Most ectomycorrhizal (EM) plants associate with many unrelated ectomycorrhizal fungi (EMF), and many EMF associate with several different plant hosts (e.g. ‘multihost’ or ‘generalist’ fungi; Bruns et al., 2002; Kennedy et al., 2003). This generalist approach may enhance plant success, because seedlings dispersing into diverse habitats may find compatible EMF and may therefore have a competitive advantage (Bruns et al., 2002; Nara & Hogetsu, 2004). Association with many EMF may also enhance a plant's access to nutrients, particularly if certain EMF are adapted to unique microhabitats or have access to different nutrients (Baxter & Dighton, 2001). Roy et al. (2008) estimated that multihost EMF comprise between 12% and 90% of all species, and suggested that multihost EMF increase the chance of carbon transfer between different hosts, even when present in low frequencies. Although multihost EMF dominate many ecosystems (Horton & Bruns, 2001; Selosse et al., 2006), there are also important specialist taxa restricted to distinct host plant groups. EMF host specificity has been well documented in several genera, including Leccinum (den Bakker et al., 2004), Strobilomyces (Sato et al., 2007), Alnicola (Moreau et al., 2006), Alpova (Nouhra et al., 2005), Chondrogaster (Montecchi & Sarasini, 2001), Terfezia and Tirmania (Díez et al., 2002), and many Suillineae (Suillus, Rhizopogon, Truncocolumella and Gomphidius) (Bruns et al., 2002). It is also well established that some individual EMF species have strong host preferences (e.g. Lactarius obscuratus–Molina, 1979; Tuber melanosporum –Murat et al., 2004; Tricholoma matsutake–Lian et al., 2006), although these taxa may or may not be truly host specific.
To better understand host effects on EMF communities, we have previously documented the EMF as sporocarps and on root tips of Quercus douglasii Hook & Arn. and Quercus wislizeni A. DC. (Smith et al., 2007; Morris et al., 2008). These Quercus species are dominant, endemic EM trees throughout a large area of interior, low-elevation woodlands in California (Pavlik et al., 1991). Quercus douglasii is deciduous and usually has a sparse litter layer, whereas Q. wislizeni is evergreen and generally has deep litter. We hypothesized that these differences in physiology and litter deposition between the two oaks would influence their EMF communities. The same dominant EMF species occurred on both Quercus hosts, yet, paradoxically, the overall communities were distinct. Quercus wislizeni showed a greater diversity and frequency of EMF with epigeous sporocarps, whereas Ascomycota were more frequent on Q. douglasii (Morris et al., 2008). Thus, despite the close phylogenetic relationship between these plants, each hosted a unique EMF community. Multihost EMF were detected on both hosts, but some EMF taxa or functional groups apparently ‘preferred’ one host over the other.
Across most of their range, these Quercus species co-occur with another widespread endemic, Pinus sabiniana Douglas (Graves, 1932). These three species are the dominant EM hosts across a geographically large area, but no studies have yet examined the EMF of P. sabiniana. The objective of this study was to determine the dominant EMF on P. sabiniana roots and to compare them with the extensively documented EMF on the co-occurring Quercus species. We wanted to examine whether multihost EMF would be frequently shared among the three hosts, or whether pine-preferring EMF would be dominant on P. sabiniana. Several recent studies have shown that plant host identity strongly influences EMF communities and that this effect is exacerbated with increasing phylogenetic distance of the hosts (Ishida et al., 2007; Tedersoo et al., 2008). As the closely related Q. douglasii and Q. wislizeni had unique EMF communities, we hypothesized that the EMF community on the distantly related P. sabiniana would be even more divergent.
EMF sampling of P. sabiniana was conducted at the site studied by Smith et al. (2007) and Morris et al. (2008). The UC Sierra Foothill Research and Extension Center is located in a low-elevation woodland (400–600 m) with a strong Mediterranean climate in Yuba County, CA, USA (39°17′N, 121°17′W). We sampled EM roots on 1 April 2005 from eight randomly selected P. sabiniana < 50 m from previously sampled Quercus plots (Smith et al., 2007; Morris et al., 2008; ‘plot 2’). Pinus were small to large trees found c. 2–15 m from Quercus within the contiguous mosaic of woodland–savanna. We intended to sample P. sabiniana as in Smith et al. (2007) because we assumed that Pinus roots would be easily distinguished on the basis of size and color. However, these differences were not obvious and preliminary sampling yielded many roots that could not be unambiguously assigned to host genus. Therefore, one lateral root system per tree (main root > 3 cm in diameter) was randomly selected and cut at the base. We then excavated the root system and extracted several 50–200-cm root sections. Each extracted root system was placed in a plastic bag, stored at 4°C and processed within 5 d following the protocols of Smith et al. (2007) with minor amendments. We were unable to gather 100 EM roots for two trees, and so we used all root tips for these samples (pines 1 and 6 had 82 and 32 EM tips, respectively). Tuberculate EM roots formed by Rhizopogon were large relative to other EM tips, and so several EM tips per tubercule were added to the pooled root tips from that sample. Taxon naming and molecular analyses of roots followed Smith et al. (2007), except that 94 instead of 48 clones were restriction fragment length polymorphism (RFLP) screened from each sample.
Morris et al. (2008) collected samples based on focal trees, whereas Smith et al. (2007) collected samples from a plot with many trees. In this study, we used the individual samples; although some autocorrelation may be present for samples taken around focal trees, this should not affect the host preference aspects of the data. EMF species’ composition data were analyzed using principal component analysis (PCA) and multiple-response permutation procedures (MRPP) to collapse the large EMF dataset into groups and illustrate the separation of EMF communities by host. PCA depicts differences in species’ composition in relation to host, whereas MRPP is a nonparametric procedure that uses randomization to test the significance of pre-assigned variables (e.g. tree species). PC-ORD v. 4.20 was used to calculate the PCA axes based on correlation coefficients with the Sørensen distance measure, because it performs well with presence–absence data (McCune & Mefford, 1999).
We have previously sampled EMF at this site by extensive surveys of sporocarps and EM roots (Q. douglasii, 110 root cores with c. 11 000 roots; Q. wislizeni, 32 root cores with c. 3200 roots) (Smith et al., 2007; Morris et al., 2008). In this study, we sampled eight P. sabiniana (717 EM roots and 752 clones; mean of 89.6 roots per tree). We detected 33 EMF species, with two to nine species per lateral root (mean of 6.1 species). Basidiomycota were dominant on P. sabiniana, with 25 species accounting for 77.6% of the relative frequency (Table S1, see Supporting Information). We detected eight Ascomycota, but these only accounted for 22.4% of the relative frequency, and only two Ascomycota were detected on more than one Pinus. By contrast, 37–39% of the EMF species on Quercus were Ascomycota, and they accounted for 31–49% of the relative frequency. Furthermore, many dominant Quercus EMF were Ascomycota (e.g. Cenoccocum, Tuber, Peziza). Nineteen of the 33 EMF detected on P. sabiniana (c. 58% of the taxa, 65% of the relative frequency) had not been detected previously on Quercus, despite sampling > 140 root cores (14 000+ roots). This was true for seven of the 10 most common Pinus EMF, including all species that occurred on three or more Pinus. Five species in two lineages that associated only with Pinus (Rhizopogon, Atheliaceae) were major contributors to this pattern; together, these taxa accounted for 28% of the relative frequency on Pinus. PCA visually segregates the EMF communities by host plant, although there is significant overlap between Quercus species (Fig. 1). MRPP confirmed the separation (t-test; P < 0.0001), indicating greater variance between than within groups. To ensure that these host preferences were not strongly affected by the small P. sabiniana sample size, we performed 10 additional MRPP tests with reduced Quercus datasets. For each MRPP test, we randomly selected data from 10 Q. wislizeni and 10 Q. douglasii cores, and analyzed them with the entire P. sabiniana dataset. In each of the reduced datasets, MRPP confirmed the separation between the Quercus and Pinus ECM communities (P < 0.001), but the effect size remained low (R < 0.10). These low effect values are not surprising considering the smaller sample size and the nature of community data, yet the statistical analysis still confirmed higher homogeneity within groups than between. This pattern of host structured ECM communities was detected regardless of whether or not species of the Pinaceae specialist genus Rhizopogon were included in the PCA.
Although the majority of Pinus-associated EMF were only detected on Pinus, 14 EMF co-occurred on the roots of at least one Quercus species (c. 42% of taxa, 35% of relative frequency); all EMF species shared by both Pinus and Quercus were Basidiomycota. Five ‘generalist’ EMF were found on all three hosts (c. 15% of taxa, 10% of relative frequency). Pinus sabiniana shared about the same number of EMF with Q. douglasii (nine) as with Q. wislizeni (10). However, most of the EMF detected on both Pinus and Quercus tended to have low relative frequency on Quercus (< 2% relative frequency). Two notable exceptions, Laccaria bicolor (4.3% relative frequency on Q. wislizeni) and Thelephoraceae (3.5% relative frequency on Q. douglasii), may be strong ‘multihost’ fungi. Species of Laccaria may be preadapted as ‘multihost’ EMF because they are often pioneer species, regularly establish new genets from spores, are readily cultured in the laboratory and have great potential to become invasive (Gherbi et al., 1999; Kropp & Meuller, 1999; Fiore-Donno & Martin, 2001; Díez, 2005; Roy et al., 2008; Vellinga et al., 2009).
Many of the higher level EMF lineages detected on P. sabiniana were the same as those on Quercus (e.g. Thelephoraceae, Pezizales), but, at the species level, many EMF were different on the two host genera. Although based on a small Pinus sample, these data suggest the possibility that some EMF groups may be more or less likely to exhibit host preferences. For example, Pezizales appeared to be strongly structured by host; they were common on Pinus and Quercus, but none of the seven Pinus-associated pezizalean EMF species were ever detected on Quercus. Interestingly, we detected a similar pattern of host structuring for pezizalean hyphae at the same site (M. M. Hynes et al., unpublished). By contrast, other groups appeared to be less likely to exhibit host preference. For example, four of five Russulaceae were found on both Pinus and Quercus. This result for Russulaceae is similar to that presented by Ishida et al. (2007) where, of the 22 species that occurred more than once, 17 Russulaceae were found on more than one host plant species. These patterns of host preference may be system specific, however, as many studies have indicated that some Russulaceae exhibit strong host preferences and some Pezizales are host generalists (e.g. Hutchison, 1999; Kennedy et al., 2003; Ishida et al., 2007; Twieg et al., 2007; Tedersoo et al., 2008). Although it would be interesting if some EMF lineages tended to have stronger host affinity than others, this hypothesis requires further testing.
To date, we have detected 225 species of EMF from sporocarps and EM roots of three host species in this dry woodland habitat. This is closely comparable with the 205 EMF species reported by Ishida et al. (2007) from a mesic, closed-canopy forest containing 12 host species in six genera. In terms of community composition, however, the Pinus EMF community in this dry woodland was similar to other dry, inland Pinus EMF communities (Bidartondo et al., 2001; Hubert & Gehring, 2008) and to the post-fire spore bank EMF community in coastal Pinus muricata (Taylor & Bruns, 1999). It was dominated by species with inconspicuous sporocarps, and the hypogeous genera Rhizopogon and Tuber were abundant and species rich. We only detected 10 species with epigeous sporocarps (29% of the relative frequency), and only two of these (Hygrophorus cf. gliocyclus, Clavulina cf. cristata) inhabited multiple trees. Dominant genera from mature coastal pines, such as Suillus and Amanita spp., were notably absent on P. sabiniana.
Unfortunately, because of the destructive techniques needed to complete this study, we were unable to sample large numbers of Pinus roots or individuals. With further sampling, we would probably have detected more EMF species that are occasionally shared between Quercus and Pinus. However, we believe that the overall pattern would remain; except for a relatively small number of ‘multihost’ taxa, most of the dominant Pinus EMF rarely associate with or have a low frequency on Quercus, and most dominant Quercus EMF rarely associate with or have a low frequency on Pinus.
The prevailing ideas about EMF host preference may be changing. Early studies of Northern Hemisphere ecosystems, mostly dominated by Pinaceae and with routinely overlapping canopies, reported high diversity and abundance of shared EMF taxa between hosts, suggesting that ‘multihost’ fungi are dominant (Horton & Bruns, 2001; Horton et al., 1999, 2005; Cullings et al., 2000; Kennedy et al., 2003; Richard et al., 2005; Twieg et al., 2007; Hubert & Gehring, 2008) However, on a global scale, most EM host plants and ecosystems remain unstudied or undersampled (Dickie & Moyersoen, 2008). Sampling EMF communities with different sets of host plants (Ishida et al., 2007; Tedersoo et al., 2008) and in different habitats (e.g. tropical forest –Tedersoo et al., 2007; woodland–savanna –Morris et al., 2008) might change our view of host preference and, consequently, the perceived importance of ‘multihost’ fungi. The emerging picture is one in which many EMF appear to display some level of host preference, and most EM plants host some fraction of the locally available EMF. These complex interactions appear to be governed by some combination of host plant relatedness (Ishida et al., 2007; Tedersoo et al., 2007), plant life-history traits and physiology (Morris et al., 2008; Tedersoo et al., 2008) and successional strategies of both hosts and fungi (Horton et al., 2005; Twieg et al., 2007; Roy et al., 2008). The end result is that ‘multihost’ EMF fungi may be much less dominant than previously thought.