P = 0.050), whereas Tomentella pilosa (P = 0.009) and a species of the Russulaceae (P = 0.012), whose overall abundances were, however, low (range from 0 to 1.6%), increased. The latter two species also showed significant positive regression curves with fructose (P = 0.002 and P = 0.016, respectively).
The starch concentration was significantly correlated with glucose (P = 0.027) and fructose (P = 0.007) concentrations in fine roots, suggesting a tight connection between these pools. Soil nitrogen was relatively stable (total N, 7.7 ± 0.7 and 8.6 ± 0.6 mg g−1 soil [dry weight] on girdled and control plots, respectively; dissolved organic N, 29 ± 6 and 32 ± 6 μg g−1 soil [dry weight] on girdled and control plots, respectively), whereas dissolved organic carbon was lower in soil from the girdled plots than in that from nongirdled plots (16). H′ was positively correlated with DOC/DON (Fig. (Fig.6B)6B) but not with soil organic carbon (P = 0.178) or soil nitrogen (P = 0.638).
The main questions studied were how root colonization, EMF species composition, and EMF species diversity were affected when assimilate flux to the roots was suppressed by girdling. For the healthy trees of this study, the EMF species richness was similar to that for other stands of adult European beech (10, 13, 26, 53, 61) and showed a typical distribution, with few abundant and many rare species (4). EMF species composition fluctuated between sampling dates (Fig. (Fig.5),5), probably because of changing environmental conditions, such as temperature and soil humidity (10, 13, 59). After girdling, root colonization remained unaffected as expected because the internal carbon resources decreased but were not depleted during the time course of this study. This is an important finding because the rapid decreases in extramatrical hyphal respiration, which were observed with pine and beech after girdling (5, 32), might have indicated that EM fungi were cut off from the carbon supply chain.
However, despite the presence of significant carbohydrate pools in the root system and the evidence of EMF abundance, which suggests that there was still a carbon supply, the EMF diversity was strongly diminished. This is in contrast to previous results which determined that girdling of young beech trees affected neither EMF abundance nor EMF diversity (20). However, in the previous study the trees were grown in compost soil, where the typical EMF flora of forest trees did not develop and where EMF species richness was three to four times lower than that found in the present investigation. Johnson et al. (35) have discussed that plants may regulate the diversity and community structure of mycorrhizal fungi. Our study suggests that this may occur via the carbon supply, because diversity and species richness were correlated with glucose and fructose concentrations and also with starch in fine roots. The latter observation was surprising at first glance but may simply reflect the fact that soluble pools of fructose and glucose were fed directly by starch degradation. After girdling, the decrease in sucrose, which is the major transport form for carbon, was most pronounced; however, it was not correlated with diversity. This appears reasonable because EM fungi can use monosaccharides, such as glucose or fructose, but not the disaccharide sucrose as the carbon source for their nutrition (48).
Notably, girdling affected mainly subordinate taxa. It is possible that they disappeared as a direct consequence of their missing food source or that they were outcompeted by the dominant EMF species. The abundance of the most dominant fungus, C. geophilum, was inversely correlated with plant carbohydrates, which points to the low carbon demand and high competitiveness of this species. Since C. geophilum can also produce mycotoxins (6), it may shape EMF communities by affecting its competitors. Indeed, negative relations between C. geophilum and other EMF taxa have been observed previously (39). In our study, C. geophilum was negatively correlated with UEM 2 and Tomentella lilacinogrisea. However, we cannot distinguish whether C. geophilum suppressed these species or whether these species were reduced because the assimilate flow was abolished. The latter option cannot be excluded, because the abundance of T. lilacinogrisea was linked with root sugar concentrations.
Regardless of the underlying mechanisms, our data demonstrate the importance of recent photoassimilates for EMF species richness. This contention is also supported by the observation that tree seedlings grown close to undisturbed trees maintained higher EMF species richness and diversity than those grown close to dead stumps or wood debris (18, 34). The community of fungi close to stumps or wood debris was shifted toward EM fungi with saprotrophic features (34), in support of the idea of a biotrophy-saprotrophy continuum of EM fungi (38). However, we have no evidence for such shifts, probably because carbohydrate reserves had not yet been used up. Significant effects on EMF diversity of roots of vital stumps have also been observed after clear-cutting (36). However, this treatment affects microclimatic conditions, and therefore, the significance of recent photoassimilates for EMF diversity is unclear.
Changes in plant carbon production may affect soil properties (23, 31) and thus affect EMF diversity by feedback effects through changes in soil composition. High soil fertility caused by high nitrogen decreases mycorrhizal fungal diversity (43, 49). In our study, the influence of nitrogen was excluded because the total soil N content and dissolved organic nitrogen did not change in response to girdling, whereas the DOC content decreased after girdling (16). The decrease in DOC was probably related to decreased root exudation, because a substantial portion of the DOC was found to originate from photoassimilates (23). Therefore, the correlation between H′ and DOC/DON is unlikely to reflect a direct influence of properties of the soil solution on EMF diversity but suggests that both DOC and EMF species composition were governed by carbon allocation to roots.
An important result of our study was that maintenance of EMF species was selective. As the three most abundant species, C. geophilum, L. blennius, and T. lapida, were unaffected by girdling despite considerable differences in the mantle surfaces (see no. 1, 2, and 3 in supplement S6 in the supplemental material), our analysis does not support that voluminous EMF structures per se were more sensitive to reduced carbon flow than those with smaller surfaces. Both EMF taxa with rhizomorphs and EM fungi with smooth mantles disappeared more frequently than those with hyphae (see supplement S6 in the supplemental material). In general, there is no correlation between the amount of extramatrical mycelium of an ectomycorrhiza-forming fungal species and its abundance on root tips (37), and therefore the influence of a reduced carbon supply on soil-foraging EMF mycelia is difficult to assess. A preferential allocation of carbon to more-efficient extramatrical mycelia has been shown previously (51). While better service to the plant as afforded by higher outreach for nutrient foraging may be associated with fitness cost and lead to selective species loss, it must be noted that the reduced carbon flow in our study did not cause shifts between the different EMF categories, since only rare species disappeared. The independence of the functional categories from actual carbon flow suggests that the EMF community has adapted to maintain nutrient flow under the prevailing ecological conditions.
The disappearance of ca. 50 fungal taxa, which colonized only about 8% of the root tips, suggested that they thrive on an excess of recent photosynthate. These fungi may be a legacy for adaptation of the forest to changing environmental conditions. This would support the insurance hypothesis of diversity, because rare species may become important for maintaining the nutrient supply of beech when microclimatic or soil conditions change. Perhaps rare or undetected species at roots of healthy trees, such as Cortinarius sp. (similar to C. solis-occasus) and Tuber puberulum, which became more abundant at roots of girdled trees, fulfill such functions. The host range of these species must be wider than previously thought because they have so far been described as EM fungi only of conifers (17, 47).
The observed instability of many individual EMF species-host interactions in our study may indicate that beech is not the preferential host. One possible ecosystem benefit would be to facilitate the establishment of other tree species in future stages of the forest. This idea is supported by the fact that some of the rare taxa have previously been reported to form interactions with conifers (Humaria hemisphaerica and Hygrophorus discoxanthus) or with other angiosperms, like Carpinus betulus or Tilia sp. (Sebacina incrustans), but not with beech (17). In conclusion, we suggest that a surplus carbon “donated” to nonhost EMF taxa would prevent extinction of these communities and constitute an important ecosystem service of mature beech trees by maintaining an EMF reservoir.
Funding of this work by the German Research Foundation/Deutsche Forschungsgemeinschaft (DFG) within the framework of Beech Research Group 788 is gratefully acknowledged.
We thank Oliver Itzel and Saskia Knillmann for help with the sugar analysis and Katharina Platner for help with the morphotyping.
Published ahead of print on 22 January 2010.
†Supplemental material for this article may be found at http://aem.asm.org/.