• Ectomycorrhizal (ECM) symbiosis is a widespread plant nutrition strategy in Australia, especially in semiarid regions. This study aims to determine the diversity, community structure and host preference of ECM fungi in a Tasmanian wet sclero-phyll forest.
• Ectomycorrhizal fungi were identified based on anatomotyping and rDNA internal transcribed spacer (ITS)-large subunit (LSU) sequence analysis using taxon-specific primers. Host tree roots were identified based on root morphology and length differences of the chloroplast trnL region.
• A total of 123 species of ECM fungi were recovered from root tips of Eucalyptus regnans (Myrtaceae), Pomaderris apetala (Rhamnaceae) and Nothofagus cunninghamii (Nothofagaceae). The frequency of two thirds of the most common ECM fungi from several lineages was significantly influenced by host species. The lineages of Cortinarius, Tomentella–Thelephora, Russula–Lactarius, Clavulina, Descolea and Laccaria prevailed in the total community and their species richness and relative abundance did not differ by host species.
• This study demonstrates that strongly host-preferring, though not directly specific, ECM fungi may dominate the below-ground community. Apart from the richness of Descolea, Tulasnella and Helotiales and the lack of Suillus–Rhizopogon and Amphinema–Tylospora, the ECM fungal diversity and phylogenetic community structure is similar to that in the Holarctic realm.
Semiarid sclerophyll habitats are the most widespread ecosystems in Australia and thus, much of the mycorrhizal research has focused on these habitats. The ecology of ECM fungi in moist coastal and submontane forests has been relatively little studied (Reddell et al., 1999), although rain forest habitats prevailed before 30–25 million yr ago (Mya). The opening of the Tasman sea and rapid northward movement of the Australian continent affected the ocean currents, which in turn caused progressive global cooling and drying particularly in the Southern Hemisphere (Crisp et al., 2004; Hill, 2004). From 25 Mya, these changing climatic conditions have enabled the evolution and radiation of present-day sclerophyll communities from rain forest-inhabiting ancestors (Ladiges et al., 2003, 2005; Steane et al., 2003; Crisp et al., 2004). Wet forests dominated by gymnosperms and ECM Nothofagus were successively replaced by sclerophyll and scrub vegetation dominated by ECM Leptospermoideae (including Eucalyptus), Casuarinaceae and Mimosaceae (Crisp et al., 2004; Hill, 2004). Therefore, certain Nothofagus-associated ECM fungi were hypothesized to have switched to Leptospermoideae and other plants following major changes in vegetation (Horak, 1983; Bougher & Malajczuk, 1985; Bougher et al., 1994).
Many stipitate, ‘agaricoid’ genera comprise a large number of secotioid or hypogeous-fruiting members in Australia (Bougher & Lebel, 2001). A number of these Australian hypogeous taxa have been described as entirely new genera or families (e.g. Trappe et al., 1996). Their high abundance in Australian semiarid woodlands has been attributed to seasonal climate and coevolution with small marsupials that consume and distribute these taxa (Johnson, 1996; Trappe & Claridge, 2005). Surprisingly, most nonhypogeous Australian ECM fungal genera are shared with the Holarctic realm, indicating vicariance/dispersal events (May & Simpson, 1997). Watling (2001) hypothesized that many ECM boletes may have followed the migrating vegetation from Indo-Malay and New Guinea to Australia via Pleistocene and earlier land bridges, which explains their wide distribution both in Southeast Asia and Australia. Conversely, other fungal parasites and symbionts are shared between Nothofagus forests in New Zealand and southern South America, suggesting ancient vicariant distribution or more recent long-distance dispersal (Bougher et al., 1994; Watling, 2001; Moyersoen et al., 2003). Certain taxa such as Mesophelliaceae (Trappe et al., 1996), Descolea (Bougher & Malajczuk, 1985) and Rozites (Bougher et al., 1994) are far more diverse in Australia compared with other continents, suggesting their Australian origin.
Based on the history of Australian vegetation, we hypothesize that ECM fungal communities are highly diverse and lack host specificity in a Tasmanian wet sclerophyll forest that comprises both temperate rain forest and true sclerophyll elements. This study further aims to uncover the relative importance of host species (i.e. direct root association) and host vicinity (the identity of nearest host tree) effects on ECM fungal community structure. Combining anatomotyping and sequencing, we demonstrate that the ECM fungal community of this forest is species-rich, phylogenetically diverse and substantially influenced by host trees.
Materials and Methods
Sampling was performed at Tall Trees Walk, Mt Field National Park, Tasmania (geocode 42°40.9′S, 146°42.2′E) in August, 2006. Mt Field National Park has a long history of conservation and recreational management, being first reserved in 1885 and proclaimed a national park in 1917. The vegetation of the study site forms a tall open forest with no logging history. Eucalyptus regnans F. Müll. (ECM host) forms a canopy at approx. 60 m. The subdominant canopy layer consists of Pomaderris apetala Labill. (ECM host), Acacia verniciflua A. Cunn. (Confirmed nonECM at this site), Nothofagus cunninghamii (Hook.) Oerst (ECM host), Atherosperma moschatum Labill., Olearia argophylla (Labill.) Benth. and a few Acacia melanoxylon R. Br. (Putative ECM host). The understorey is covered by tree ferns (Dicksonia antarctica Labill.), Pittosporum bicolor Hook. and Coprosma quadrifida (Labill.) Rob. Forest floor is covered by ferns, including Histiopteris incisa (Thunb.) J. Sm., Hypolepis rugosula (Labill.) J. Sm. and Blechnum spp., and various bryophytes. Decaying boles and branches of all decomposition stages are abundant. Soils are derived from Permian mudstone and siltstone parent material and are deep gradational clay loam over light brown clay. The mean annual rainfall averages 1224 mm and mean daily minimum and maximum temperatures range between 5.3°C and 16.2°C (Maydena Post Office Station # 095063, 1992–2004).
Three 1-ha plots were established 100–500 m apart in sites where at least three ECM host trees –E. regnans, N. cunninghamii and P. apetala– co-occurred. Plots 1 and 2 were situated on a south-easterly aspect of a slope of approx. 1–5°, whereas plot 3 was situated on steep (slope 10–25°), eastern and western banks of a stream. In plots 1 and 2, all three hosts grew mixed, whereas in plot 3, N. cunninghamii and E. regnans inhabited low river banks and elevated sites, respectively. From each plot, five root samples (15 × 15 cm to 5 cm depth) were collected from 0.2–1.5 m distance to trunks of each three host species. Roots were separated from remaining soil particles in tap water. After careful examination, ECM roots from each sample were sorted by plant species based on colour, ramification pattern and thickness. Using a stereomicroscope, ECM morphotypes were distinguished based on colour, roughness of mantle surface, occurrence of rhizomorphs, emanating hyphae and cystidia. The ECM root tips were assigned to morphotypes and scored for relative abundance (percentage of all ECM root tips) on each putative host species separately. Several ECM clusters of each morphotype were mounted into 1% CTAB (cetyltrimethylammonium bromide) DNA extraction buffer (100 mm Tris-HCl (pH 8.0), 1.4 m NaCl, 20 mm ethylenediaminetetraacetic acid (EDTA), 1% CTAB) for storage and transportation. Roots were processed within 5 d of collection. Several CTAB-stored root tips from each morphotype per host were further anatomotyped. To improve precision, anatomotypes were determined in each plot separately. One or more root tips of each anatomotype per host and plot were carefully cleaned from adhering debris, dense clumps of extraradical mycelium and rhizomorphs, and frozen until molecular analyses.
The DNA extraction was performed using a High Pure polymerase chain reaction (PCR) Template Preparation Kit for Isolation of Nucleic Acids from Mammalian Tissue (Roche Applied Science, Indianapolis, IN, USA) as outlined in Tedersoo (2007). Microscopical examination revealed frequent ascomycete hyphae in the ECM mantle and PCR often resulted in multiple or no product using universal fungal primers. Therefore, new taxon-specific primers were developed in rDNA nuclear large subunit (nLSU) to selectively amplify the putative ECM mycobiont. All primers were manually designed based on a common alignment comprising plant, ascomycete and basidiomycete sequences and to perform specifically at 55°C (calculated TM 56–60°C).
The rDNA internal transcribed spacer (ITS) and ca. 350 bp of nLSU were amplified using a primer ITS1F (5′-cttggtcatttagaggaagtaa-3′) in combination with a newly designed basidiomycete-specific primer LB-W (5′-cttttcatctttccctcacgg-3′) or ascomycete-specific primer LA-W (5′ cttttcatctttcgatcactc 3′) following the protocol of Tedersoo (2007). Tomentella–Thelephora, Sebacina and Tulasnella morphotypes were amplified using a primer ITS1F or ITS5 (5′-ggaagtaaaagtcgtaacaagg-3′) in combination with newly designed primers LR5-Tom (5′-ctaccgtagaaccgtctcc-3′), LR5-Seb (5′-attcgctttaccgcacaagg-3′) or LR3-Tul (5′-bactcgcatgcaaggtgca-3′), respectively. Later, ITS1F was substituted by another primer 1–2 bp downstream (5′-acttggtcatttagaggaagt-3′) that apparently differs from the recently published primer ITS0F-T (Taylor & McCormick, 2008) by lacking a single nucleotide in the 5′-terminus and is similarly experimentally proven to amplify fungi, including Tulasnellales. The nLSU was amplified using a primer LR0R (5′-acccgctgaacttaagc-3′) in combination with newly designed basidiomycete-specific primers LB-Y (5′-tttgcacgtcagaatcgcta-3′) or LB-Z (5′-aaaaatggcccactagaaact-3′), ascomycete-specific primer LR3-Asc (5′-cacytactcaaatccwagcg-3′) or fungal-specific primer LR5-F (5′-cgatcgatttgcacgtcaga-3′). All these primers are characterized in Fig. 1 and the Supporting Information Text S1. Since the ITS and nLSU of several anatomotypes could not be amplified, mitochondrial LSU and nuclear rDNA Small Subunit amplifications were attempted instead, using a combination of several primers, but these recovered just a few extra species.
The PCR products were checked on 1% agarose gels under UV-light and purified using Exo-Sap enzymes (Sigma, St Louis, MO, USA). Sequencing was performed using primers ITS4 (5′-tcctccgcttattgatatgc-3′), ITS5 and/or a newly designed universal sequencing primer LF340 (5′-tacttgtkcgctatcgg-3′) for the ITS region; ctb6 (5′-gcatatcaataagcggagg-3′) and/or LR5 (5′-tcctgagggaaacttcg-3′) for the nLSU. Contigs were assembled using Sequencher 4.7 (GeneCodes Corp., Ann Arbor, MI, USA). A value of 97.0% ITS region identity (excluding flanking 18S and 28S rDNA sequences) was used as a DNA barcoding threshold (molecular species criterion; Tedersoo et al., 2003). For Cortinarius and Laccaria, a 98.0% threshold was used instead, because the ITS region is more conserved in these genera (Frøslev et al., 2005; L. Tedersoo, pers. obs.). All unique ITS sequences were submitted to UNITE (Kõljalg et al., 2005) and the European Molecular Biology Laboratory (EMBL) databases. blastn searches were performed against public sequence databases NCBI, etc., and UNITE to provide as precise identification for the ECM fungi as possible.
To confirm the identity of a host tree, the plastid trnL region of root tip DNA was amplified using primers trnC (5′-cgaaatcggtagacgctacg-3′) and trnD (5′-ggggatagagggacttgaac-3′). Two or more root tips per putative host and soil sample were checked to confirm host identification. Size differences of the trnL region, as revealed from 1% agarose gels, effectively distinguished the three host trees from each other (not shown).
Using a computer program estimates ver. 8 (Colwell, 2006), sample-based rarefaction curves and minimal total species richness estimates Chao2, Jackknife2 and ACE were calculated for each host species and the whole community. In these analyses, root samples were used as sampling units and fungal species were sampled randomly without replacement.
To study the effects of host species and host vicinity on frequency of ECM fungal species, Fisher's Exact tests were calculated at significance level α = 0.05. Differences in species richness (square root-transformed) and relative abundance (arcsine square root-transformed) of ECM fungal lineages by host species were tested using anova. To control false discovery rate and reduce familywise error rate associated with multiple testing, a sharpened procedure of Benjamini–Hochberg correction was used instead of classical Bonferroni correction, as implemented in Verhoeven et al. (2005). The combination of these tests required at least three and four observations of each species to obtain statistically significant results for studying host species and host vicinity effects, respectively.
Using pc-ord ver. 5.04 (McCune & Mefford, 2006), Detrended correspondence analysis (DCA) was employed to unravel the effects of host species, host vicinity and plot on ECM fungal community structure. Root samples with binary-encoded species data and downweighting rare species were used in the analysis. All factors were transformed to dummy variables because of multiple levels and unordered status.
Combining DNA barcoding and anatomotyping, 123 taxa of putative ECM fungi were recovered from root tips of the three host species (see the Appendix). Among these, six distinct anatomotypes remained unamplified using various primers on several nuclear and mitochondrial DNA regions. Certain species of Tulasnella, Endogone and Amanita proved most difficult to amplify. Many additional sequence types identified as members of Helotiales or Sordariales of the ascomycetes were recovered from ECM root tips, but excluded from the analyses, because they were considered root endophytes or saprobes. These ‘contaminants’ were separated from putatively ECM Helotiales by comparing mantle anatomy and phylogenetic affinities (not shown).
Neither rarefaction curves nor minimal species richness estimates (except Chao2) approached an asymptote with increasing sample size (Fig. 2), indicating the need for enhanced sampling effort. Of the estimators, Chao2 produced more stable and consistent estimates than Jackknife2. The total ECM root-associated fungal community at Tall Trees Walk was estimated to comprise between 210 (Chao2) and 247 (ACE) species (Fig. 2).
The ECM fungal community comprised a few frequent species and a large number of rare species (Fig. 3). In particular, 68 (55.3%) of the taxa were found only once. Laccaria sp1 and Lactarius eucalypti were the most frequent, colonizing 44.4% and 42.2% of root samples, respectively. Cortinarius (incl. hypogeous members; 28 spp.), Tomentella–Thelephora (18 spp.), Russula–Lactarius (10 spp.), Clavulina (9 spp.), Descolea (including Setchelliogaster and Descomyces; 8 spp.) and Laccaria (5 spp.) were the most species-rich lineages of ECM fungi. Of ascomycetes, both Cenococcum sp., members of Pezizales (2 spp.) and Helotiales (5 spp.) were detected as ECM symbionts, but, in total, they colonized just 9.0% of root tips.
All three host tree species supported multiple ECM fungi and displayed no significant specificity for any fungal lineage in terms of species richness or abundance on root tips (see the Supporting Information, Fig. S1a,b). Based on rarefaction curves, roots and soil of P. apetala supported more species of fungi compared to E. regnans and N. cunninghamii, but overlapping confidence intervals suggested that this effect was nonsignificant (not shown).
The strongly overlapping host species and host vicinity effects of P. apetala respectively contributed 69.3% and 55.7% to the primary axis of DCA (eigenvalue 0.80) that explained 14.4% of variance in species data (see the Supporting Information, Fig. S2). The DCA main axis effectively separated ECM fungal communities of P. apetala root tips from these of E. regnans and N. cunninghamii. Based on arrow length, the combined host effects appeared more important than plot effect in explaining the ECM fungal community structure.
Most ECM fungi (56.4%) that were observed more than once colonized root tips of a single host species (Fig. 3). Differential presence of species on root tips and in the vicinity of host trees and plots could be statistically tested in 28, 19 and 19 of the species, respectively, given the integrated power of Fisher's exact test and sharpening of the reduced familywise error rate. The frequency of 10 species (52.6%) was significantly different in the vicinity of a certain host tree, whereas 19 (65.5%) species were significantly affected by roots of a host plant (Fig. 3). Pomaderris apetala, E. regnans and N. cunninghamii accounted for significant differences in 13, 4 and 2 of these host-biased fungal species, respectively. Host species effect was substantially stronger than host vicinity effect in 12 out of 15 cases (Fisher's exact test: n = 12, P = 0.030; see the Appendix). Plot had no significant effect on any species.
The ECM fungi formed a phylogenetically diverse community in a Tasmanian wet sclerophyll forest that compares well with Holarctic ecosystems (Horton & Bruns, 2001; Richard et al., 2005; Tedersoo et al., 2006; Ishida et al., 2007). The minimal species richness estimates of 210–247 spp. are probably strong underestimates of the local community richness, because richness estimators and rarefaction curves had not begun to approach an asymptote and the low sample size. Cortinarius, Tomentella–Thelephora, Russula–Lactarius, Clavulina, Descolea and Laccaria lineages were the most species-rich and abundant in the ECM fungal community of Tall Trees Walk in Tasmania. Except for Descolea, these lineages also form a substantial part of the Holarctic ECM fungal communities (Horton & Bruns, 2001; Izzo et al., 2005; Tedersoo et al., 2006; Ishida et al., 2007). In paleotropical ecosystems, the Boletaceae–Sclerodermataceae (represented by two rare species in this study), Tomentella–Thelephora and Russula–Lactarius account for the majority of ECM fungal taxa (Sirikintaramas et al., 2003; Riviere et al., 2007; Tedersoo et al., 2007), whereas Cortinarius, Descolea and Laccaria are uncommon (Peintner et al., 2003 but see Onguene, 2000; Tedersoo et al., 2007). In the Holarctic realm, individual species of Cortinarius and Laccaria usually occur below ground in low abundance and frequency owing to small genetic individuals and highly clumped distribution of ECM root tips (Gherbi et al., 1999; Genney et al., 2006). In Tasmania, however, Laccaria and Cortinarius were among the most abundant ECM taxa, colonizing, respectively 23.8%, and 10.9% of ECM root tips in this study, and 47% and 15% in the forest floor of an old-growth Nothofagus forest in Victoria state (Tedersoo, 2007). This phenomenon suggests that these lineages may have different ecological roles and importance compared with Holarctic ecosystems. The ECM fungal lineages of Descolea (Bougher & Malajczuk, 1985), Tulasnella and members of Helotiales (except Meliniomyces bicolor) commonly observed in Tasmania are seldom recorded in the Northern Hemisphere. The putatively ECM-forming species of Helotiales are closely related with many endophytic and/or ericoid mycorrhizal taxa (Vrålstad et al., 2002; Hambleton & Sigler, 2005) that rendered their distinction ambiguous solely based on DNA sequence data. Consistent features in mantle anatomy enabled us to separate these putative ECM taxa from root endophytes and saprobes, highlighting the importance of integrating these approaches when studying poorly known taxa such as Ceratobasidiaceae, Sebacinaceae, Tulasnellaceae and various ascomycetes that comprise closely related ECM and other root associated taxa. Based on this unreplicated study site, it is unwise to conclude on the absence of certain ECM lineages in Tasmanian wet sclerophyll forests or Australia in general. Indeed, two additional lineages, Elaphomyces and Sordariales, were found from Victoria despite the substantially lower sampling effort (Tedersoo, 2007). Nevertheless, the Suillus–Rhizopogon group of the Boletaceae–Sclerodermataceae lineage and Amphinema–Tylospora (Atheliales), which are relatively abundant in the Holarctic realm (Taylor et al., 2000; Horton & Bruns, 2001), were not observed below ground in Tasmania or Victoria. In agreement with this, none of these taxa are reported as native to Australia (May & Simpson, 1997).
Root morphology combined with length difference of chloroplast trnL region was successfully employed to distinguish the three host species below ground. Because most samples comprised roots of a single host species that grew the closest, separation of host root and host vicinity effects proved difficult using both ordination and statistical methods. Nonetheless, statistical analyses of individual species revealed that host preference was consistently stronger than host vicinity effect for individual species and the whole community. As suggested by DCA and the frequency of individual species, the ECM community of P. apetala differed from E. regnans and N. cunninghamii even in the vicinity of another host. Similar effects of host species on the ECM fungal community have been demonstrated in mixed forests of Japan (Ishida et al., 2007) and woodlands of California (Morris et al., 2008). Ishida et al. (2007) demonstrated that host effects on ECM community are stronger with increasing taxonomic distance and successional status of hosts.
Unlike previous studies, as many as 66% of the common ECM fungal species displayed statistically significant host preference, although exclusive specificity was less common among the most frequent species. Similarly, Ishida et al. (2007: 430) concluded that ‘a significant proportion of ECM fungi exhibited host specificity’, demonstrating that eight of 55 common species (15%) had statistically significantly biased host preference in a Japanese mixed forest. In the Holarctic realm, strictly host-specific taxa are usually restricted within a few fungal lineages (e.g. Alnicola, Leccinum, Rhizopogon, Suillus and Lactarius sect. Dapetes), whereas lineages of Laccaria, Tomentella–Thelephora, Cenococcum and Clavulina usually comprise promiscuous species (Horton & Bruns, 1998; Kennedy et al., 2003; Richard et al., 2005; Nara, 2006; Ishida et al., 2007; Morris et al., 2008; Tedersoo et al., 2008).
Host preference at any taxonomic level may provide new niches and hence support higher local species richness (Ishida et al., 2007), whereas host generalist ECM fungi are considered important drivers of forest succession by facilitating seedling establishment of late-successional host trees (Horton et al., 1999; Kennedy et al., 2003; Dickie et al., 2004; Richard et al., 2005; Nara, 2006). Following this hypothesis, P. apetala and E. regnans, both pioneer, fire-dependent tree species may effectively exclude each other through priority effect and hardly compatible ECM symbionts. Kope & Warcup (1986) reported no apparent host specificity of various Australian ECM plants (Pomaderreae and Nothofagus not studied) and fungi from early successional habitats in pure culture synthesis trials. Late-successional N. cunninghamii may be facilitated especially by shared ECM symbionts of eucalypts.
The ECM fungi from different lineages had biased association with P. apetala. In agreement, a computer program unifrac (Lozupone & Knight, 2005) revealed no significant phylogenetic difference in the frequency of ECM fungal species and lineages among the three hosts and plots (T. Suvi & K. Abarenkov, unpublished), indicating repeated, independent evolution towards host preference. The evolution of host preference may depend on ecological or genetic specificities such as substantial differences in phytochemistry and nutrient content between Pomaderris aspera Sieber ex DC. and E. regnans (Ashton, 1975a,b). Experimental synthesis trials involving more species of Eucalyptus and Pomaderreae are required to rule out the plant species effects on ECM specificity.
In conclusion, this study provides first evidence for the presence of plant communities with predominately host-biased, but not directly host specific ECM relationships. The underlying causes and mechanisms of this remain unknown, but deserve attention to learn the basic features of biogeography and host shifting in ECM symbiosis.
We thank G. Gates, G. Kantvilas, D. Ratkowsky, D. Puskaric and N. Ruut for support in Tasmania. This study was funded by Estonian Science Foundation grants nos 6606, GLOOM7434 and GDHLM0092J, the Doctoral School of Environmental Sciences and Kristjan Jaak scholarship. Three referees provided constructive comments on the manuscript.
Table Appendix. Identification and host preference of ectomycorrhizal fungi
Best blastn/fasta3 matches of the entire internal transcribed spacer (ITS) region to database sequences are shown. The P-values for host root and host soil preference are indicated. Statistically significant effects following the sharpening procedure of Benjamini–Hochberg correction are shown in bold type.
1 nd, not determined because of consistent failure of polymerase chain reaction (PCR) or insufficient statistical power; 2 Identification based on rDNA mitochondrial large subunit (mtLSU) sequence; 3 Identification based on rDNA nuclear large subunit (nLSU) sequence; 4 Partial, identification based on partially alignable internal transcribed spacer (ITS) sequence match.