Elevated atmospheric CO2 alters root symbiont community structure in forest trees

Authors

  • Petra M. A. Fransson,

    Corresponding author
    1. Department of Forest Mycology and Pathology, SLU, Box 7026, S-75007 Uppsala, Sweden
      Author for correspondence:Petra M. A. Fransson Tel: +46 18 672797 Fax: +46 18 309245 Email: petra.fransson@mykopat.slu.se
    Search for more papers by this author
  • Andrew F. S. Taylor,

    1. Department of Forest Mycology and Pathology, SLU, Box 7026, S-75007 Uppsala, Sweden
    Search for more papers by this author
  • Roger D. Finlay

    1. Department of Forest Mycology and Pathology, SLU, Box 7026, S-75007 Uppsala, Sweden
    Search for more papers by this author

Author for correspondence:Petra M. A. Fransson Tel: +46 18 672797 Fax: +46 18 309245 Email: petra.fransson@mykopat.slu.se

Summary

  •  Changes in below-ground ectomycorrhizal (ECM) community structure in response to elevated CO2 and balanced nutrient addition were investigated in a 37-yr-old Picea abies forest.
  • Trees in whole-tree chambers were exposed to factorial combinations of ambient/elevated CO2 (700 ppm) and fertilization (+/–). ECM fungal community structure was determined in 1997 and 2000 using a combination of morphotyping and molecular analyses. Samples were taken both from chambers and from reference trees receiving the same fertilization treatments but without chambers.
  • Significant effects on ECM community structure were found in response to elevated CO2. Neither elevated CO2 nor fertilization altered species richness; however, there was considerable variation among samples, which may have masked treatment effects on individual species. After 3 yr, the effects of elevated CO2 on community composition were of the same magnitude as those seen after 15 yr of fertilization treatment.
  • Our results show that increasing atmospheric CO2 concentrations affect the community structure of root symbionts colonizing forest trees. The potential effects of altered ECM community structure on allocation and turnover of carbon and nutrients within forest ecosystems are discussed.

Introduction

Within European forests there is a fine balance between the quantity of carbon (C) fixed via photosynthesis and that released through respiration (Valentini et al., 2000). The result of these two processes determines the net exchange of C between the terrestrial biosphere and the atmosphere and whether forests act as a source or a sink for CO2. The balance is affected by daily, seasonal and yearly variations in climatic conditions (Hanson et al., 1993; Kirschbaum, 1995; Goulden et al., 1998; Lindroth et al., 1998), and by changes in land use patterns (Schimel, 1995; Falkowski et al., 2000). Valentini et al. (2000) showed that total ecosystem respiration, most notably that of roots and soil microorganisms, was the main determinant of the C balance of a forest. Among soil microorganisms, symbiotic mycorrhizal fungi are of particular significance since they have a direct influence on both the sequestration and emission of carbon. Boreal forest trees rely on ectomycorrhizal (ECM) fungi for their nutrient uptake, and almost 100% of forest tree short roots are colonized by ECM fungi (Taylor et al., 2000). The fungi, in turn, depend strongly on current assimilates from their plant hosts (Högberg et al., 2001), and up to 20% of the photosynthetically fixed carbon may be allocated by the tree to the fungus (Finlay & Söderström, 1992). Högberg et al. (2001) found a large and almost immediate decrease in root associated respiration, which constituted 54% of the total soil respiration, in combination with a large decrease in ECM fruit body production, following girdling of a pine stand illustrating the importance of current assimilates to ECM fungi.

Individual ECM fungal species are likely to differ in how they allocate plant derived C between growth, respiration and exudation. The allocation patterns can in turn be affected by the actual supply of C from the host plant to the fungus. Increased levels of CO2 in the atmosphere are known to increase plant photosynthesis and subsequent C supply into the soil (O’Neill, 1994; Rey & Jarvis, 1997; Ceulemans et al., 1999). Whether this increased supply of C below-ground actually leads to significant, long-term net carbon sequestration in forest soils is however, questioned (Schlesinger & Lichter, 2001). Laboratory studies of tree seedlings have demonstrated that Suillus bovinus and Laccaria bicolor respond differently to elevated CO2 concentrations in the way they partition assimilates between fungal biomass and respiration (Gorissen & Kuyper, 2000). These authors also showed that the fungi differed in their capacity to process and transfer N to their tree hosts. In general, diversity in physiology and function – both among and within different fungal species – might be large (Cairney, 1999), but the dominance of a few species is commonly seen in investigations of below-ground ECM community composition (Gardes & Bruns, 1996; Erland et al., 1999; Peter et al., 2001). Changes within these dominant species thus have a potentially large influence on any physiological response changing community structure. The results of Gorissen & Kuyper (2000) also underline the importance of nutritional effects and their likely interactions with C allocation. Changed ECM community structure may thus contribute to changes in the C balance of boreal forest ecosystems. It is therefore important to establish how individual species respond to elevated levels of CO2 under field conditions.

There are few published studies of the effects of elevated CO2 on ECM mycorrhizal community structure. Rey & Jarvis (1997) found indications that the mycorrhizal species composition based on morphology (morphotypes) shifted towards those species characteristic of later successional stages in young birch seedlings (Betula pendula) after growing in open-top chambers under elevated levels of CO2 for 4.5 yr. They interpreted this as an acceleration of tree ontogeny, which may lead to the trees supporting ECM fungal species with a higher carbon demand. Godbold et al. (1997) transferred paper birch (Betula papyrifera) and Eastern white pine (Pinus strobus) saplings from a forest to growth chambers and observed, after 24 wk of elevated CO2 exposure, an increase in the proportion of morphotypes that produced rhizomorphs and large amounts of extraradical mycelium. Rygiewicz et al. (2000) planted 2-yr-old Douglas fir (Pseudotsuga menziesii) seedlings in chambers, and after 4 yr of elevated CO2 exposure they found small overall effects on ECM fungal diversity, based on gross morphological traits. The authors did find an increase in ECM root tip proliferation during summer. Markkola et al. (1996) were unable to detect any significant changes in ECM morphotype community structure growing on roots of Pinus sylvestris as a result of CO2 enrichment. The authors used pine seedlings, which were inculated with Piloderma croceum, and then grown in pots with natural mor humus for up to 11 months. As far as we know, there are no studies of the effects of elevated CO2 on ECM fungi associated with large forest trees growing under field conditions. In the present study we investigated changes in the ECM fungal community structure in a 37-yr-old Norway spruce forest after factorial combinations of elevated CO2 and balanced nutrient addition.

Materials and Methods

Study site

The study was conducted at the Flakaliden field site, northern Sweden (64°07′ N; 19°27′ E; alt. 310 m above sea level (asl)). The site was planted with Picea abies[L.] Karst. after clear-felling in 1963. A fertilization experiment was started in 1987 in which plots were supplied with a balanced solution of macro- and micronutrients on a daily basis throughout the growing season. Since the start of the experiment the fertilization treatment has included a total of 1125 kg N ha−1, added as both ammonium and nitrate. In 1996, 12 whole tree chambers were installed around trees. Six of the chambers were placed in an unfertilized plot and six were placed in a fertlized plot. The CO2-treatment started in 1998 and continued for 3 yr. One half of the enclosed trees on each plot received ambient levels of CO2 (350 ppm), the other half received elevated levels of CO2 (700 ppm). The enclosed trees are hereafter referred to as ambient trees and elevated trees. Three reference trees (without chambers) on each plot were also included in this analysis. The enclosed trees were irrigated via a tubing system with amounts of water equivalent to those intercepted by the same area outside the chambers. The mean distance between individual trees in the plots was c. 3 m. Two of the chambers (ambient tree no. 2 in the unfertilized plot and ambient tree no. 2 in the fertilized plot) were moved to new trees in 1997 and 1998 as the trees died as a result of aphid and fungal attacks.

Sampling and identification of ectomycorrhizal fungi

In late August in 1997, before the CO2-treatment started, and at the same time in 2000, before the final harvest of the chamber trees, five soil cores (2.8 cm diameter) were taken from the organic soil layer beneath each of the chamber trees (6 + 6) and the reference trees (3 + 3). Forty-five soil cores were thus taken from each plot, giving a total of 90 soil cores. Soil samples were placed in separate plastic bags and maintained at 4°C, until examined. The depth of the organic layer in each core was measured before soaking the sample in water for 30 min. Live root tips were then extracted by wet sieving using a combination of 1.0 mm and 500 µm sieves. Root tips were examined and classified into mycorrhizal morphotypes, using macroscopic and microscopic features (Agerer, 1986–98). The term morphotype is used here to designate a recognizable group of mycorrhizal root tips. Morphotypes were either identified to species or genus level, or unidentified. The unidentified morphotypes were given numbers as they appeared in the processing. The total abundance (total number of mycorrhizal root tips per core) and relative abundance (number of each morphotype/total no. of living mycorrhizal root tips) of each mycorrhizal type were recorded. Nonmycorrhizal root tips were also recorded.

Subsamples of morphotypes were kept in the freezer for internal transcribed spacer-restriction fragment length polymorphism (ITS-RFLP) analysis as a supplement to the morphotyping to confirm consistency within and between single morphotypes. DNA was extracted from the mycorrhizal root tips according to Gardes & Bruns (1993), excluding the initial freeze-thawing step. The ITS region of the rDNA was amplified by PCR following the modified protocol by Henrion et al. (1994). PCR was performed using High Fidelity DNA polymerase and High Fidelity buffer (Roche). The reaction mix had final concentrations of each nucleotide at 0.2 mM, each primer at 0.3 µM, polymerase at 0.026 µ µl−1, Mg2Cl at 3.1 mM. DNA template was added as 25% of the final reaction volume. Cycling parameters were modified to 1 cycle at 94°C for 3 min followed by 35 cycles of 30 s at 94°C, 45 s at 50°C, 1 min at 72°C ending with 1 cycle of 7 min at 72°C. The primers used were ITS1-ITS4 as described by White et al. (1990). PCR products were separated on a 1% agarose gel in 0.5 X TBE-buffer at 4.7 Vcm−1, stained with ethidium bromide and visualized by UV-light. The amplified DNA was digested using the restriction enzymes Hinf1, Mbo1 and Taq1 (Promega Corporation, Madison, WI, USA), for 2.5 h in 37°C (Hinf1, Mbo1) and 65°C (Taq1). Digestion products were separated on a 2.3% Metaphor gel in 0.5 X TBE-buffer at 4.7 Vcm−1, stained and visualized as previously. Fragment lengths were estimated and band patterns analysed with Taxotron® software system (Institute Pasteur, Paris, France).

Statistical analysis

To investigate community structure and possible treatment effects, Detrended correspondence analysis (DCA) and Canonical correspondence analysis (CCA) were carried out on the total numbers of mycorrhizal root tips in each morphotype (ter Braak & Smilauer, 1998). DCA represents an indirect gradient analysis, based on the morphotype abundance data, that examines the total variation in the community samples. CCA represents a direct gradient analysis, where the morphotype abundance data is explained by environmental variables and part of the total variation in the community samples is thus examined. In this case the environmental variables are the treatments, that is no fertilization – fertilization, no chamber – chamber and ambient CO2– elevated CO2. An interyearly comparison of community structures was made by DCA analysis. The two ambient trees where the chambers were moved between the two samplings were excluded from the analysis. Arrows are used in the DCA ordination graph to indicate the shift in fungal community structure between 1997 and 2000. Data were transformed (log10) before analysis, and the analyses were performed with CANOCO 4.0 (ter Braak & Smilauer, 1998). Monte Carlo permutation tests (n = 199) were performed to test the significance of the relationship between morphotype data and treatments. In the CCA ordination diagram vectors are included to make the presentation clearer, although these are not strictly applicable since the environmental variables are not continuous.

Possible treatment effects of elevated CO2 and chambers on ECM morphotype richness, total root tip numbers and individual ECM morphotype abundance (total) were analysed for each plot using one-way ANOVA. Morphotype richness was tested using both the numbers of identified morphotypes, that is taxa identified to species or genus level, and the total numbers of morphotypes, that is both identified and unidentified taxa. The data from 2000 were tested for both the elevated CO2 and the chamber factors, while the data from 1997 were only tested for the chamber factor.

To test whether the effects of CO2 and chamber treatments were the same for both fertlized and unfertilized plots, that is whether the possible effects of CO2/chamber and fertilization treatments upon individual components of the community structure interact, a two-way ANOVA was performed. Since the fertilization treatment was not replicated, the significance levels for the possible effects of the fertilization treatment alone upon individual components cannot be used. It is however, statistically correct to test the interaction and this part of the results from the two-way ANOVA is reported (B. Vergerfors-Persson, pers. comm.). The significance levels for the possible effects of CO2/chamber treatment upon individual components are not reported since they were already tested through one-way ANOVA as described in the above section.

Mean values ± standard errors for total numbers of morphotypes, number of identified and unidentified morphotypes and levels of colonization are reported.

Results

Ordination analyses of community structure

The Canonical correspondence analysis showed significant effects of elevated CO2 on ECM fungal community structure in the year 2000 (Fig. 1a). 20.9% of the variation in the species data set can be attributed to the fertilization-, chamber- and CO2-treatments. The eigenvalues for CCA axes 1 and 2 were 0.371 and 0.303. The first two CCA axes together display c. 15% of the total variance in the species data set, and this part of the variation is shown in Fig. 1(a). The Monte Carlo permutation tests showed that there were statistically significant (P = 0.005) relationships between ECM morphotype composition and environmental variables. When the test was repeated for one environmental variable at a time, the results were also significant for fertilization and CO2 effects, but not for the chamber effect.

Figure 1.

Ordination diagram based on Canonical correspondence analysis (CCA) of the ectomycorrhizal (ECM) community in a 37-yr-old Norway spruce stand subject to elevated CO2 (700 ppm) and optimal fertilization for 3 and 15 year, respectively. The diagram shows the part of the variation within the communities (21%) that can be attributed to the treatments. Arrow length indicates the relative importance of environmental variables. (a) Open symbols, unfertilized trees; closed symbols, fertilized trees. Ambient CO2 trees with (open diamonds) and without (open circles) chamber and elevated CO2 with chamber (open diamond within square). (b) Species scores for ECM taxa that appeared on three or more of the sampled trees. Abbreviation: A. byssoides, Amphinema byssoides; C. geophilum, Cenococcum geophilum; P. gelatinosa, Piceirhiza gelatinosa; P. rosa-nigrescens, Piceirhiza rosa-nigrescens; P. byssinum, Piloderma byssinum; P. croceum, Piloderma croceum; Russula spp.; tomentelloid group 1, clamped tomentelloids; tomentelloid group 2, unclamped tomentelloids; T. fibrillosa, Tylospora fibrillosa; Unknown, unknown basidiomycete.

The positioning of the samples relative to the end point of the vectors in the CCA ordination diagram relates individual trees to different treatments (Fig. 1a). The ordination diagram shows the elevated CO2 trees clustered together in the upper left part. For the unfertilized plot, elevated CO2 trees are clearly separated from the ambient CO2 trees and reference trees. The fertilized plot shows a similar pattern, although the fungal communities on trees receiving elevated CO2 levels were not so clearly separated from ambient and reference trees.

Eleven taxa occurred on three or more of the trees examined and the CCA species scores are plotted together with the environmental variables in Fig. 1(b). Of these, Amphinema byssoides, Cenococuim geophilum, tomentelloid group 1 and Tylospora fibrillosa, all seemed to be favoured by the fertilization treatment. Piloderma croceum and tomentelloid group 2 are grouped near the origin and between the two plot treatment clusters, and can thus not be related to any of the treatments. Piloderma byssinum seemed to be negatively affected by the fertilization treatment, and the unknown basidiomycete appeared to be favoured by the CO2 treatment.

In 1997, before the CO2 treatment started, the CCA analysis showed a clear separation between the unfertilized and fertilized plots, and 14.4% of the total variance of the species could be explained by the fertilization and chamber treatments (figure not shown). The significance of the treatments was tested as previously and the ECM morphotype distribution was significantly (P = 0.01) related to fertilization and chamber treatments. Variation in community structure in the unfertilized plot was larger compared to the variation in the fertilized plot. Eigenvalues for CCA axes 1 and 2 were 0.50 and 0.38, respectively. Seven taxa in 1997 occurred on three or more of the trees. Cenococcum geophilum, T. fibrillosa and tomentelloid group 1 were favoured by the fertilization treatment, while Piloderma species and Inocybe spp. were mainly associated with the trees in the unfertilized plot.

Inter-yearly variation in community composition

A comparison of the ECM community structure between years was made using DCA analysis. The results showed that the composition of fungi colonizing tree roots in the unfertilized plot remained different from that in the fertilized plot (Fig. 2a), with an exception. The elevated CO2 treatment seemed to drive the community structure in the unfertilized plot towards that in the fertilized plot (Fig. 2a). This is explained by changes in the relative abundance of some taxa, but the specific changes, that is which of the fungal taxa are responsible for the shift in community structure, are not easily distinguishable. In this analysis no apparent shift in community structure was seen as a result of the CO2 treatment in the fertilized plot. All trees in the fertilized plot are grouped together in the DCA diagram (Fig. 2b). Despite the variation in ECM community structure between years, especially for the reference trees in the unfertilized plot, ambient trees and reference trees in the unfertilized plot remain separated from the fertilized plot. In general the variation both within and between years was considerable.

Figure 2.

Ordination diagram based on Detrended correspondence analysis (DCA) of the ectomycorrhizal (ECM) community in a 37-yr-old Norway spruce stand, showing the interyearly variation between 1997 and 2000. The Norway spruce stand was subject to elevated CO2 and optimal fertilization treatments. The direction of arrows in the ordination diagram indicates the shift in community structure for individual trees between the two years, and long arrows indicates larger differences than small arrows. (a) All trees in the unfertilized plot remained separated from the trees in the fertilized plot, with one exception. The fungal community structures of elevated trees, that are marked in red, seem to become more similar to those of the fertilized plot (b). The division between the unfertilized plot and fertilized plot is indicated by the dashed line. Red/blue, elevated trees in unfertilized/fertilized plot; orange/turqoise, ambient trees in unfertilized/fertilized plot; black, reference trees.

Effects of elevated CO2 and fertilization treatment on individual components of the community structure

In 1997, a total of 58 ECM morphotypes was found, 14 which were identified to species or genus level and 44 that were unknown (Table 1). In 2000, a total of 42 morphotypes was found, 14 were identified and 27 were unknown types (Table 2). For all investigated trees in 2000, the number of identified ECM morphotypes per tree (5.3 ± 0.5) exceeded the number of unknown morphotypes (1.7 ± 0.4). In 1997, the number of unidentified morphotypes was higher (2.7 ± 0.5), and the number of identified morphotypes lower (3.7 ± 0.3). Between 1 and 14 morphotypes were found per tree with a mean value of 6.4 ± 0.8 morphotypes per tree in 1997 and 7.1 ± 3.0 in 2000. ITS-RFLP analyses showed satisfactory consistancy between and among tested morphotypes.

Table 1.  Relative abundance of ectomycorrhizal (ECM) morphotypes, sampled in 1997 from Norway spruce plots receiving no treatment or fertilization. Individual trees were either covered with whole tree chambers, not recieving CO2 treatment in 1997, or free-standing reference trees receiving no additional treatment
Tree treatment Replicate tree no.Control plotFertilization plot
Chamber treesReference treesChamber treesReference trees
12134561231223456123
  1. Abundances are expressed as percentage of root tips examined. 1,2The chamber was moved to a new tree in 1997 and 1998, respectively.

Amphinema sp.3.8000000000000001.400
Cenococcum008.814.30< 11.814.31.903.614.91.27.512.119.68.614.7
Cortinarius spp.003.0064.2025.59.3< 100000001.70
Dermocybe spp.00000< 101.8000000003.42.0
L. rufus025.0000034.100000000000
Lactarius spp.88.500000< 100< 1040.40011.0000
Piceirhiza bicolorata0000008.1000000001.400
P. rosa-nigrescens00000003.30000000000
P. byssinum00011.10005.024.50000008.700
P. croceum0010.001.104.19.09.20010.606.821.1000
Russula spp.075.00000000000000000
Tomentelloid007.100000085.204.210.817.35.318.107.3
Tylospora sp.000010.50003.24.988.06.471.148.1034.167.443.3
Unknown tips7.7014.012.721.168.911.651.049.606.06.4010.545.816.718.910.7
Nonmyco. tips0057.161.93.129.914.36.311.19.62.417.116.99.84.70022.0
Total no. tips785617063952414403352163121674783133190138175150
Types id.225232566324344644
Types unid.101322518022054336
Colonization (%)10010042.938.196.970.185.793.788.990.497.682.983.190.295.310010078.0
Table 2.  Relative abundance of ectomycorrhizal (ECM) morphotypes, sampled in 2000 from Norway spruce plots receiving no treatment or fertilization. Individual trees were treated with ambient or elevated levels of CO2 in whole tree chambers, or were free-standing reference trees recieving no additional treatment
Tree treatment Replicate tree no.Control plotFertilization plot
AmbientElevatedReference treesAmbientElevatedReference trees
12131231231223123123
  1. Abundances expressed as percentage of root tips examined. CO2 concentrations: ambient (350 p.p.m) and elevated (700 p.p.m). 1,2The chamber was moved to a new tree in 1997 and 1998, respectively.

Amphinema sp.000004.506.019.926.91.4008.101.06.58.7
Cenococcum0011.103.710.53.713.13.720.1018.035.018.310.231.48.821.4
Cortinarius spp.69.2010.20010.95.66.125.01.5000001.000
Dermocybe cf. cinnamomeus00000006.6000000000< 1
Dermocybe spp.000001.4< 100000000000
L. deterrimus000000003.2000000000
Lactarius spp.0000025.0000000001.3000
Piceirhiza gelatinosa00007.41.40000014.610.0031.56.9< 12.4
P. rosa-nigrescens006.500001.62.8021.40000000
P. byssinum080.0005.1010.003.700008.60000
P. croceum11.60020.54.4020.7000012.60043.40021.0
Russula spp.00000< 1< 105.1000000000
Tomentelloid0020.456.80003.8006.9< 101.11.30< 11.6
Tylospora sp.15.4050.004.44.151.812.025.942.360.745.140.040.98.820.551.837.7
Unknown3.820.01.89.174.334.57.649.710.79.29.68.715.021.53.538.217.66.8
Non-myco. tips00013.6< 17.301.1000< 101.501.014.10
Totals (no. tips)2651084413622030118321613014520620186226102170252
Types is316358788446357558
Types unid101156022222031111
Colonization (%)10010010086.410092.710098.910010010010010098.510099.085.9100

The level of colonization was lower and more variable in 1997, with a mean of 85.2% ± 4.3, compared to 2000 when the mean colonization level was 97.9% ± 1.1. ECM morphotype richness and total root tip numbers were generally not affected by the treatments, with a few exceptions (see below). In Tables 1 and 2, the relative abundances of Norway spruce root tips colonized by each ECM morphotype are listed for each tree. Unknown morphotypes were assigned numbers as they occurred during processing, and the category called tomentelloid fungi includes both Tomentella and Pseudotomentella species, that is both clamped and unclamped tomentelloids. No individual ECM morphotype, except P. croceum and C. geophilum (see below), showed a significant response to the different treatments in either of the years. This apparent lack of response by individual ECM morphotypes to treatments was mainly due to the high variability within the data.

Significantly (F1,7 = 6.27, P= 0.04) more nonmycorrhizal root tips and unidentified ECM morphotypes (F1,7 = 7.00, P= 0.03) were found on the elevated CO2 trees compared to ambient trees and reference trees in the unfertilized plot in 2000. No significant effects of elevated CO2 were detected for the fertilized plot.

Significantly fewer root tips were found in the chambers in the unfertilized plot both in 1997 (F1,7 = 12.25, P= 0.01) and in 2000 (F1,7 = 7.17, P= 0.03). Significantly fewer identified ECM morphotypes (F1,7 = 15.75, P= 0.005) and fewer Piloderma croceum root tips (F1,7 = 16.99, P= 0.004) were found in the chambers compared to reference trees in the unfertilized plot in 1997. Significantly fewer Cenococcum geophilum root tips (F1,7 = 6.08, P= 0.04) were found in the chambers in the fertilized plot in 1997.

Interactions between elevated CO2 and fertilization treatments

In 1997, significant interactions were found between the fertilization treatment and the chamber treatment for the total number of root tips (F3,14 = 6.69, P= 0.02) and for root tips colonized by P. croceum (F3,14 = 10.91, P= 0.005). No significant interactions were found between elevated CO2 treatment and fertilization treatment in 2000.

Discussion

Increased levels of CO2 in the atmosphere are known to affect both host plants (Saxe et al., 1998; Ceulemans et al., 1999; Norby et al., 1999) and mycorrhizal fungi (Hodge, 1996; Fitter et al., 2000; Treseder & Allen, 2000). The effects of elevated CO2 on ECM community structure is important to establish since individual ECM fungal species may differ greatly in both physiology and function (Cairney, 1999). In our study the ECM community structure of Norway spruce trees changed in response to elevated levels of CO2. Canonical correspondence analysis was performed in order to compare the response of the fungal community as well as single ECM taxa to elevated CO2 and fertilization. Almost 21% of the total variation in the species data set could be attributed to the CO2 and fertilization treatments, reflecting the fact that the fungal community composition was affected by other environmental variables than the present treatments. Using CCA analysis, this is as large a part of the total variation one can normally expect to explain (ter Braak & Verdonschot, 1995). High variablility among individual samples was obvious here and is a general feature of ECM community data (Horton & Bruns, 2001), but despite this a significant effect of the CO2 treatment could be detected. Samples from individual trees and chambers may have contained tree roots from other, nearby trees, but we consider the samples to have been representative of the conditions experienced by the trees. Changes in ECM community structure due to elevated CO2 levels have been reported previously for seedlings and saplings (Godbold et al., 1997; Rey & Jarvis, 1997; Rygiewicz et al., 2000), but not for large forest trees grown under field conditions as far as we are aware.

The shift in community structure in the present study was mainly due to a change in abundance of a few common species. Similar responses were reported by Godbold et al. (1997) for paper birch saplings exposed to elevated levels of CO2, where the frequencies of dominant morphotypes changed significantly. Responses to elevated CO2 in the present study were stronger in the unfertilized plot, compared with the fertilized plot, as shown in Fig. 1. This was possibly due to the same ECM morphotypes responding in a similar way both to elevated CO2 and fertilization, thus making the actual effect of CO2 on community composition in the fertilized plot less pronounced. Fifteen years of fertilization may already have shifted the community in the fertilized plot so that the CO2 treatment had little effect once started. The DCA ordination diagram of interyearly differences in ECM community structure showed that morphotype composition for two out of three elevated trees in the unfertilisated plot became more similar to those found in the fertilized plot, indicating such a response.

In the present study the morphotype richness was high and this is in accordance with previous reports on below-ground ECM diversity (Erland et al., 1999; Rygiewicz et al., 2000; Peter et al., 2001). Richness was not affected by elevated CO2 in accordance with the studies by Rygiewicz et al. (2000) and Godbold et al. (1997). Rey & Jarvis (1997) did not report on the total numbers of taxa found, or the abundance of these, in their study of CO2 effects on ECM fungi. In the present study, balanced fertilization also had no effect on morphotype richness. Different fertilization treatments has earlier been shown to either decrease ECM richness (Alexander & Fairley, 1983; Rühling & Tyler, 1991; Peter et al., 2001) or not to affect richness at all (Kårén & Nylund, 1997; Jonsson et al., 1999).

Godbold et al. (1997) suggested that birch saplings grown under elevated CO2 levels could support mycorrhizal species with a higher C-demand. Twelve morphotypes were found by Godbold et al. (1997) on paper birch and Eastern white pine, with an average of six morphotypes per individual birch sapling. They observed an increase in the number of morphotypes forming extensive extraradical mycelia and rhizomorphs. Increased production of mycelium due to elevated CO2 has been reported for several different ECM species (Ineichen et al., 1995; Rouhier & Read, 1998a) as well as for arbuscular mycorrhizal (AM) fungi (Rouhier & Read, 1998b; Sanders et al., 1998). The increase in fungal biomass is sustained by the extra C fixed by and supplied from the host plant under enriched CO2 regimes (Godbold et al., 1997). However, the turnover of mycelium is likely to be faster compared to roots, and an increase in the amount of mycelium could lead to a faster turnover of C within the plant-fungus system (Fitter et al., 2000). In the present study it was impossible to say what happened with the extraradical mycelial production.

Rey & Jarvis (1997) found indications that the mycorrhizal species composition in birch shifted towards later successional stages after CO2 treatment, and they interpreted this as an acceleration of tree ontogeny. This may lead to the trees supporting ECM fungal species with a higher carbon demand, in accordance with the conclusions of Godbold et al. (1997). Rey & Jarvis (1997) also reported that trees grown in elevated CO2 invested more C into fine roots, and did not experience any decline in nutrient concentrations in plant tissues. The authors interpret the increased fine root density as likely to have met the nutrient demands of the trees. In addition, the above mentioned increased production of mycelia under elevated CO2 may also contribute to more efficient nutrient acquisition.

Due to the dry conditions in the chambers in our study, the number of root tips decreased compared to reference trees, although attempts were made to irrigate. This also affected the colonization levels to some extent (see below), but we may consider that the irrigation problem should not have affected the overall response of the ECM community structure to elevated CO2 because in the CCA ordination diagram, the chamber effect was small and not significant. Irrigation has previously been shown not to affect the community composition at the Flakaliden field site (Fransson et al., 2000), since the site is not water limited. Water limitation may influence the response of host plants and associated mycorrhizal fungi to increasing CO2, as reported for longleaf pine (Runion et al., 1997).

An increase in colonization of ECM root tips due to elevated levels of CO2 has often been reported in the literature. This has been demonstrated both from different types of pot studies (O’Neill et al., 1987; Godbold & Berntson, 1997) and from studies conducted in field soil (Tingey et al., 1997), although Lewis et al. (1994) and Walker et al. (1997) found no effect. The degree of ECM root tip colonization in boreal forests, however, is often close to 100% (Taylor et al., 2000) and in those types of systems elevated CO2 will have no effect on colonization levels. The response of seedlings germinated and grown according to nursery practices or other seminatural growth conditions might be expected to differ from those seen with rooted trees in a more undisturbed environment. What may happen in the field is that the production of fine roots increases as a result of elevated CO2, fine roots being more responsive to CO2 compared to the rest of the root system, but it is not clear if this response would persist (Norby et al., 1999). Norby et al. (1999) did not find any support for an increased root : shoot ratio in their analysis of open-top chamber field experiments. In the present study the degree of colonization in 1997 before the CO2 treatment started was variable, due to irrigation problems in some of the chambers. Increases in the numbers of nonmycorrhizal root tips can be observed after dry periods and subsequent addition of water when the root tips start to proliferate. In the sampling in 2000 all levels of colonization were high as expected, and in an earlier study at the same field site high levels of colonization were also observed (Fransson et al., 2000).

Elevated CO2 affected few individual components of the ECM community, only nonmycorrhizal root tips and unidentified ECM morphtypes in the unfertilized plot increased significantly with CO2 treatment. The high variablility among samples may have masked significant effects of CO2 on individual taxa. The below-ground ECM community is often dominated by a few common species (Gardes & Bruns, 1996; Erland et al., 1999; Peter et al., 2001) with a long tail of rare species in the distribution curve. Although there is still little known about differences in carbon use efficiency between ECM fungal species, there may be differences in their partitioning of C, and in their response to elevated CO2 concentrations (Dosskey et al., 1990; Rouhier & Read, 1998a; Gorissen & Kuyper, 2000). Changes in abundance of dominating fungi may thus affect the C-allocation pattern of the ECM community.

Variation in ECM community composition both within seasons and among years has been reported (Malajczuk & Hingston, 1981; Wu et al., 1993), and was also evident in the present study. Rygiewicz et al. (2000) found seasonal patterns in the number of ECM root tips and in two of the dominant morphotypes (the Rhizopogon group and C. geophilum) to respond differentially to CO2. Effects of elevated CO2 on ECM fungi have also been shown to vary over time by O’Neill et al. (1987). Samples in our study were taken at the end of the growth season at both sampling occasions to avoid some of the variation attributed to seasonal changes in community structure. Despite the considerable interyearly variation in community composition a clear pattern was revealed in the DCA analysis. The ECM community structure of all fertilized trees was clearly separated from that of the unfertilized trees, with the exception of the elevated trees in the unfertilized plot. These became more similar to the fertilized plot after three years of CO2 treatment.

After 3 yr, the effects of CO2 on community composition were of the same magnitude as those seen after 15 yr of fertilization treatment (indicated in Fig. 1 by the similar length of CO2 and fertilization vectors). Trees are seen to respond within hours-days to elevated CO2, and the response of the below-ground symbiont will be mediated through the host plant. The dependence of ECM fungi on current assimilates have been shown both in laboratory and field experiments (Söderström & Read, 1987; Lamhamedi et al., 1994; Högberg et al., 2001). Physiological responses by ECM fungi to changes in C-supply are therefore likely to take place shortly after the plant responds. Shifts in community composition, however, may take a longer time to be induced. We do not know if the induced change in ECM community structure observed in the present study is transient in the present study, but evidence from long-term studies on the effects of N-fertilization upon ECM communities suggests that once initiated, changes in community structure may be more enduring compared to physiological responses.

C- and N-nutrition within ECM fungi are intimately linked. The response by the fungus to elevated CO2 levels is most likely to be affected by availability, uptake and allocation of N. We found no detectable interactions between elevated CO2 and fertilization for individual components of the community, as might have been expected. Interactions between elevated CO2 and N treatments were reported by Runion et al. (1997), with increased colonization, fine root lengths and numbers of ECM root tips under CO2 enrichment only under high N conditions. The authors suggested that sink/source relationships were a major factor regulating ECM fungal responses to elevated CO2.

Klironomos et al. (1997) found increased growth of extra-matrical mycelium of AM fungi colonizing Populus tremuloides under low N-availability/high CO2 treatment, and a decrease under high N-availability. Körner et al. (1997) suggested that microbial activity of a late successional alpine grassland ecosystem was colimited by C supply and N availability. We still know very little about individual species responses.

The issue of whether or not trees will experience a down-regulation or acclimation of photosynthesis as a response to long-term exposure to elevated CO2 has been debated over the years. If a down-regulation occurs in the field it might have large effects on the capacity of forests to sequester CO2 and act as the predicted sink for atmospheric C. Results from short-term pot studies have shown considerable down-regulation of photosynthetic rates within days-weeks (Curtis & Wang, 1998; Ward & Strain, 1999). This down-regulation seems to be a result either of sink-limitations (accumulation of starch and succrose) or environmental limitations (root restrictions due to pot size, nutrient availability). For trees in the field, the situation seems to be quite different, and down-regulation has usually not been seen in longer-term experiments (Saxe et al., 1998; Curtis & Wang, 1998; Norby et al., 1999). But although down-regulation in the field may not generally be considered to be as pronounced as for short-term pot experiments, it cannot be dismissed as a potentially important factor in the over-all C-cycling. In the Flakaliden experiment, down-regulaton of net photosynthesis was observed (G. Wallin, pers. comm.).

Fitter et al. (2000) suggested that the response of mycorrhizal fungi could counteract down-regulation by providing the plants with more nutrients, but only up to a certain point. Eventually the whole system will come to a new equilibrium since some factor will always be limiting in ecosystem processes. In conclusion, our results show that elevated CO2 can have an impact on the community structure of ECM fungi in mature forests, and potentially alter carbon and nutrient allocation and turnover within forest ecosystems. Since soil fertility has been shown to limit C sequestration by forest ecosystems in an enriched CO2 environment (Oren et al., 2001) the role of ECM fungi could be even more important than earlier believed.

Acknowledgements

This work was supported by the Swedish National Energy Administration (STEM) and the Knut and Alice Wallenberg foundation. We wish to thank Jörg Brunet for valuable help with the ordination analysis, Birgitta Vegerfors-Persson for evaluating the statistical analysis and Björn Lindahl for discussing the ordination results.

Ancillary