• Plant host species is considered an important factor influencing ectomycorrhizal (EM) communities. To gain insights into the role of host species in structuring EM communities, EM communities on sympatric oak (Quercus) species were compared in the Sierra Nevada foothills of California.
• Using molecular methods (polymerase chain reaction, cloning, restriction fragment length polymorphism and DNA sequencing), EM fungi on roots of deciduous Quercus douglasii and evergreen Quercus wislizeni trees were identified from 64 soil cores.
• The total EM species richness was 140, of which 40 taxa were detected on both oak hosts. Greater diversity and frequency of EM fungi with epigeous fruiting habit were found on Q. wislizeni, while taxa in the Ascomycota were more frequent and diverse on Q. douglasii.
• Using ordination, it was determined that both soil extractable phosphorus and oak host species explained a significant proportion of the variation in EM species distribution. These results indicate that plant host species can be an important factor influencing EM fungal community composition, even within congeneric trees.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
In arid and semiarid ecosystems, the concentration of resources beneath individual trees or shrubs can result in the formation of ‘resource islands’ or ‘islands of fertility’ (Padien & Lajtha, 1992; Belsky et al., 1993). Increased concentrations of nutrients, organic matter and soil microbial biomass under trees or shrubs in arid and semiarid environments indicate that individual long-lived plants can have a significant effect on soil resources under their canopies (Belsky et al., 1993; Schlesinger et al., 1996; Burke et al., 1998). Blue oak woodlands are a mosaic of scattered individual trees or clusters of trees surrounded by grassland (Allen-Diaz et al., 1999). The dominant overstory tree, winter-deciduous blue oak (Quercus douglasii), often occurs in association with evergreen interior live oak (Quercus wislizeni) (Nixon, 2002). Within this savanna–woodland mosaic, isolated trees can influence the environment under their canopies, creating islands of enhanced soil fertility, and potential EM fungal habitat, as compared with adjacent grasslands (Holland, 1973; Callaway et al., 1991; Frost & Edinger, 1991; Dahlgren et al., 1997).
We hypothesized that there might be differences in the belowground EM communities on co-occurring Q. douglasii and Q. wislizeni for two reasons. First, we observed differences in fruiting patterns of EM fungi under these two oaks hosts. We frequently found a diverse array of epigeous EM fruiting bodies under Q. wislizeni, but epigeous species were less frequent and diverse under Q. douglasii (M. E. Smith & M. H. Morris, pers obs). Secondly, ecological and functional differences between deciduous Q. douglasii and evergreen Q. wislizeni could influence EM communities. Rates of photosynthesis, foliar nutrient content, litter decomposition and carbon production often vary between evergreen and deciduous species (Monk, 1966; Chabot & Hicks, 1982). Litter layer depth and herbaceous vegetation also differ markedly between the understories of Q. wislizeni and Q. douglasii (Frost & Edinger, 1991; Downie & Taskey, 1997; Rejmánek et al., 2005).
This study investigated belowground EM communities on root tips of two co-occurring Quercus species in a blue oak woodland. DNA sequences from root tips were compared to a fruiting body sequence database of taxa collected from the site over a 4-yr period (Smith et al., 2007). We predicted that there would be a greater frequency of epigeous EM species on the roots of Q. wislizeni than on Q. douglasii based on observations of fruiting patterns and the differences in ecology between the two Quercus hosts. The objectives of our study were (1) to document diversity and structure of EM communities on two Quercus hosts in a blue oak woodland; (2) to compare EM communities on Q. douglasii and Q. wislizeni; and (3) to investigate the influence of abiotic soil characteristics on the distribution of EM fungi.
Materials and Methods
The research was conducted in a blue oak woodland located in the Koch Natural Area at the Sierra Foothill Research and Extension Center (SFREC) in Yuba County, California, USA (39°14′N, 121°18′W). Within the protected natural area, we selected two adjacent sites approx. 250 m apart. Before 1960 the area experienced cattle grazing and frequent fires, but little disturbance has occurred since that time (J. M. Connor, unpublished). The dominant overstory tree, Quercus douglasii Hook. & Arn, and the subdominant Quercus wislizeni A. DC. co-occur in a mosaic ranging from open savanna to dense woodland. In the study area, tree architecture differs between the two oaks; Q. wislizeni is shrubby in stature with multiple stems while Q. douglasii is arboreal. Individual Pinus sabiniana Douglas trees are scattered throughout the landscape. Common shrubs include the non-EM Rhamnus tomentella Benth. ssp. tomentella and Toxicodendron diversilobum Torrey & Gray. Assorted exotic annual grasses and forbs form arbuscular mycorrhizas and dominate the understory (He et al., 2006).
The Mediterranean climate is characterized by cool, wet winters (January–March) and hot, dry summers (June–September). Mean annual precipitation is 75 cm (range 23 to 132 cm) and mean annual temperature is 17.8°C (range 10–43°C). At the study area, elevation ranges from 400 to 600 m. Soils are derived from basic metavolcanic rocks and are fine-loamy, mixed, thermic, Mollic Haploxeralfs (Alfisols) with pH ranging from 5.2 to 7.1 (Huang, 1997).
At site 1, we selected four pairs of Q. douglasii and Q. wislizeni trees scattered over a 25-ha area. Trees were estimated to be ≥ 50 yr old and were selected based on distance from each other and distance from other EM host species. Several studies suggest that oak roots in blue oak woodlands are concentrated beneath the tree canopy (Jackson et al., 1990; Callaway et al., 1991; Huang, 1997). In order to reduce the possibility that roots from other EM species were present within the sampling areas, the distance separating individual Q. douglasii and Q. wislizeni trees was 20–50 m. All Q. wislizeni trees had multiple stems (mean = 5.3 stems per tree) while all Q. douglasii trees had a single main stem that branched above c. 1.5 m. The average diameter at breast height (DBH; diameter at 1.5 m height) for Q. douglasii and Q. wislizeni was 47 cm and 43 cm, respectively. For trees with multiple basal stems, DBH was calculated as 2 × sqrt (sum of the basal area of each stem/3.14). For each tree, four soil cores were collected on 20 April 2004 in cardinal directions at approximately half canopy.
At site 2, sampling was the same as at site 1 except that Q. douglasii soil cores were collected from four 16 × 16 m plots within a stand of 41 Q. douglasii trees (mean DBH = 30 cm; mean stems per tree = 1.3). Sampling was modified at site 2 because the distribution of Q. douglasii trees necessitated that cores be collected from plots around a cluster of trees and not from around individual isolated trees. A portion of the data presented here is a subset of the data collected by Smith et al. (2007): 16 cores from a single sampling date. Sampling of Q. wislizeni trees was the same as at site 1; a total of 16 soil cores were collected from four Q. wislizeni trees (mean DBH = 48 cm; mean stems per tree = 4.5) in cardinal directions at approximately half canopy. The four Q. wislizeni trees were between 25 and 140 m from the edge of the Q. douglasii plot and trees were estimated to be ≥ 50 yr old. Soil cores from site 2 were collected on 3 May 2004.
Sampling of ectomycorrhizal fine roots
Rocky soils made soil coring difficult, so we collected a standard volume of soil (900 cm3) to an approximate depth of 10–12 cm. We sampled to this depth because the majority of oak fine-root (< 0.5 mm) biomass is located at depths of 0–20 cm and decreases below 20 cm (Millikin & Bledsoe, 1999). Soil samples were taken to the laboratory, stored at 4°C and processed for EM root tips within 2 wk. Fine roots were removed from soil samples and hand-washed over a 0.12-mm sieve. From each sample, healthy EM root tips were severed from roots and placed in a Petri dish with water. We randomly selected 100 EM root tips from the collection of tips in the Petri dish by picking tips and systematically swirling the Petri dish to redistribute roots tips between selections. We considered root tips to be colonized by EM fungi based on absence of root hairs and presence of a fungal mantle when examined under a dissecting microscope. EM roots were cleaned of debris with deionized water and freeze-dried for DNA analysis.
Samples were collected and litter depth was measured in late April and early May, when soils were moist and relatively easy to penetrate. After EM root tips had been removed, soils were analyzed for pH (saturated paste in H2O and 0.5 M CaCl2), soil water content (gravimetric), extractable phosphorus (P) (Bray-P) and total carbon (C) and nitrogen (N) (Carlo Erba C/N analyzer; combustion method). Measurements of soil N, P and C were performed by the Division of Agricultural and Natural Resources Laboratory, University of California, Davis.
DNA extraction and PCR
From each sample of 100 pooled EM root tips, DNA was extracted by a modified cetyltrimethylammonium bromide (CTAB) method (Gardes & Bruns, 1993) followed by purification with the UltraClean soil DNA kit (MO BIO Laboratories, Carlsbad, CA, USA). The internal transcribed spacer (ITS) region and part of the 28s large subunit were amplified with primers ITS1F and LR3 (Gardes & Bruns, 1993; Hopple & Vilgalys, 1994). Amplifications were performed in a GeneAmp PCR System 9700 thermocycler (Applied Biosystems, Foster City, CA, USA) with initial denaturation at 94°C for 5 min, followed by 20 cycles of 94°C for 1 min, 55°C for 1 min, and 72°C for 4 min and final extension at 72°C for 7 min. Amplifications consisted of 15-µl reactions containing 1 Unit Platinum Taq High Fidelity (Invitrogen, Carlsbad, CA, USA), 1× polymerase chain reaction (PCR) buffer, 2 mM MgSO4, 0.2 µM dNTP, 0.38 µM primers and 2–3 µl of DNA template.
Cloning and restriction digest screening
PCR products from four replicate reactions were pooled and cloned using the TOPO TA cloning kit (Invitrogen, Carlsbad, CA, USA) following the manufacturer's protocols. For each sample, 48 transformed Escherichia coli colonies were randomly selected and used as templates for PCR amplification with ITS1F and LR3. Analyzing 48 clones from each sample allowed us to efficiently detect the common species in each sample. In a previous study with bulk soil, Landeweert et al. (2003) determined that analyzing 30 clones per soil sample was likely to be sufficient for detecting the most common EM species. Plasmids were amplified as described above except for minor modifications; the initial denaturation time was increased to 10 min, the extension time was decreased to 2 min, the number of cycles was increased to 25, the reaction volume was increased to 30 µl and recombinant Taq polymerase was used (Invitrogen). PCR products were digested with restriction enzymes HinfI and AluI. Fragments were run simultaneously on 1.5% agarose gels using SYBR Green I (Applied Biosystems) and scored visually as described by Smith et al. (2007). Restriction fragment length polymorphism (RFLP) patterns were not compared across cores.
Sequencing and identification of EM root tips
From each sample, one to six representative clones for each unique RFLP pattern were sequenced with ITS1F using the BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems) on an ABI 3730xl capillary sequencer (Applied Biosystems) at the College of Agriculture and Environmental Sciences Genomics Facility, University of California, Davis. When sequences from a single RFLP pattern resulted in multiple taxa, all clones from that RFLP type were sequenced. Sequences were edited using sequencher v4.2 (Gene Codes, Ann Arbor, MI, USA) and examined by BLAST searches against GenBank (National Center for Biotechnology Information) and a database of sequences from fruiting bodies collected at the site (Smith et al., 2007).
Names were assigned to EM root sequences using the methodology of Smith et al. (2007). Briefly, EM root sequences matching fruiting bodies are designated by genus and species epithet. In cases where species identity was unclear, a genus name is followed by a Sierra Research Center (src) number. For species that did not match fruiting bodies, names are based on a combination of BLAST searching, sequence alignment, and advice from taxonomic experts (see Smith et al., 2007). When ITS sequences were insufficient for phylogenetic placement, we sequenced part of the 28s large subunit using LR3 and performed a similar analysis as described above.
ITS sequences with ≥ 97% identity were grouped together. This cut-off level is based on error rates generated by PCR, cloning, sequencing and interspecific variability within ITS and has been employed in other studies using ITS for EM species identification from roots and soil (Izzo et al., 2005; O’Brien et al., 2005; Walker et al., 2005). We used a traditional species concept for assigning names to sequences because almost half of the EM root tips were identified based on matches to fruiting bodies from the site.
Generation of chimeric sequences is a potential problem when amplifying mixed-template environmental samples (Qiu et al., 2001; Jumpponen, 2003). We checked for chimeras using both alignments and the Recombination Detection Program (RDP; Martin & Rybicki, 2000) as described by Smith et al. (2007). Six sequences were identified as possible chimeras and removed from the data set.
Data from four soil cores from each tree or plot were combined for statistical analyses. EM taxa were categorized into four fruiting habits: epigeous (aboveground fruiting), hypogeous (belowground fruiting), resupinate (crust-like) and unknown/asexual. We calculated the relative frequency for each group (e.g. epigeous, hypogeous or resupinate) as the number of occurrences of species from each group divided by the number of occurrences of all taxa on a per tree/plot basis. Differences in the relative frequency of Ascomycota taxa and the relative frequency of EM taxa with epigeous, hypogeous and resupinate fruiting habits on Q. douglasii and Q. wislizeni were assessed using analysis of variance (ANOVA) with terms for oak species, site and the interaction (oak × site). The primary overall significance level was controlled at P < 0.05 and pairwise comparisons between oaks were controlled at the P < 0.0125 level using Bonferroni correction. To improve homogeneity of variances, proportions were arcsine transformed. We calculated first- and second-order jackknife estimates of species richness using pc-ord 4.0 (MjM Software Design, Gleneden Beach, OR, USA).
Differences in means for the seven soil variables from under the canopies of Q. douglasii and Q. wislizeni were also analyzed using ANOVA with terms for oak species, site and the interaction (oak × site). The primary overall significance level was controlled at P < 0.05 and pairwise comparisons between oaks were controlled at the P < 0.007 level using Bonferroni correction. All ANOVA analyses were conducted in sas (SAS Institute, Cary, NC, USA).
Relationships between EM community composition and environmental variables were analyzed with multivariate procedures (canoco v4.5; Microcomputer Power, Ithaca, NY, USA) using the number of times each unique RFLP type was detected per tree as an approximate measure of abundance. The number of unique RFLP types is an estimation of abundance because this measure may be influenced by PCR and cloning bias. Site was used as a covariable; oak species were treated as nominal environmental variables. Two types of ordination methods that can be used with species composition and environmental data are canonical correspondence analysis (CCA) and redundancy analysis (RDA). These ordination methods are based either on a linear (RDA) or unimodal (CCA) response of species to environmental gradients. In deciding whether to use linear or unimodal ordination methods, beta diversity in community composition along the ordination axes (as measured by gradient length) should be taken into account (Lepš & Šmilauer, 2003). If the longest gradient length is < 3.0 then linear method is most appropriate; if the longest gradient is > 4.0 then unimodal methods should be used (Lepš & Šmilauer, 2003). In this study the gradient length was 3.85. Therefore, we used both CCA and RDA. For RDA analyses, data were log + 1 transformed. Environmental variables (soil pH, total N, total C, extractable P, soil moisture and litter depth) were tested for their contribution to the variation in the EM species data (Monte Carlo permutation test). In order to determine the influence of rare species on the ordination analysis we conducted CCA and RDA both with and without singletons. To quantify the proportional contribution of the significant environmental variables, we conducted variance partitioning as described in Lepš & Šmilauer (2003).
Ectomycorrhizal fungal diversity
Based on sampling of 6400 root tips, RFLP analysis of 3072 clones and sequencing of 1788 clones, total EM species richness on fine roots of Q. douglasii and Q. wislizeni was 140 taxa (Supplementary Material, Table S1). We identified 57 taxa (41%) based on sequence matches to fruiting bodies (sclerotia for Cenococcum) collected from the study area and surrounding woodlands. One species (Amanita phalloides) was identified based on > 99% similarity of the ITS region to sequences in GenBank. The remaining 82 taxa were identified to the level of genus (40), family (34) or order (eight). Sequence information from the 28s region was used for taxonomic placement of 57 taxa; all others were determined by ITS alone. Two root-associated fungi in the Helotiales (Cadophora sp. 1 and cf. Lophiostoma sp. 1) were also detected in five soil cores. Because these two species could not be unambiguously confirmed as EM symbionts, they were excluded from analyses.
Ectomycorrhizal communities on Q. douglasii and Q. wislizeni were similar in species richness and community structure at both sites (Supplementary Material, Table S2). A total of 88 EM taxa occurred on Q. douglasii roots, of which 40 (45%) were detected as fruiting bodies. A total of 92 EM taxa were found on Q. wislizeni roots, of which 44 (48%) were detected as fruiting bodies. Many species were rare (found in only one soil core). A total of 46 (52%) EM taxa were found in only one soil core from Q. douglasii and 44 (48%) taxa were found in only one soil core from Q. wislizeni. The mean number of species in a soil core was 6.5 and the range was 2 to 12. The mean number of species detected around a tree (four soil cores) was 19.6 and the range was 16 to 25.
Ectomycorrhizal community composition
Basidiomycota species richness was more than double that of Ascomycota (Basidiomycota = 100 taxa; Ascomycota = 40 taxa). Basidiomycota species richness was 54 taxa on Q. douglasii and 67 taxa on Q. wislizeni, while species richness of Ascomycota was 34 taxa on Q. douglasii and 25 taxa on Q. wislizeni. There was a greater relative frequency of Ascomycota on Q. douglasii than on Q. wislizeni (49% versus 31%, respectively; n = 8, F = 10.8, P = 0.0066) and site effects and the interaction were not significant. At the family level, the most frequently detected taxa on both oak species were in the Thelephoraceae (Table 1). Taxa in the Pyronemataceae, Tuberaceae and Cortinariaceae (mostly Inocybe) occurred frequently on roots of both oak hosts. By contrast, taxa in the Russulaceae were frequent and diverse on Q. wislizeni (relative frequency = 14.5%; species richness = 12) but less frequent and diverse on Q. douglasii (relative frequency = 3.7%; species richness = 4; Supplementary Material, Table S1). However, Kendall's rank correlation coefficient showed no significant difference between the frequencies of higher level taxonomic groups (families and orders) on Q. douglasii and Q. wislizeni (Kendall's τ = 0.29, P = 0.21; Table 1).
Table 1. Relative frequency of ectomycorrhizal families from root tips of Quercus douglasii and Quercus wislizeni trees
Relative frequency was calculated as the number of occurrences of species from each family divided by the total number of occurrences of all taxa. Within each phylum, taxa are ordered in decreasing frequency on Q. douglasii.1Number of genera is approximate because of taxonomic uncertainties.
Sebacinales is at the level of order because of taxonomic uncertainties within this group and recent evidence indicating that Sebacinaceae constitutes a new order, Sebacinales (Weiss et al., 2004).
Other Basidiomycota include taxa from 12 genera in 10 families, altogether accounting for 5% (Q. douglasii) and 12% (Q. wislizeni) of the relative frequency. Genera in this category include: Amanita, Boletus, Clavariadelphus, Clavulina, Entoloma, Gautieria, Hebeloma, Hygrophorus, Hysterangium, Laccaria, Melanogaster and Octavianina.
Cenococcum is listed at the genus level because it is an asexual ascomycete with unresolved classification.
Helotiales is listed at the level of order because we were unable to identify taxa to family.
Patterns of overall community structure were similar at the two sites; however, some individual taxa differed between sites. For example, Laccaria cf. bicolor was the dominant species on Q. wislizeni at site 1 (detected on all trees) but was detected on Q. wislizeni from only one core at site 2. Similarly, the second most frequent species on Q. douglasii at site 1, Tuber californicum, was not detected on Q. douglasii at site 2. Although some frequently occurring taxa differed between sites, many of the common species were found on both Q. douglasii and Q. wislizeni (Fig. 1). For example, Cenococcum geophilum (lineage 2) (Douhan & Rizzo, 2005) and Tuber cf. candidum were detected frequently on both oak species. Overall, 40 taxa were found on both hosts and Q. douglasii and Q. wislizeni shared 13 out of the 16 most frequently occurring EM fungi (Fig. 1).
Differences in fruiting habit
Most EM taxa found on oak roots formed fruiting bodies that were epigeous (49) or resupinate (48). Less common fruiting types were hypogeous (30), asexual (3) and unknown (10). The roots of Q. wislizeni had a significantly higher relative frequency of taxa with epigeous fruiting than Q. douglasii (n = 8, F = 21.41, P = 0.0006; Fig. 2a,b); site effects (n = 8, F = 0.55, P = 0.471) and oak × site interaction (n = 8, F = 2.87, P = 0.116) were not significant. Two epigeous taxa, Laccaria cf. bicolor and Lactarius xanthogalactus, were common on Q. wislizeni and infrequent or absent on Q. douglasii (Fig. 1). The species richness of epigeous fruiting types was 25 taxa on roots of Q. douglasii and 32 on roots of Q. wislizeni. The relative frequency of taxa with hypogeous fruiting on the two oak species differed between sites, as indicated by a significant oak × site interaction (n = 8, F = 13.41, P = 0.003). Analysis of the interaction showed that Q. douglasii had a significantly higher relative frequency of hypogeous taxa at site 1 (P = 0.0023; Fig. 2a) but there were no significant differences between hypogeous taxa on oak species at site 2 (P = 0.9716; Fig. 2b). The relative frequency of taxa with resupinate fruiting did not differ significantly between Q. douglasii and Q. wislizeni (n = 8, F = 0.09, P = 0.770) and site effects (n = 8, F = 0.12, P = 0.737) and oak × site interaction (n = 8, F = 1.51, P = 0.243) were not significant.
Soil samples from under the canopy of Q. wislizeni had a deeper litter layer than those from under the canopy of Q. douglasii (n = 8, F = 35.5, P < 0.0001; Supplementary Material, Table S3). There were no significant differences in total C, total N or extractable P in soil samples from Q. douglasii and Q. wislizeni (Supplementary Material, Table S3). Differences in soil moisture between the two oak species were not significant after Bonferroni correction (n = 8, F = 6.05, P = 0.030). However, there was a significant site effect; soil moisture was higher at site 1 than at site 2 (n = 8, F = 34.75, P < 0.0001). Soil pH measured in water was higher under Q. douglasii at site 1, but showed no pattern at site 2 (significant oak × site interaction; n = 8, F = 4.73, P = 0.050).
Direct gradient ordination of EM fungal communities using RDA showed that the conditional effect of extractable P (i.e. the effect in addition to the other variables included in the model) was highly significant (P = 0.004) in explaining the distribution of EM species (Fig. 3). Oak species was the second most significant variable (P = 0.024). Variance partitioning revealed that extractable P explained 11.3% and oak species 9.6% of the variation in the species data. The two variables together explained 20.9% of the variance, demonstrating that there is no shared effect of these two variables, which is consistent with the corresponding vector and centroids in Fig. 3. The oak × extractable P interaction was not significant (P = 0.12), indicating that soil extractable P concentrations are essentially independent of oak species identity. Other environmental variables had no significant conditional effect. However, the independent effect of soil pH in CaCl2 was almost significant (P = 0.054) and this variable explained an additional 7.4% of the variance. Results were similar both with and without singletons and results from CCA were consistent with those from RDA (data not shown).
Previous studies have suggested that closely related host species, such as trees from the same genus or family, support similar EM communities (Horton & Bruns, 1998; Cullings et al., 2000; Ishida et al., 2007). In this study, many of the most frequent EM species were found on both Quercus hosts; however, the EM communities on the two hosts differed in several notable ways. We found a greater diversity and frequency of taxa that produce epigeous sporocarps on the roots of Q. wislizeni as compared with Q. douglasii. This finding is consistent with observations of aboveground fruiting patterns; we observed greater fruiting of epigeous taxa under Q. wislizeni than under Q. douglasii (M. E. Smith & M. H. Morris, pers obs), suggesting that environmental conditions under Q. wislizeni may be more favorable for epigeous taxa. This pattern is exemplified by taxa in the family Russulaceae, which were primarily epigeous and tended to be more frequent on Q. wislizeni (relative frequency = 14.5%) than on Q. douglasii (relative frequency = 3.7%). Smith et al. (2007) also found a similarly low relative frequency (2.9%) of taxa in the Russulaceae on Q. douglasii after intensively sampling roots from 94 soil cores over a 2-yr period.
By contrast, Ascomycota were significantly more frequent on the roots of Q. douglasii as compared with those of Q. wislizeni. Taxa in the Ascomycota are increasingly recognized as an important component of EM communities (Vrålstad et al., 2002; Dickie & Reich, 2005; Tedersoo et al., 2006; Smith et al., 2007), and this study demonstrates the prevalence of Ascomycota as EM fungal partners on both Q. douglasii and Q. wislizeni. We detected Ascomycota from a wide range of lineages (Pyronemataceae, Tuberaceae, Pezizaceae, and others) and the three most common species were Ascomycota (accounting for 14% of species occurrences). This evidence suggests that Ascomycota are not only diverse but may also play an important ecological role in this ecosystem.
Although we did not experimentally explore the reasons for the differences between the EM communities on the roots of the two Quercus species, these hosts differ in several important ways. Evergreen versus deciduous habit is one important difference between the two oak species. In addition, Q. wislizeni has a deeper litter layer, less herbaceous vegetation and fewer exotic species in the understory than Q. douglasii (Frost & Edinger, 1991; Rejmánek et al., 2005; this study). It is likely that these physiological and ecological differences between the two host species act in concert to influence their associated EM communities. For example, leaf litter can strongly influence EM communities (Baar & de Vries, 1995; Conn & Dighton, 2000; Dighton et al., 2000; Cullings et al., 2003) and increased litter accumulation under Q. wislizeni, but not Q. douglasii, may reduce soil evaporation (Williams et al., 1990). Although soil moisture was not significantly different between oaks at the end of the rainy season, moisture differences are likely to be more pronounced at other times of the year. Not only the amount but also the composition of leaf litter has been shown to influence EM fungi (Koide et al., 1998; Conn & Dighton, 2000). Evergreen and deciduous leaves are known to differ in chemical composition, with evergreen leaves having higher lignin concentrations than deciduous leaves (Aerts, 1995). Reduced lignin and litter accumulation in the canopy under Q. douglasii may favor Ascomycota taxa that are considered less efficient at decomposition of lignified organic matter (e.g. litter and wood) than Badisiomycota (Brundrett, 2002).
Of the eight environmental variables in the RDA ordination, species distributions were most strongly associated with extractable P and secondarily with oak host species. The importance of EM fungi in P uptake is well known (Smith & Read, 1997), but the role of P in influencing EM community composition is not well understood. Heterogeneity of extractable P may create niches for different species of fungi. Harrington & Mitchell (2005) found that soil organic matter and extractable P contributed significantly to explaining the distribution of EM types on Dryas octopetala in grass heaths. Independent of extractable P, the distribution of EM fungi was also influenced by oak host species, suggesting that both variables influence the EM communities, but it is unclear how these factors are related. Future research on the complex interactions among host plant, soil properties and soil biota is needed in order to better understand the direct and indirect effects of host species on EM communities. This research demonstrates significant differences in EM communities on two congeneric tree species, indicating that in this oak woodland phylogenetically similar plant hosts can create distinct ecological niches for EM fungi.
We thank the University of California Sierra Foothill Research and Extension Center for logistical support in the field, M. A. Perez-Perez for assistance with data collection and R. M. Davis for advice and suggestions. We are especially grateful to G. W. Douhan for his valuable contributions to this research. Three anonymous reviewers provided helpful comments on earlier drafts of this manuscript. This research was supported by a grant to C. S. Bledsoe from the National Science Foundation Biocomplexity program (DEB-9981711).