• Ectomycorrhizal (ECM) fungal communities of Douglas-fir (Pseudotsuga menziesii) and paper birch (Betula papyrifera) were studied along a chronosequence of forest development after stand-replacing disturbance. Previous studies of ECM succession did not use molecular techniques for fungal identification or lacked replication, and none examined different host species.
• Four age classes of mixed forests were sampled: 5-, 26-, 65-, and 100-yr-old, including wildfire-origin stands from all four classes and stands of clearcut origin from the youngest two classes. Morphotyping and DNA sequences were used to identify fungi on ECM root tips.
• ECM fungal diversities were lower in 5-yr-old than in older stands on Douglas-fir, but were similar among age classes on paper birch. Host-specific fungi dominated in 5-yr-old stands, but host generalists were dominant in the oldest two age classes. ECM fungal community compositions were similar in 65- and 100-yr-old stands but differed among all other pairs of age classes.
• Within the age range studied, site-level ECM fungal diversity reached a plateau by the 26-yr-old age class, while community composition stabilized by the 65-yr-old class. Simple categories such as ‘early stage’, ‘multi stage’, and ‘late stage’ were insufficient to describe fungal species’ successional patterns. Rather, ECM fungal succession may be best described in the context of stand development.
Ectomycorrhizas are ubiquitous on the roots of ectomycorrhizal (ECM) host trees in northern forest ecosystems and are essential for tree survival and productivity. Because ECM fungal colonization patterns are dependent on host distribution, and because physiologies of hosts and mycobionts are so strongly linked, ECM fungal succession patterns should integrate with floristic succession of forested stands. Patterns of above-ground vegetation succession following wildfire and clearcutting have been well characterized in forested systems (Oliver & Larson, 1996), but ECM fungal succession has been less well studied (Jones et al., 2003).
The earliest uses of the term ‘succession’ generally referred to primary succession and were applied to model patterns of community development that exhibited discrete replacement of species or species groups over time (Chapin et al., 2002). However, recent functional definitions of this term include ‘community development following disturbance or formation of a new habitat’ (Anderson, 2007), and can refer to either secondary or primary succession. Egler (1954) argued that patterns in which discrete community replacements occur (termed ‘relay floristics’) are not common in many systems, except during primary succession. He proposed the ‘initial floristics’ model of community development to describe secondary succession, when all species present over the entire developmental period may be present immediately following disturbance, but shift in terms of dominance over time. Forest succession following most natural disturbance regimes, including those examined in this study, follows variations of the initial floristics model. This application of the term ‘succession’ has been used previously to describe the process of ECM fungal community development (Mason et al., 1983; Visser, 1995). Succession of ECM fungi was first described on young birch used to reforest agricultural land of the United Kingdom, where fungal taxa categorized as ‘early stage’ were thought to be replaced by taxa that were termed ‘late stage’ (Deacon & Fleming, 1992). The terms ‘early stage’ and ‘late stage’ may oversimplify patterns of fungal succession in native forest ecosystems, and colonization patterns of particular ECM fungal species may be better described with reference to selection theory, life history strategies (Deacon & Fleming, 1992), competition and resource use than by temporal aspects alone (Taylor & Bruns, 1999).
In temperate North American forests, general trends in ECM fungal species composition and community structure have been observed from chronosequences representing forest succession (Kranabetter et al., 2005). In jack pine forests, for example, Visser (1995) found that ‘multi-stage’ fungi occurred in all stand ages, and these fungi were augmented, rather than replaced, by a community of ‘late-stage’ fungi in older stands. Only a few ‘early-stage’ fungi present in young stands were absent from older ones (Visser, 1995). In older mixed conifer stands, sporocarp biomass and frequency of several ‘late-stage’ fungi were greater than in younger stands (Smith et al., 2002; Kranabetter et al., 2005). While Visser (1995) and Kranabetter et al. (2005) found that ECM fungal diversity was much lower in young, open stands than in older stands, Smith et al. (2002) found no strong differences in cumulative species richness of ECM sporocarps among stand ages.
Host specificity of ECM fungi is important to their succession patterns in mixed forests. Generalist fungal taxa comprise substantial proportions of the ECM community in mixed conifer-broadleaf (Molina et al., 1992) and mixed conifer forests (Horton & Bruns, 1998). Climax tree species may associate with host generalist fungi to the greatest extent during their establishment period because these fungi are already present in ECM communities of seral tree species and facilitate maximal colonization of seedlings (Kropp & Trappe, 1982; Horton et al., 2005). Certain ECM fungi appear to colonize new hosts more successfully when those hosts are near established host trees (Hagerman et al., 1999).
Ectomycorrhizal communities have been well studied on conifer seedlings regenerating within 10 yr of clearcutting (Jones et al., 2003). However, few ECM community comparisons have been made between stands of different disturbance histories. Lazaruk et al. (2005) found fewer active roots and slightly lower ECM fungal species richness and diversity in a patchily burned forest than in unburned clearcuts. Mah et al. (2001) found ECM fungal community composition differences on hybrid spruce (Picea engelmannii Parry ex Engelm. ×Picea glauca (Moench) Voss), but no differences in diversity or percentage colonization in burned vs unburned clearcuts.
The objectives of this study were to characterize successional patterns of the below-ground ECM community in a chronosequence of mixed forests, and to improve on earlier succession studies by using molecular identification of ECM fungi and a well-replicated study design. The hypotheses were: ECM fungal species diversity increases with stand age; ECM fungal community composition and diversity differ between forests of clearcut and wildfire origin; previously described ‘early-stage’ fungi decrease in abundance with increasing stand age, while ‘late-stage’ fungi increase, and ‘multi-stage’ fungi have no distinctive pattern; and the proportion of the ECM community comprising ECM fungi shared between Douglas-fir (Pseudotsuga menziesii) and paper birch (Betula papyrifera) decreases with stand age (see Kropp & Trappe, 1982).
Materials and Methods
Site description and study design
The study sites were located in two moist, warm variants (ICHmw2, ICHmw3) and one moist, cool variant (ICHmk2) of the Interior Cedar-Hemlock (ICH) biogeoclimatic zone of southern interior British Columbia (Lloyd et al., 1990). In these variants, forests regenerate following wildfire or clearcutting in a single cohort, with Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco), paper birch (Betula papyrifera Marsh) and lodgepole pine (Pinus contorta var. latifolia Doug. Ex Loud.) forming the seral canopy during the first 100 yr (all variants), and western redcedar (Thuja plicata (Donn ex D. Don) Spach) (all variants), western hemlock (Tsuga heterophylla (Raf.) Sarg.) (ICHmw2 and ICHmw3), or hybrid spruce (ICHmk2) gaining dominance later in succession. The ICHmk2 is a transition zone located on elevational gradients between the ICHmw variants and Montane Spruce zones and has fewer growing-degree days than the ICHmw2 and ICHmw3.
Single cohort stands of four stand age classes, representing a chronosequence after stand-replacing fire and clearcutting, were selected for sampling. The four stand age classes represent important stages of stand development (Oliver & Larson, 1996): 4–6 yr old (stand initiation stage, when Douglas-fir and paper birch establish concurrently); 21–30 yr old (stem exclusion stage, when Douglas-fir and paper birch compete intensely for resources); 60–70 yr old (stand re-initiation stage, when birch starts to senesce, creating canopy gaps); and 90–103 yr old (stand re-initiation stage, when Douglas-fir dominates over remaining paper birch). These age classes are referred to as 5-, 26-, 65-, and 100-yr-old stands, respectively, and stands of wildfire origin are referred to as burned. The only two sites located in the ICHmk2 were two of four replicates of the 26-yr-old burned stands; no other appropriate sites could be found in the ICHmw variants. For the 5- and 26-yr-old age classes, both clearcut and burned sites were sampled, but for the 65- and 100-yr-old age classes, only burned stands were available. These six stand types (represented by age class and disturbance type) were sampled on four replicate sites, each in a completely randomized design.
Sites were selected using the following criteria: Douglas-fir and paper birch comprised at least 75% of the total canopy cover, which ranged between 25% in the youngest age class to 90% in the oldest age class; moisture regime was submesic to mesic and site series was zonal (Lloyd et al., 1990); and replicate stands were at least 800 m apart and within 500–1200 m elevation. Site attributes are listed in Table S1 (Supplementary material). The two oldest stand types had regenerated naturally, while the two youngest age classes had been planted to Douglas-fir and regenerated naturally to paper birch. For the youngest age class, age represents the number of years since planting. One 30 m × 30 m plot was established at each replicate site. Soils were Podzols or Brunisols with loamy texture and moder humus form.
Sampling for ectomycorrhizas
Soils were sampled for ectomycorrhizas twice in 2004 – late May to early June (spring) and late September to early October (autumn) – providing a more complete look at the entire growing season than a single sampling period. Soil sampling locations were chosen randomly within each plot but constrained to locations where roots of both hosts were likely to be found, with a minimum distance of 2 m between samples. Each sample was 9 cm × 9 cm, included forest floor and the top 20 cm of mineral soil, and was removed using a machete and trowel. Eight soil samples were taken from each site at each sampling period.
Soil samples collected from 5-yr-old burned sites in spring yielded insufficient root tips for analysis. Therefore, in the autumn, we also sampled the entire root systems of three Douglas-fir seedlings from each of these sites as well as from the 5-yr-old clearcut sites. Examination of excavated seedling root systems thereafter showed that the roots of Douglas-fir from burned sites did not grow radially outward from the tap root nearly as much as those from 5-yr-old clearcuts, which explains the lack of root tips in soil samples from the burned sites. Both soils and seedlings were sampled from 5-yr-old clearcut sites in the autumn, but only seedlings were sampled from 5-yr-old burned sites. Because of this, 5-yr-old burned stands could only be compared with 5-yr-old clearcut stands, and only based upon the ECM of the excavated Douglas-fir seedlings.
Sorting of root tips and morphotyping
For each sample, all roots were sieved from the soil, placed in distilled water, and cut into 1.5 cm segments. Obvious ECM tubercles were sliced in half to help equalize their chance of being sampled, as the tips within are far more clumped than other growth forms. Segments were randomly selected for viewing under a dissecting microscope. Viability of root tips was determined based on color and turgidity. Ectomycorrhizal status was determined by the presence of a fungal mantle and/or Hartig net. Douglas-fir and paper birch roots were differentiated by the size of root tips. When this was ambiguous, ECM paper birch root tips were identified by their radially elongated epidermal cells when viewed in cross-section.
Successive root segments were examined until 100 live Douglas-fir and 100 live paper birch root tips were counted per sample, and percentage ECM colonization was determined from the ratio of live mycorrhizal to total live tips. Ectomycorrhizal roots were then cut into 5–10 mm segments and these segments were randomly subsampled from a 5 mm grid until 50 tips were selected per host for morphotyping (50 tips × 2 hosts = 100 root tips per soil sample); individual root tips were counted from tubercles. All ECM subsamples were examined in detail under dissecting and compound microscopes according to Goodman et al. (1996). For each subsample, one to five root tips were freeze-dried and stored at –80°C until DNA extraction. Roots of seedlings from 5-yr-old sites were washed, cut into 1.5 cm pieces, and randomly sampled until 400 viable root tips were classified as mycorrhizal or nonmycorrhizal. Morphotypes were determined for 200 randomly selected ECM tips per seedling.
Roots of other broadleaf trees may have erroneously been counted as birch roots and those of other conifer species counted as Douglas-fir, so error in host tree identification was estimated by amplifying plant choloroplast DNA from ECM root tips and using RFLP analysis as described in Brunner et al. (2001). To estimate the error, we subsampled (from ECM tip genomic DNA also used to identify fungi) 10 Douglas-fir and 15 paper birch root tips from soil samples taken within 5 m of other ECM host species; 15 more subsamples of Douglas-fir and paper birch root tips were taken from randomly selected samples.
Molecular identification of ECM root tips
DNA extraction and PCR amplification One subsample per morphotype per soil sample was subject to DNA extraction and PCR amplification of the fungal ITS region of nuclear rDNA (exceptions noted later). Extraction methods followed Baldwin & Egger (1996), except that samples were pulverized with a ceramic bead in a beater machine instead of pulverizing manually. Primer pairs used in PCR amplifications were: NSI1 and NLC2; ITS1f and ITS4; and ITS1 and NL6Bmun or ITS4B, depending on their preceding success with the specific morphotype of the sample.
RFLP and DNA sequence analysis Accuracy of within-soil sample morphotype sorting was checked by PCR-RFLP analysis on two to three samples each of 15 morphotypes. Seven of these 15 morphotype selections included subsamples from both host species in the same soil samples. Restriction digests were performed and RFLPs visualized on gels as described by Hagerman et al. (1999). Several Rhizopogon tubercles were also checked by RFLP analysis as described in Kretzer et al. (2003).
Samples were sequenced using the Big Dye Terminator Kit (Applied Biosystems, Foster City, CA, USA). Sequencing was performed on a 3730S capillary sequencer or a 3130 × 1 capillary sequencer (Applied Biosystems). Forward and reverse sequences were aligned and manually corrected in Sequencher 4.2 (GeneCodes, Ann Arbor, MI, USA). Sequences were blast searched (Altschul et al., 1997) against GenBank and unite databases to suggest taxonomic affinities of the samples. A database taxon was considered a proper species match to a root tip sample if their sequences had 98% or greater similarity and aligned over at least 450 base pairs. Samples that sequenced poorly in one direction were analyzed with a single-pass sequence, and were considered proper matches at 97% similarity or better as a result of error rates of single-pass sequencing (Izzo et al., 2005). If no proper species match was made to a sample, then taxonomic placement was made according to the consensus of the 10 top-scoring matches. Next, all sample sequences were aligned by genus or family along with sporocarp sequences from Durall et al. (2006). Alignments were performed in ClustalX (Thompson et al., 1997), and pairwise sequence similarities were calculated in a current version (3.63) of phylip (Felsenstein, 1989). The same criteria for blast species matching were used to match unknown samples to each other, similar to the approach taken by Parrent et al. (2006) to delineate ITS ‘phylotypes’ of ECM fungi; here, ‘unique sequence types’ and ‘fungal species’ are used interchangeably to describe functional species-level units defined by sequence match criteria. One representative of each unique sequence type was deposited in GenBank (accession numbers EF218740-218845). Some morphological data were used along with DNA information for the final classification of samples. No attempt was made to analyze molecular data for Cenococcum geophilum, because the taxonomy of this species complex is not well delineated by variations in the ITS region alone (Douhan & Rizzo, 2005).
While 16 total soil samples were taken from each site, several samples from each site contained few or no live root tips, and many contained root tips of only one host. Therefore, ECM from 10 soil samples (five spring, five autumn) were used from each site for analyses. Samples were chosen to standardize the analysis to six samples with 50 tips each of both hosts, two samples containing only 100 Douglas-fir root tips each, and two samples containing only 100 paper birch tips each (1000 total tips per site). One of the 100-yr-old sites only yielded nine samples that contained live root tips.
Statistical analyses were carried out using SAS version 9.1 (SAS Institute Inc., Cary, NC, USA), with the exception of nonmetric multidimensional scaling (NMS) ordinations, diversity calculations, and species-sample unit curves, which were performed in pc-ord, version 4 (McCune & Mefford, 1995–2002). Species-sample unit curves by stand type were constructed by pooling the four replicates for each stand type. Mean site-level Shannon diversity index of ECM fungal communities, as well as relative abundance of shared ECM fungi, were compared among stand types and between host species (Douglas-fir and paper birch) using split-plot analysis of variance (anova). For the combined fungal community of both hosts, diversity measurements were compared among stand types with one-way anova. Relative abundance was calculated for each site as the number of root tips colonized by a taxon divided by the total number of ECM root tips identified at that site. Frequency was calculated as the number of soil samples in which a taxon occurred at a given site (site-level analyses were standardized to 10 soil samples).
Generalized linear models (GLMs) of abundance (number of root tips colonized in each soil sample) of dominant fungal taxa vs all possible combinations of stand age, initiation type, and ICH variant (accounting for the different climax tree composition in the ICHmk2) were created using the negative binomial distribution and a log link. These models were used partly because relative abundance data from most taxa, with the exception of Cenococcum geophilum, Russula, and Rhizopogon vinicolor-type, violated anova assumptions, even after arcsine-square-root transformation. Relative abundance of each taxon was then estimated by dividing the predicted values by the average number of ECM root tips of the appropriate host per soil sample. Quadratic and inverse transformations of the independent variable, stand age, were included to identify likely response curve shapes. Akaike's information criterion (AIC) was used to select the best model from the set of candidates (Burnham & Anderson, 2002).
ECM from soil samples
ECM colonization and fungal identification Mean ECM colonization was at least 97% for all stand types. On average, 476 Douglas-fir and 462 paper birch ECM root tips per site were identified with sufficient morphological and/or DNA support. Out of the 541 total DNA sequences analyzed, 83% were of sufficient quality and length to place into genotypes, but 13% of these were identified by blast search as nontarget, co-occurring fungi (mostly either saprotrophic or ‘MRA’ fungi); the vast majority of samples producing nontarget ‘MRA’ sequences were of distinct morphotypes (e.g. Rhizopogon vinicolor-type, Lactarius torminosus, and Hebeloma leucosarx) that could not be confused with that of ‘MRA’ fungi. It is possible that some nontarget sequences came from tips that were not viable at the time of sampling or were colonized by two ECM fungi, even though only ECM tips that appeared turgid, healthy and exhibited only one distinct morphotype were sampled.
In total, 105 unique ECM fungal ITS sequence types (hereafter referred to as ‘species’) were used for analysis. The average usable sequence length was 706 base pairs. All samples with ambiguous taxonomic placement were omitted from analyses. Within-soil sample morphotyping was shown to be accurate by all pertinent RFLP and DNA sequence information. All ECM fungal species, as well as their blast search results and mean relative abundance by stand type, are listed in Table S2.
Rhizopogon vinicolor and R. vesiculosus (sensu Kretzer et al., 2003), tuberculate species difficult to distinguish by morphology, were encountered frequently. We could not analyze DNA from all observations of these sister species. Since both species were identified by DNA analysis in numerous samples from all stand types, they were lumped as Rhizopogon vinicolor-type for analyses. There were two distinct Piloderma sequence types, both of which shared the Piloderma fallax-like morphotype, but limited resources did not permit sequence analysis of many samples. Hence, these two types were lumped as Piloderma in the analyses.
Molecular methods showed that host identification of Douglas-fir root tips was correct for all of the 25 samples selected. However, for paper birch, two out of 15 samples taken ≤ 5 m from other ECM broadleaf host trees produced RFLPs matching the genus Salix (Brunner et al., 2001). One out of 15 samples > 5 m from other ECM broadleaf host species also matched Salix. However, the 15 samples < 5 m from another broadleaf species accounted for 83% of the samples that were taken from this distance. Since most samples were taken > 5 m from other ECM broadleaf species, the overall error in birch identification was likely quite low.
ECM community diversity and shared fungi Based on relative abundance, ECM fungal diversity on paper birch did not differ among stand types, but on Douglas-fir it was lower in the 5-yr-old clearcuts than all other stand types (Fig. 1a). In 5-yr-old clearcuts, the Shannon diversity index was three times higher for paper birch than for Douglas-fir (F = 44.65; P < 0.0001). There was a significant interaction between stand type and host species (F = 6.86; P = 0.0024) because stand type affected diversity in only the Douglas-fir community. Where frequency data were used instead of abundance for diversity estimates, the same patterns emerged among stand types (Twieg, 2006). Species-sample unit curves show that sampling was inadequate to account for all species present in each of the stand types, but the curve levels off more for the 5-yr-old stands than the older stands (Fig. 2).
Of the 105 ECM fungal species observed in soil samples, 42 occurred on both hosts (i.e. were shared ECM fungi), 23 were found only on Douglas-fir, and 40 were found only on paper birch. Mean relative abundance of shared ECM fungi was significantly higher (by 75%) on paper birch than Douglas-fir (F = 23.34; P = 0.0002 with stand types pooled) and was affected by stand type (F = 9.49; P = 0.0005) (Fig. 1b). Mean relative abundance of shared ECM fungi on Douglas-fir was over five times greater in 65- and 100-yr-old stands than in 5-yr-old stands. There was no interaction between stand type and host species (F = 0.63; P = 0.6454); therefore, pairwise mean comparisons were not made between hosts within stand types or among stand types for each host individually.
ECM community composition and structure Cenococcum geophilum was the most frequently encountered ECM fungus, followed by Rhizopogon vinicolor-type (Douglas-fir specific), Piloderma, and Leccinum scabrum (birch specific). Rhizopogon vinicolor-type was the most abundant ECM taxon on Douglas-fir in every stand type. Relative abundance of fungal taxa colonizing at least 5% of ECM root tips in at least one age class are shown in Fig. 3. Genera are included where individual species colonized less than 5% of ECM root tips by age class; Thelephoraceae was included as a family because one sequence type grouped equally well with two genera (i.e. Thelephora and Tomentella) within that family. Thelephora terrestris was found on Douglas-fir seedlings, but it was found only on paper birch in soil samples, and only on 5-yr-old clearcut sites.
Cenococcum geophilum mean relative abundance was unaffected by stand type (F = 2.02; P = 0.144), but was about four times higher on paper birch than on Douglas-fir (17 vs 4.4%; F = 37.2; P < 0.0001). Mean relative abundance of Russula differed among stand types (F = 7.25; P = 0.0019), but not among host species (F < 0.01; P = 0.964). Most Russula species occurred infrequently and in low abundance, precluding meaningful analyses at the species level. One exception, Russula 9, increased with stand age (χ2 = 5.37; P = 0.021).
Generalized linear models of taxon abundance are summarized in Table S3. The genera Piloderma and Russula, in the combined ECM community of both hosts, increased with stand age (Fig. 4a; P < 0.001 for each), and both were less abundant in the ICHmk2 than ICHmw variants (P = 0.0001 and 0.0904, respectively). However, Russula abundance was predicted only marginally better by variant and stand age together than by stand age alone. On Douglas-fir, Rhizopogon vinicolor-type was more abundant in 5-yr-old clearcuts (P = 0.0326), and Suillus lakei was an important part of the ECM community in 26-yr-old stands but not in any other age class (Fig. 4b; P = 0.0013). Birch-specific Lactarius pubescens and Leccinum scabrum decreased with stand age (P < 0.001 for each), but the decline was steeper for L. pubescens (Fig. 4c).
A NMS ordination of the combined ECM fungal community showed that stand age was strongly correlated with the same axis, explaining the greatest proportion of variation in relative Sorensen distance (based on frequency of ECM fungal species) among the sites (Fig. 5). Clearcut 5-yr-old stands grouped closely with each other, and clearcut and burned 26-yr-old stands grouped reasonably well. Sixty-five- and 100-yr-old stands grouped well with each other and were separated from the other age classes.
ECM from seedling samples
Mean ECM colonizations of Douglas-fir seedlings were 96 and 98% on burned and clearcut 5-yr-old sites, respectively. There was a tendency for greater ECM fungal richness, evenness, and Shannon diversity of the ECM communities on burned than on clearcut sites, but differences between stand types were not significant (t = –1.9, –2.1, and –2.1; P = 0.11, 0.08, and 0.08, respectively). Rhizopogon vinicolor-type was dominant in clearcuts, while R. rudus was of roughly equal abundance as R. vinicolor-type in burned stands (Fig. S1). NMS ordinations based either on abundance or frequency data showed no significant structure in the 5-yr-old Douglas-fir seedling ECM fungal communities (Monte Carlo test; P = 0.21 and 0.25, respectively), nor did they group sites according to initiation type (ordinations not shown).
Mean relative abundance of ECM fungal taxa on excavated seedlings from burned and clearcut 5-yr-old sites are listed in Table S2. Four taxa occurred on seedlings from 5-yr-old clearcuts that were found in soil samples only from older sites: Lactarius rubrilacteus, Phallales 1, Piloderma fallax and Russula 9. On the single seedling where it occurred, Russula 9 colonized 66% of the root tips.
Our first hypothesis, that ECM fungal diversity increases with stand age, was supported strongly by our results on the Douglas-fir and combined communities, but not on the paper birch community. The greatest increase in average site-level diversity occurred in the 5- to 26-yr-old age class, a period corresponding with tree canopy closure, and increased only slightly thereafter, agreeing with the results of Visser (1995) and Kranabetter et al. (2005). At canopy closure, tree growth rates are rapid and leaf area maximal (Simard et al., 2004), with correspondingly high potential for carbon allocation to roots and mycobionts. Host roots are also more abundant and regularly distributed after canopy closure, corresponding with higher diversity of fruiting ECM fungi (Kranabetter et al., 2005) which likely results in positive feedback to below-ground diversity. ECM fungal species diversity tended to be slightly higher in 65- than in 26-yr-old stands, and 65- and 100-yr-old stands had similar diversities.
The ECM fungal community of paper birch was more diverse than that of Douglas-fir in young stands, and changed less dramatically with stand age. Although ECM fungal species richness on paper birch was slightly lower in 5-yr-old stands than in older ones, patterns in species richness were very similar to the stand type patterns in diversity index (Twieg, 2006). Paper birch roots remain intact and healthy following cutting or burning of shoots, providing a large carbon source and ECM legacy for stump sprouts, as well as large ECM inoculum potential for seedlings establishing nearby (Simard et al., 2004). Paper birch trees of stump-sprout origin were abundant in 5-yr-old stands and were considerably more vigorous than those of seed origin. By contrast, Douglas-fir does not sprout from old stumps, and seedlings are often not replanted until a few years after logging, requiring inoculation of seedlings from other plants, hyphae or spores. Jones et al. (1997) found that when both tree species were planted, richness and evenness of the ECM fungal community did not differ between Douglas-fir and paper birch after 16 or 28 months. In that study, all stumps had been removed with excavators before planting, suggesting that sprouting birch stumps are important for maintaining higher ECM fungal diversity on birch seedlings. Durall et al. (2006) similarly found no difference in epigeous ECM sporocarp diversity among recently planted birch, Douglas-fir, and mixed stands on the same study sites as Jones et al. (1997). Many studies, by contrast, show a strong discrepancy between above- and below-ground ECM community composition and structure, and this can occur for several reasons (reviewed by Horton & Bruns, 2001).
There is a possibility that the total numbers of root tips from which ECM were sampled were lower in 5-yr-old stands than in older stands and that this would affect diversity and community composition measures. However, species-sample unit curves showed that the total number of ECM fungal species in all 40 soil samples from the four 5-yr-old sites combined was roughly equal to the mean number of species in 10 soil samples from 26-yr-old clearcut, 65- and 100-yr-old sites, and that older sites were undersampled to a greater degree than young ones. While total numbers of root tips were not counted in soil samples, many soil samples from older sites contained no live root tips, and the only site that did not have at least 10 samples yielding 100 root tips each was a 100-yr-old site. Furthermore, Rhizopogon was similarly dominant on excavated Douglas-fir seedlings as on soil samples from 5-yr-old sites, even though a few ECM fungi frequent in older stands were absent from the soil samples but occurred on seedlings.
Our second hypothesis, that ECM fungal diversity differs between clearcut and burned forests, was largely rejected in this study. While differences may exist in stands younger than those sampled in the youngest class, ECM community diversity was similar among 5-yr-old stands regardless of whether they originated from fire or clearcutting. These results suggest that fungal inoculum was not limiting on these sites, even though the fires appeared intense (based on the lack of forest floor layers and totality of stand destruction) and would have likely reduced inoculum of many ECM fungal species (Baar et al., 1999). Although ECM fungal diversity tended to be slightly lower in burned than in clearcut 26-yr-old stands, this probably resulted from half the burned sites occurring in an ICH moist cool variant, which has a shorter growing season and different climax forest composition than moist warm ICH variants, despite their close geographical proximity.
ECM community composition and structure
Ectomycorrhizal community composition varied substantially with stand age, supporting our third hypothesis regarding patterns of fungal succession. Our results show that some fungal succession patterns are clearer at the genus than species taxonomic level. For example, Russula, one of the three most speciose genera in this study, increased in abundance and frequency (Twieg, 2006) with stand age. This is consistent with other recent, similar studies (Visser, 1995; Smith et al., 2002; Kranabetter et al., 2005). Russula 9, R. brevipes, R. aeruginea, and R. roseipes all occurred infrequently, but had high root tip abundance where found. Patchy distributions are generally expected for ECM fungi (Dickie et al., 2002; Lilleskov et al., 2004). Russula species were absent from young stands on Douglas-fir roots, except for extensive colonization of a single Douglas-fir seedling by one Russula species.
Piloderma also increased in frequency and abundance with stand age, agreeing with patterns observed by Visser (1995) and Smith et al. (2000). Species of Cortinarius tended to increase in frequency and abundance with stand age after 5 yr, although Cortinarius was not found as frequently here as in ICH sporocarp studies (Kranabetter et al., 2005; Durall et al., 2006). High fruiting body frequency and abundance of Cortinarius species has previously been found to coincide with low frequency and abundance on ECM tips (Dahlberg et al., 1997). Lower frequency and abundance of Russula, Piloderma, and Cortinarius in 5-yr-old stands could be the result of many factors, including reduced spore dispersal and germination or lower survival of spores and mycelia. While unique occurrences of Russula and Piloderma on excavated seedlings suggests their abundance in 5-yr-old stands may have been underestimated by soil samples, that their frequencies were also much lower in young stands than in older ones (Twieg, 2006) suggests that more exhaustive sampling would similarly find low relative abundance in young stands.
Rhizopogon vinicolor-type was far more dominant on Douglas-fir roots in 5-yr-old than in older stands, from both soil and seedling samples. Other studies also show that Rhizopogon species are common colonizers in bioassays of soil from disturbed and undisturbed forests (Jones et al., 1997; Taylor & Bruns, 1999). Rhizopogon spores are known to persist as viable inocula for long time periods, even in the absence of hosts (Horton et al., 1998), but rhizomorphs of R. vinicolor and R. vesiculosus are also likely important in colonization of new hosts (Simard et al., 1997a).
Douglas-fir seedlings in this study were dominated by R. vinicolor-type more than in other nearby studies. For example, the relative abundance of R. vinicolor-like ECM on 28-month-old field-grown Douglas-fir seedlings was only 37% (Jones et al., 1997), roughly half the average for the 5-yr-old stands in this study. In Jones et al. (1997), fungi forming E-strain mycorrhizas (represented by Wilcoxina rehmii in the current study) and Thelephora occupied considerable portions of the ECM fungal community on Douglas-fir, but this was not the case in the current study. Those fungi are often considered ‘early stage’ (Visser, 1995), commonly colonizing nursery stock before planting, along with other fungi such as Laccaria spp. and Amphinema byssoides (Hunt, 1991), but they may not compete well with other fungi after a few years in the field. Thelephora terrestris and Laccaria spp. were only present on Douglas-fir in the 5-yr-old sites, and could have been introduced from the nursery, but their relative abundances were low.
Rhizopogon rudus was more dominant in burned 5-yr-old stands, where its abundance was roughly equal to that of R. vinicolor-type, than in clearcut 5-yr-old stands. Soil moisture availability can play a role in the amount of colonization by Rhizopogon species (Baar et al., 1999), and soils are often dry after wildfire because moderate to severe burns can cause persistent hydrophobicity of the uppermost soil layers (Huffman et al., 2001). Timing of colonization (Kennedy & Bruns, 2005) or removal of forest floor layers by the fire (Twieg, 2006) may also contribute to the difference. Cline et al. (2005) found that R. rudus colonized uninoculated Douglas-fir seedlings in the nursery before planting. While Hunt (1991) states that Suillus-like morphotypes found as nursery colonizers could actually be Rhizopogon species, he reminds us that Rhizopogon is not likely to colonize in nurseries because their spores are primarily dispersed by mammals.
Two birch-specific fungi, Lactarius pubescens and Leccinum scabrum, were abundant in 5-yr-old stands. These results contrast with Mason et al. (1983), who refer to strand-forming Lactarius pubescens and Leccinum as ‘late-stage’ fungi. The ‘late-stage’ description arises from bioassay studies (Deacon & Donaldson, 1983; Fox, 1983), which demonstrated that these two fungal taxa do not readily inoculate birch from spores or mycelium dislocated from live hosts. This suggests that vegetative spread was important in young stands after limited initial colonization by spores or fragmented mycelia, or that these fungi were legacies of predisturbance birch roots from which sprouts arose. The probability that these two fungi are relics of the previous mature stands is higher for Leccinum scabrum than for Lactarius pubescens; Durall et al. (2006) found Leccinum scabrum sporocarps frequently in nearby, mature Douglas-fir–paper birch forests, but Lactarius glyciosmus and L. torminosus were the only common Lactarius species in those forests. It is not surprising that the dominant fungi on both hosts in 5-yr-old stands were rhizomorph-forming; this feature would have facilitated inoculation of host roots that were sparse and separated by large distances.
Existing fungal succession models are inadequate to describe the complexity we observed in our study. While the term ‘multi-stage’ adequately described Cenococcum geophilum, this was not the case for Rhizopogon vinicolor-type, which, although important in all stages, was more dominant in young stands. One could call Russula a ‘late-stage’ fungus simply because it was rare in the youngest sites and a significant component of all older age classes. However, the ‘late-stage’ classification does not differentiate this taxon from one that is abundant immediately following canopy closure but nearly absent in older stands (e.g. Suillus lakei on Douglas-fir). It may also overshadow the fact that some Russula species occur in young stands, even though they are generally more frequent and abundant in older stands. Care is also advised in generalization of successional patterns to the genus level, as this study and Kranabetter et al. (2005) found unique patterns among different Lactarius species. Likewise, the pattern exhibited by Hebeloma species in this study did not agree with that of Hebeloma sacchariolens described by Mason et al. (1983). Since Lactarius pubescens dominated 5-yr-old stands, it could be referred to as an ‘early-stage’ species, but its life history strategy does not agree with that of true ‘early-stage’ fungi that colonize new hosts effectively from spores and fragmented vegetative material (see Mason et al., 1983).
Using forest stand development stages to help characterize ECM community succession patterns may be more useful than classifying fungi into their own successional categories. Sites in the stand initiation stage (5-yr-old) had distinctive ECM community composition on both hosts, and also supported low ECM diversity. Stem exclusion sites (26-yr-old) had higher ECM diversity, but community composition was intermediate between 5-yr-old and older sites. For instance, Rhizopogon vinicolor-type was much less dominant and Lactarius pubescens was almost absent in this age class, while Russula and Piloderma were significant components of the community. However, other fungi, such as Suillus lakei, Hebeloma spp., and Cortinarius spp., were prominent in 26-yr-old stands, but were largely supplanted by the stronger presence of Russula and Piloderma in older stands. Sites in the stand re-initiation stage (65- and 100-yr-old) were similar to each other, more so than to other age classes, in ECM diversity and community composition and structure. These patterns encompass principles of both initial and relay floristics models of succession (Egler, 1954), and the complex patterns of ingress and shifting dominance by different fungi reflect the mixed fire regime and gap phase disturbances that create substrate complexity in these ecosystems (Simard et al., 2004).
Network potential between host species
The results of this study do not support our fourth hypothesis that the proportion of Douglas-fir and paper birch ECM root tips colonized by shared fungi decreases with stand age. By contrast, there was a lower proportion of shared ECM fungi in 5-yr-old than in older stands because of the high relative abundance of a few host-specific fungi; Douglas-fir, in particular, substantially increased its proportion of shared fungi with stand development. Nevertheless, shared fungi occupied a significant proportion of birch roots in young stands. Douglas-fir seedlings may have been dominated by shared species immediately following disturbance as suggested by Jones et al. (1997) and Simard et al. (1997b), but our study shows that this does not persist to 4–6 yr. The potential for common mycorrhizal networks (CMNs) to form is still high in young stands, but this study suggests they are considerably more extensive in older stands. Our findings did not support the hypothesis that later seral tree species (e.g. Douglas-fir) associate more with fungi shared by early seral tree species (e.g. paper birch) during establishment, as would parallel with the hypothesized pattern between climax and seral tree species (Kropp & Trappe, 1982). That said, while Douglas-fir in these ecosystems occurs at early and later stages of succession, it is not considered a climax species like western redcedar and western hemlock (Lloyd et al., 1990; Simard et al., 2004). We speculate here, however, that the colonization insurance provided by host generalists early in succession may be overshadowed by other advantages brought to individual tree species by the more abundant host-specific fungi dominating soon after disturbance. These ideas require further scrutiny and validation in different ecosystems and across different successional pathways.
We are grateful to Drs Melanie Jones and Gary Bradfield for extensive editorial suggestions and input on study design, and also to Dr Antal Kozak for statistical advice. We thank the BC Ministry of Forests, Riverside Lumber, Canoe Lumber, and LP Engineering for help in establishing and protecting research sites. Lenka Kudrna, Tanis Gieselman, Danielle Larsen, and Lydia Stepanović were extremely helpful in the laboratory and the field. We also thank Jean Roach for selection of sites. This work was supported by an NSERC Discovery Grant to DMD and a Forest Innovation Investment – Forest Science Program grant to SWS.