•The absence of co-evolved mutualists of plants invading a novel habitat is the logical corollary of the more widely recognized ‘enemy escape’. To avoid or overcome the loss of mutualists, plants may co-invade with nonnative mutualists, form novel associations with native mutualists or form associations with native cosmopolitan mutualists, which are native but not novel to the invading plant.
•We tested these hypotheses by contrasting the ectomycorrhizal fungal communities associated with invasive Pinus contorta in New Zealand with co-occurring endemic Nothofagus solandri var. cliffortioides.
•Fungal communities on Pinus were species poor (14 ectomycorrhizal species) and dominated by nonnative (93%) and cosmopolitan fungi (7%). Nothofagus had a species-rich (98 species) fungal community dominated by native Cortinarius and two cosmopolitan fungi.
•These results support co-invasion by mutualists rather than novel associations as an important mechanism by which plants avoid or overcome the loss of mutualists, consistent with invasional meltdown.
An important factor in nonnative plant invasions is ‘enemy escape’, with plants in novel habitats having fewer pathogens, herbivores and parasites than in their home range or compared with co-occurring native species, increasing the ability of the plant to become invasive (Mitchell et al., 2006; Reinhart & Callaway, 2006). Less attention has been given to an inevitable corollary, that plants in novel habitats will encounter fewer coevolved mutualists (Pringle et al., 2009). This process may have significant ecological importance, as the vast majority of plants rely on one or more mutualisms, including mycorrhizal associations with fungi, nitrogen-fixing symbioses with bacteria and actinomycetes, foliar endophytic fungi, and pollination and seed dispersal mutualisms, among others (Christian, 2001; Rudgers et al., 2005; Tylianakis, 2008; Brundrett, 2009). The higher frequency of nonmycorrhizal plant families among exotic compared with native plants in some regions suggests that dependence on mutualists can be an important limitation on plant invasion (Pringle et al., 2009). In some cases, plants may evolve to become less dependent on mutualists during invasions (Seifert et al., 2009). Otherwise, plants must either form novel associations with native or previously arrived organisms or co-invade along with mutualists from the plant’s native range (Pringle et al., 2009).
Novel associations are frequently reported for plant invasions, potentially reflecting the relatively low specificity of many mutualisms (Richardson et al., 2000). In forming novel associations, there is potential for plants to gain disproportionate benefits, termed the ‘enhanced mutualisms hypothesis’ by Reinhart & Callaway (2006). It is also possible for invasive plants to acquire mutualists from previously nonsymbiotic native organisms. For example, DNA sequence divergence suggests that the N-fixing symbiotic Bradyrhizobium bacteria associated with nonnative legumes in New Zealand were previously present only as free-living organisms (Weir et al., 2004). Not all associations of nonnative plants with native mutualists would represent novel associations. Where a native mutualist has a cosmopolitan distribution, it may be native but not novel to the nonnative species. The occurrence of cosmopolitan associations would be consistent with the long-held concept that for fungi ‘everything is everywhere’, yet the applicability of this concept to ectomycorrhizal fungi is increasingly questioned (Peay et al., 2007; Lumbsch et al., 2008).
Other invasive plants appear to co-invade along with their mutualists, with a particularly good example in the ectomycorrhizal plant family Pinaceae. The Pinaceae is the most invasive family of shrub and trees, with Pinus being the most invasive genus within the family (Richardson & Rejmánek, 2004). Pinus seedlings establishing into novel habitats frequently coinvade with two closely related fungal genera, Rhizopogon and Suillus, both of which show host-specificity to the Pinaceae (Ashkannejhad & Horton, 2006; Collier & Bidartondo, 2009; Nuñez et al., 2009). Pinus plantations in the southern hemisphere also are dominated by nonnative ectomycorrhizal fungi, including Rhizopogon and Suillus, along with Thelephora, Pisolithus and a small number of other genera (Garrido, 1986; Chapela et al., 2001; Tedersoo et al., 2007). Ironically, initial plantings of Pinus in many parts of the southern hemisphere failed because of a lack of ectomycorrhizal fungi (reviewed in Marx, 1991; Pringle et al., 2009). The global introduction and spread of ectomycorrhizal inoculum has largely overcome this barrier to pine invasion (Vellinga et al., 2009), suggesting the absence of coevolved mutualists was an important limitation to pine establishment and growth before coinvasion by fungi. Nonetheless, much of the southern hemisphere has native ectomycorrhizal trees; including the Myrtaceae (e.g. Eucalyptus), Nothofagaceae, Pomaderidae (Rhamnaceae) and Dipterocarpaceae (Brundrett, 2009). This raises the possibility that where native ectomycorrhizal plants are present, Pinus could be forming novel associations as an alternative mechanism for avoiding mutualist loss.
Here we studied Pinus contorta invasions in high country ecosystems of the South Island of New Zealand. The ectomycorrhizal tree Nothofagus solandri var. cliffortioides once dominated these ecosystems, but human-induced fire and grazing has resulted in conversion of large areas to grassland ecosystems over the past c. 800 yr (McGlone et al., 1996). Baylis (1980) showed that Nothofagus establishment into these grasslands is limited by a lack of ectomycorrhizal fungi, with seedling establishment constrained to within the root zone of established trees. Intriguingly, this same grassland environment is now undergoing major invasion by ectomycorrhizal P. contorta, which shows no sign of being limited by a lack of ectomycorrhizal fungi. We tested whether Pinus, in this situation, has overcome the loss of mutualists by:
• forming novel associations with native ectomycorrhizal fungi,
• forming cosmopolitan associations with native, cosmopolitan fungi, or
• coinvading along with ectomycorrhizal fungi from the northern hemisphere.
Materials and Methods
We sampled a site where self-established invasive P. contorta Loudon and N. solandri var. cliffortioides (Hook f.) Poole co-occurred in a spatially interdigitated patches at the Craigieburn Forest Park, Canterbury, New Zealand (43°9′12′°S, 171°43′48′°E, approx. 1000 m elevation, 1447 mm mean annual rainfall with > 100 mm in most months, mean summer maximum 32.9°C, mean winter minimum −8.6°C; soils 10–15 cm depth yellow-brown earths over greywacke/argillite rock; Ledgard & Baker, 1988). The site has extensive native Nothofagus forests, along with small areas of the native myrtaceous ectomycorrhizal tree Leptospermum scoparium and forestry plantations (since the 1950s) of various nonnative ectomycorrhizal tree species, including several Pinaceae, Betula pendula, and various Salix, Alnus and Eucalyptus species. Self-establishing P. contorta seedlings are therefore likely to encounter a wide range of potential mutualist propagules.
Within the site we located five self-established stands of P. contorta between 9 yr and 20 yr of age, and five stands of N. solandri. Nothofagus stands were multi-aged, typically comprising one or more, large, older trees surrounded by a mixed-age cohort of younger trees. From each stand we collected five soil samples. Extreme soil rockiness prevented use of a soil corer, but we sampled roughly equal volumes (12 × 12 × 12 cm) using a narrow-blade spade. Samples were then returned to the laboratory (same day) and stored at 4°C for up to 5 d before processing.
We washed all roots over a sieve under running tap water, spread all roots from each sample in a dish, and took three samples of each distinct morphology of ectomycorrhizal root tip within each sample, also recording the number of root tips that were of that morphology (i.e. every single root tip was examined and classified on the basis of stereo and compound microscopy). The purpose of sorting by morphological groups rather than sampling root tips randomly was to overcome the highly uneven distribution of morphological groups within a core (on average a single morphotype comprised 67% of root tips within a sample, and the lowest abundance morphotype, where more than one type was detected, comprised < 4% of root tips on average); morphotypes were not used in analysis of data or to combine types across samples but only as a method to guide sampling of roots for molecular analysis. On average, Pinus samples contained 94 root tips (minimum 26, maximum 208) with a mean 1.5 morphotypes per sample and Nothofagus samples contained 535 root tips (minimum 199, maximum 1007) with a mean 3.5 morphotypes per sample. We used two samples of each morphotype group for subsequent molecular analysis, with a third used where PCR failed on both of the first two samples. After sampling individual morphotypes we sampled 60 ectomycorrhizal root tips (or all remaining tips for the five samples with < 60 remaining root tips) by repeatedly sampling at random to form a single pooled sample. To randomize, the collection dish was rotated more than a full rotation under a stereo microscope, the ectomycorrhizal root tip closest to an ocular cross-hair after rotation was sampled, then the dish was re-rotated until 60 tips were sampled. We placed samples in a 1.5 ml microtube and stored these at −20°C for up to 7 d before lyophilizing.
Species were identified through a combination of sequencing from clone libraries and terminal restriction fragment length polymorphism (T-RFLP), direct sequencing was also attempted (see next paragraph). We extracted and amplified DNA using the Sigma ‘REDExtract-N-Amp Plant PCR Kit’ (Sigma, part #XNAPR) with a modified protocol, as follows. To each individual sample we added 15 μl of supplied extraction solution, crushed the root tip with a pipette tip, incubated for 10 min at 95°C, and then added 15 μl of supplied dilution solution. Pooled samples were ground with a motorized plastic micropestle in 100 μl of extraction solution, incubated as for single root tips and diluted with 300 μl of supplied dilution solution. DNA was PCR-amplified using 0.4 μl of 50 mM fluorescently tagged 6FAM-ITS1F, VIC-ITS4 primers in a 20 μl volume with 0.4 μl genomic DNA extract and 10 μl supplied PCR mix. PerfectPrep (Eppendorf) Gel cleanup kits were used to clean successful PCR products (directly cleaning the samples without band cutting), before digestion with the restriction fragment length polymorphism (RFLP) enzymes BsuRI (Fermentas, Burlington Ontario, Canada) and HpyCH4IV (New England BioLabs, Ipswitch MA, USA). We denatured 1 μl of RFLP products at 95°C for 5 min in 10 μl HiDi formamide with 0.2 μl MapMarker 1000 ROX size standard (BioVentures, Murfreesboro TN, USA), and sized fragments in a Prism 3100 Genetic Analyzer (Applied Biosystems, Carlsbad CA, USA) in 36 cm capillaries with POP4 polymer, at the University of Canterbury Sequencing Centre (Christchurch, New Zealand). The T-RFLP patterns were matched against a database comprising over 900 patterns representing 430 T-RFLP groups obtained from sporocarps, environmental samples, and morphotype collections of native and exotic trees for species matching. For all unidentified T-RFLP patterns that occurred on more than one Pinus root sample we constructed a clone library and sequenced DNA using a cloning kit (Promega pGEM-T Vector System I A3600) modified to use a blunt-end ligation, and sequenced on a ABI3730 capillary sequencer (Allan Wilson Centre Genomic Service, Massey University, Manuwatu, New Zealand). Representative ITS DNA sequences were submitted to GenBank; accession numbers are in the Supporting Information, Table S1. No concerted effort was made to clone T-RFLP patterns found exclusively on Nothofagus as the lack of a T-RFLP match to a pattern found on Pinus was sufficient to test our hypotheses. Where T-RFLP patterns of multiple species clustered into a single group a more general group name was used (e.g. six potentially different Laccaria sp. based on sporocarp morphology produce highly similar T-RFLP patterns and therefore are clustered together). Grouping reflects lack of ITS sequence variation among species (Chase & Fay, 2009), or lack of clear species concepts for some New Zealand fungi, including potential misidentification of sporocarps, as well as limited resolution of T-RFLP (Dickie & FitzJohn, 2007).
Direct sequencing from root tips (as in Peay et al., 2009b) was attempted but abandoned because of very low quality sequence data. This likely reflects the 35% of Pinus and 55% of Nothofagus single root tip samples that contained multiple PCR products, consistent with other studies (e.g. 38% in Rosling et al., 2003). Direct sequencing studies often have a failure rate of c. 30% (e.g. Peay et al., 2009b), which is consistent with the 35% of samples where we found multiple T-RFLP patterns in Pinus. The higher frequency of multiple T-RFLP patterns in Nothofagus was largely driven by a ubiquitous Helotiales root endophyte. Although this could be avoided by the use of basidiomycete-specific probes, doing so could result in under-sampling of ascomycete mycorrhizas and a bias against some major basidiomycete groups (Tedersoo et al., 2003). In addition to multiple species in single root tip samples, further difficulty in direct sequencing may reflect the occurrence of intraspecific internal transcribed spacer (ITS) variation within individuals (Avis et al., 2006; Smith et al., 2007). For example, Suillus luteus, the second most common fungus on Pinus roots, had heterogeneous ITS sequences in both single root tip and sporocarp collections. These cause no particular difficulty in database T-RFLP analysis (both patterns are automatically pooled as representing a single species in TRAMPR analysis; FitzJohn & Dickie, 2007), or in clone libraries, but result in low quality sequence data in direct sequencing. To investigate further the presence of secondary T-RFLP patterns, we used a severe threshold (retaining only the single most intense peak in each T-RFLP enzyme–primer pair) to indicate which species were present as the ‘primary’ T-RFLP in at least one single root tip collection within a core and retained other T-RFLP patterns found as ‘secondary’ patterns. For pooled samples we expected multiple T-RFLP patterns, so did not distinguish primary and secondary patterns.
The T-RFLP patterns were considered to represent an ectomycorrhizal species or group of species if: the T-RFLP matched a sporocarp or DNA sequence of a known ectomycorrhizal genus (Rinaldi et al., 2008); or the T-RFLP pattern matched a T-RLFP profile from a previously described New Zealand ectomycorrhizal morphotype collection; or the T-RFLP was the primary pattern in at least one single root tip sample and no contrary evidence was present, assuming DNA of the ectomycorrhizal fungus would be much more abundant than DNA of any endophytic fungus that might be present. If a T-RFLP matched a sporocarp or DNA sequence of a known saprotrophic or endophytic fungus, these trophic states were assumed.
We considered a T-RFLP to represent a native species if the T-RFLP matched a native sporocarp or DNA sequence, or if the T-RFLP matched a known ectomycorrhizal morphotype collected from Nothofagus or other native plant species from this or other studies (I. A. Dickie et al., unpublished; Dickie et al., 2009b). Native species were considered to be cosmopolitan if they were also native to the northern hemisphere (the native range of Pinus). We considered a T-RFLP pattern to be nonnative if it matched a nonnative sporocarp or DNA sequence of a species known to be nonnative, or if the T-RFLP matched a known ectomycorrhizal morphotype collected exclusively from Pinus. In some cases the origin of a species was problematic to determine (cf. Pringle & Vellinga, 2006), these are also discussed individually later.
We recovered Pinus roots from 20 of the 25 samples under Pinus stands and Nothofagus roots from all 25 samples from under Nothofagus stands. In total, we collected 354 individual root tips representing 118 morphotypes within soil core combinations. More than 2500 additional root tips were sampled in pooled root tip sampling. Single root tip T-RFLP data were obtained from at least one collection of each morphotype within soil samples for all but one collection (three samples of that morphotype in that core all failed to produce PCR product). We obtained T-RFLP data from all but two pooled root samples (both from Nothofagus). Pooled and individual root tip sampling gave congruent views of fungal community composition associated with each tree species (see Notes S1 and Fig. S1). We were able to identify 88% of fungal–root associations of Pinus to taxa (species or genus level, based on number of occurrences as the primary T-RFLP pattern in single root tip T-RFLP), including all species that occurred in more than one root tip sample as the primary T-RFLP pattern.
Nineteen distinct fungal T-RFLP patterns were found associated with Pinus roots, of which we considered 14 ectomycorrhizal (Fig. 1). Nonnative fungi dominated the ectomycorrhizal fungal community on Pinus roots. The most common T-RFLP pattern found on roots of Pinus failed to match any sporocarp T-RFLP pattern, and when sequenced (representative sequences GenBank: GQ906424, GQ906425) had strong similarity to North American and European collections of ectomycorrhizal roots, but not to any DNA sequence from a sporocarp collection. Phylogenetic analysis suggested a placement in the Cantharellales, but did not place the sequence in any known family, hence we named this species ‘Cantharellales sp.’. The three next most frequent species were all nonnative –Suillus luteus, Amanita muscaria and an Atheliaceae (GenBank sequence GQ906429; likely a Tylospora sp.) – along with the less common nonnatives Tylospora sp. (GQ906431), Tricholoma sp. (GU220358), and Suillus granulatus (sporocarp T-RFLP match) and the root endophyte Phialocephala fortinii (GQ906427).
Four putatively native fungi were found on Pinus roots: two ectomycorrhizal species, one endophyte and one saprotroph. The ectomycorrhizal species Cenococcum geophilum was highly abundant on Nothofagus samples (Fig. 1) and was detected in three single root tip collections on Pinus, but not detected in pooled root tip samples. The ITS sequences of C. geophilum obtained from Nothofagus and Pinus ectomycorrhizas matched at 99% identity (397/398 bp; representative sequences in GenBank from Pinus: GQ906428, GQ906430; from Nothofagus: GQ906439). The ITS sequences of C. geophilum were inadequate to identify origin, as our sequences matched at 99% identity to other sequences from New Zealand, North America and Europe in blast searching. The other putatively ectomycorrhizal fungus was a Tomentella/Thelephora spp. (GQ906435, GQ906436; both from Nothofagus, the single Pinus sample failed in sequencing owing to problems with cloning kit cell viability and limited genomic DNA), matching at a maximum 95% identity to known sequences in GenBank and UNITE. A Helotiales root endophyte (GQ906426) was highly abundant in Nothofagus samples, and occurred in Pinus samples, but in Pinus was always a secondary T-RFLP pattern. The cosmopolitan saprotroph Bisporella citrina was also observed in one single root tip sample as a secondary T-RFLP pattern; fruiting bodies of this fungus are common in forest around the sampling area (I. A. Dickie, pers. obs.).
A total of 98 T-RFLP patterns were found on native Nothofagus solandri. Other than the shared T-RFLP patterns noted earlier in this text, the fungal community associated with Nothofagus was dominated by native Cortinarius, along with Laccaria spp., and some less common fungal groups (e.g. Russula, Austropaxillus).
The invasion of Pinus is clearly a coinvasion-dominated process, with 90% of ectomycorrhizal fungal species and 93% of ectomycorrhizal associations (soil samples × species) of Pinus being with nonnative fungi (Fig. 2). The fungal community observed was largely a subset of that associated with Pinus plantations in New Zealand (Walbert et al., 2010), although the most abundant species on invasive pine (Cantharellales sp., GQ906424) has not been recorded previously in New Zealand. Conversely, out of 97 fungal taxa associated with native Nothofagus, only three were also associated with nonnative Pinus, and only two of these were ectomycorrhizal. Cenococcum geophilum is cosmopolitan in distribution, hence the cosmopolitan hypothesis was supported for 7% of ectomycorrhizal associations of Pinus. No associations were found that unequivocally supported the novel associations hypothesis, although the origin of the Tomentella sp. remains unclear as there were no species-level matches to ITS sequence data. While we cannot rule out the possibility of novel associations occurring, we can conclude that they form, at most, a minor portion of the symbiotic associations of invasive Pinus contorta.
The widespread invasion of Pinus into grasslands in New Zealand contrasts with the failure of Nothofagus to spread into these same areas, despite an abundant local seed source. These grasslands were Nothofagus forest before human settlement (McGlone et al., 1996), indicating the potential for Nothofagus to grow on these sites. Baylis (1980) suggested that Nothofagus spread was limited by a lack of ectomycorrhizal fungi in grasslands, while others have reported similar limitation in other Fagaceae species (Dickie et al., 2005). The fungal mutualists of Pinus may be a major factor explaining the ability of Pinus to invade where Nothofagus does not establish. For example, Suillus sp. (including the second most abundant taxa on Pinus) have been reported as a dominant fungal associate of P. contorta establishing in Oregon (USA) sand dunes (Ashkannejhad & Horton, 2006), during early successional establishment of Larix kaempferi on Mt Fuji (Nara, 2006), and in Pinus mugo establishment in alpine grasslands in Switzerland (Wiemken & Boller, 2006). Suillus is animal-dispersed by deer (Ashkannejhad & Horton, 2006) giving Suillus a potentially unique ecological role in Pinus establishment in early successional and grassland habitats (Molina et al., 1992; Terwilliger & Pastor, 1999; Ashkannejhad & Horton, 2006). Deer, as well as other mammals, are introduced and abundant in New Zealand but their role in fungal dispersal has not been investigated. Further work on the ecology of Pinus-associated fungi is needed to determine to what degree the specific traits of Pinus-associated fungi contribute to the global success of Pinaceae as one of the most invasive of tree families (Richardson & Rejmánek, 2004).
The Cantharellales sp., despite being the most common fungal associate of P. contorta in this study and matching (98% identity) commonly encountered fungi in studies from North America and Europe (Landeweert et al., 2005; Ashkannejhad & Horton, 2006; Peay et al., 2009a), is known only by DNA sequence. Given the apparent ubiquity and possible ecological importance of this taxa, it would be useful to apply a taxon-specific name that could be applied consistently across research groups and provide an Internet-searchable ‘tag’ (Hibbett et al., 2009). Regrettably, the current taxonomic code has no provisions for naming fungi based only on DNA sequence data. Before this study, the fungus was not known to be present in New Zealand; having a recognized name for the species would aid invasive species control and border quarantine (biosecurity) efforts.
Although coinvasion was supported as important in Pinus invasion, a similar importance to other ectomycorrhizal plants cannot be assumed. Tedersoo et al. (2007) showed substantial variation among two ectomycorrhizal trees in the Seychelles, where novel associations with native ectomycorrhizal fungi are formed by Eucalyptus but not Pinus. Other symbioses also show variation among plant species. For example, in the actinorhizal symbiosis, Richardson et al. (2000) suggests that coinvasion by specific strains of Frankia is critical to Casuarina success, but novel Frankia associations are common in Elaeagnus and Alnus. In pollination mutualisms, novel associations appear common (Tylianakis, 2008), yet even here a lack of mutualists can limit invasion of some plant species (Stout et al., 2002). Ultimately, the importance of coinvasion likely depends on four factors: the symbiont specificity of the plant, the host specificity of local symbiont population, and the abundance and distributions of native and nonnative symbiont populations.
Six of the 19 total fungal taxa associated with pines remained unidentified, yet together these represent only 12% of the community. Our sampling was certainly insufficient to detect all species, as is typical of even far more intensive sampling efforts (Buée et al., 2009). Despite these limitations, we can safely conclude that the majority of Pinus mycorrhizal associations represent coinvasion.
Cenococcum geophilum was the most abundant fungus on N. solandri samples and occurred at lower frequencies in Pinus samples. Although C. geophilum is now recognized as a species complex (Douhan et al., 2007), the high affinity of sequences from Pinus and Nothofagus (1 bp difference) suggests a close taxonomic relationship. While C. geophilum is presently listed as an ‘unwanted regulated pest’ in New Zealand (http://www.biosecurity.govt.nz/pests/registers/uor; accessed 12 Feb 2010) the species is common in undisturbed native New Zealand forest (Mejstrik, 1970; I. A. Dickie, unpublished), and is much more abundant on native Nothofagus than on Pinus. We therefore assume that the C. geophilum on Pinus is likely to be a native cosmopolitan species, but molecular work using gene regions other than ITS would be necessary to confirm if the C. geophilum on Nothofagus and Pinus represents the same clade within the species complex (Douhan et al., 2007). Similarly, the failure to match any known sporocarp collection and a paucity of New Zealand Thelephoraceae collections in general prevents us from being certain of whether the Tomentella sp. is native. Nonetheless, there are native Tomentella in New Zealand (Cunningham, 1963) and the species was more abundant in Nothofagus than Pinus samples, suggesting a native status.
The other relatively common native fungus associated with Pinus appears to be a cosmopolitan Helotiales root endophyte, with an ITS rDNA sequence matching at > 98% identity to ascomycete strains from ectomycorrhizal and ericoid mycorrhizal hosts from North America (including Oregon, Alaska and California), Europe, subtropical China, and Australia (Ashkannejhad & Horton, 2006; Korkama et al., 2006; Taylor et al., 2007; Vincenot et al., 2008; Peay et al., 2009a; Wright et al., 2009). Along with C. geophilum, this T-RFLP type was one of the two most common fungi in Nothofagus roots, but was most frequently a secondary T-RFLP pattern and, where present as the primary T-RFLP was not associated with any particular morphological characters (morphotypes described variously as white, brown, gray, red, or black, with short hyphae or long hyphae, and as pinnate or monopoid). This makes it likely that the species is not ectomycorrhizal, but rather a widespread endophytic fungus with unknown effects on plant hosts (cf. Tedersoo et al., 2009).
While no evidence of native fungi forming novel associations with Pinus was found, A. muscaria, the third most abundant fungus on Pinus roots, is known from sporocarp surveys to have formed novel associations with native Nothofagus (Orlovich & Cairney, 2004), similar to the invasion of Amanita phalloides in North America (Wolfe et al., 2009). While this invasion was first observed only at forest edges and disturbed sites, A. muscaria is now spreading into native forest, sometimes accompanied by Chalciporus piperatus (Dickie & Johnston, 2008). This invasion appears to remain quite patchy, and no evidence of A. muscaria on Nothofagus roots was observed in the present study, despite known Amanita invasions within 1 km of the study area.
Fungal associations of native Nothofagus
In contrast to the Pinus ectomycorrhizal fungal community, the fungal community under Nothofagus was much more species rich and entirely dominated by native fungi. Many of the fungal genera (Cenococcum, Laccaria, Tomentella and Russula) present are similar to those genera found in northern hemisphere ectomycorrhizal fungal communities associated with Quercus (Dickie et al., 2009a). McKenzie et al. (2000) list 226 species of ectomycorrhizal fungi associated with Nothofagus in New Zealand, with Cortinarius being the most diverse genus (57 species), consistent with the dominance of Cortinariaceae in our results. The Cortinariaceae appear particularly important in Nothofagus forests of New Zealand, Chile, and Australia (Garrido, 1986). The genus Cortinarius does occur on Pinus contorta in its native range (Bradbury et al., 1998; Cullings et al., 2000), but no Cortinariaceae (native or nonnative) were found on Pinus samples in this study.
Conclusions and extension
Coinvasion by mutualistic symbionts such as ectomycorrhizal fungi represents a classic case of invasional meltdown (Simberloff & Von Holle, 1999). In the absence of exotic mycorrhizal fungi, Pinus plantations failed to establish in much of the southern hemisphere, supporting the loss of mutualists as a factor limiting invasion; however, once these fungi were introduced both the fungi and the plant were able to spread widely (Richardson et al., 2000). Intriguingly, Pinus spreads easily into grasslands and disturbed sites where no ectomycorrhizal trees are present and where native ectomycorrhizal trees are limited by a lack of mycorrhiza (Baylis, 1980).
A number of ectomycorrhizal trees are present in New Zealand but not yet invasive (e.g. Quercus, Eucalyptus). Of particular interest is Pseudotsuga menzeisii (Pinaceae), which is currently being planted partially on the basis of being less invasive than Pinus. As exotic fungi, particularly Rhizopogon, spread through intentional inoculation and secondary dispersal, and build up spore populations in soil, Pseudotsuga is likely to become significantly more invasive (Schwartz et al., 2006). A similar process may have occurred in Spain, where Eucalyptus has become more invasive following the introduction of Australian ectomycorrhizal fungi (Diez, 2005). These provide very real examples of the importance of understanding the role of belowground feedbacks and the network of ecological interactions among species as a potential tool for ecosystem management and invasive species control.
We thank V. Allison and K. Bonner for assistance in sample collection and processing, P. Novis and R. Smissen for help in molecular methods development, and R. Allen, M. Bidartondo, G. Grelet, P. Johnston, P. Kennedy, M. McGlone, M. St John and S. Richardson for helpful discussions and advice. This research is part of the Ecosystem Resilience OBI of the Foundation for Research, Science and Technology of New Zealand (C09X0502).