Ectomycorrhizal fungal communities in endangered Pinus amamiana forests

Interactions between trees and ectomycorrhizal (ECM) fungi are critical for the growth and survival of both partners. However, ECM symbiosis in endangered trees has hardly been explored, complicating conservation efforts. Here, we evaluated resident ECM roots and soil spore banks of ECM fungi from endangered Pinus amamiana forests on Yakushima and Tanegashima Islands, Kagoshima Prefecture, Japan. Soil samples were collected from remaining four forests in the two islands. The resident ECM roots in soil samples were subjected to molecular identification. Soil spore banks of ECM fungi were analyzed via bioassays using a range of host seedlings (P. amamiana, P. parviflora, P. densiflora and Castanopsis sieboldii) for 6–8 months. In all remaining P. amamiana forests, we discovered a new Rhizopogon species (Rhizopogon sp.1), the sequence of which has no match amoung numerous Rhizopogon sequences deposited in the international sequence database. Host identification of the resident ECM roots confirmed that Rhizopogon sp.1 was associated only with P. amamiana. Rhizopogon sp.1 was far more dominant in soil spore banks than in resident ECM roots, and its presence was confirmed in nearly all soil samples examined across the major remaining populations. While Rhizopogon sp.1 did not completely lose compatibility to other pine species, its infection rate in the bioassays was highest in the original host, P. amamiana, the performance of which was improved by the infection. These results indicate that Rhizopogon sp.1 is very likely to have a close ecological relationship with endangered P. amamiana, probably due to a long co-evolutionary period on isolated islands, and to play the key role in seedling establishment after disturbance. We may need to identify and utilize such key ECM fungi to conserve endangered trees practically.


Introduction
Forests have disappeared and deteriorated all over the world due to habitat destruction and environmental changes caused by human activity [1]. As a result, many tree species are threatened with extinction [2]. In particular, the dominant trees in a forest play critical roles in primary production and ecosystem structuring, directly and/or indirectly supporting many other organisms. Thus, the extinction of a dominant tree species can have a serious impact on a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 million years [12]. Notably, this fungus was completely absent in resident tree roots [32]. Because P. japonica is a light-demanding pioneer tree species, its seedlings are rarely found on the dark forest floor [33,34]. These findings strongly suggest that both P. japonica and R. togasawariana depend on disturbance for regenerating their populations, and that R. togasawariana plays the key role in the establishment of seedlings during regeneration. The existence of this ECM fungus specific to the endangered tree species and its potential role in establishment of endangered tree seedlings have critical implications for conservation; yet, it is difficult to generalize the results obtained from this single case. Similar studies in Pinus, which includes far more endangered species than those of Pseudotsuga, would further improve our understanding of ECM symbiosis on endangered trees.
In the present study, our objective was to characterize the ECM fungal communities of both soil spore banks and resident ECM roots in remaining P. amamiana forests. Specifically, we examined the following hypotheses: (1) P. amamiana is associated with some host specific ECM fungi that have coevolved with this endangered pine, and (2) these fungi are more frequently detected in soil spore banks than in resident ECM roots as Rhizopogon species in other ecosystems. We believe that the results of this study can provide key information for in situ conservation of P. amamiana and has many important implications for the conservation of other endangered trees.

Bioassays
Bioassay experiments were performed to assess the soil spore banks following an established method [11,12]. In these experiments, only soil samples collected in 2014 were used. After removing roots, organic debris and small stones, the soil samples were air-dried for 2-8 months (S2 Appendix). The bioassay containers were 50-mL centrifuge tubes (Ina Optika Inc., Osaka, Japan), with two drainage holes near the bottom blocked by a cotton ball to prevent soil loss. Each tube was filled with approximately 40 mL of the air-dried soil sample. We used four tree species, P. amamiana, P. parviflora, P. densiflora and Castanopsis sieboldii (Fagaceae), to evaluate the host specificity of soil propagule banks, particularly the specificity of the Rhizopogon spp., which are expected in spore banks. While P. parviflora belongs to the same subgenus (Strobus) as that of P. amamiana, P. densiflora belongs to the subgenus Pinus. C. sieboldii is a broadleaf tree species that is predominant in warm-temperate forests of the region. We were able to use only 46 P. amamiana seeds for all bioassay tubes because of the limited seed availability and low germination rate of this endangered species.
Seeds of all tree species were soaked in tap water for 48 h, surface-sterilized using a 5% sodium hypochlorite solution for > 10 min, and rinsed with tap water. To induce germination, surface-sterilized P. amamiana and P. parviflora seeds were placed on sterilized, well-moistened peat moss in an incubator (25˚C for 1 month and then 5˚C for 2 months), while P. densiflora and C. sieboldii seeds were placed on sterilized Shibanome soil (fine red granular soil of Pleistocene volcanic origin) in an incubator (25˚C). For each soil-host combination, one germinated seed was planted and covered with sterilized Shibanome soil in a bioassay tube, for a total of 283 seedlings (S2 Appendix). To monitor airborne spore contamination of ECM fungi, we prepared a control treatment using 10 tubes containing an autoclaved, randomly selected soil sample (per host). Bioassay seedlings were watered with tap water (2-3 mL per seedling) every 3-5 days depending on soil conditions and were harvested after 6-9 months of growth in a growth chamber (MLR-351; SANYO Inc. Tokyo, Japan) set to 25˚C with 16 h of light (15 fluorescent lamps: 20,000 lx) and then to 20˚C for 8 h of dark (S2 Appendix).

Identification of ectomycorrhizal fungi
Roots of resident trees collected from field soil samples and bioassay seedlings were gently washed with tap water. ECM root tips were classified into morphotypes under a dissecting microscope based on their surface color, texture and emanating hyphae, as described in previous studies [35,36]. For molecular identification of ECM fungi, triplicate ECM root tips, if available, were selected for each morphotype in a soil sample, then placed individually into 2.0-mL tubes. A total of 682 root tips from ECM were used for molecular identification on resident trees. In addition, 208, 246, 257 and 191 ECM root tips from the bioassays of the P. amamiana, P. parviflora, P. densiflora and C. sieboldii, respectively, were subjected to DNA analysis. Molecular identification of ECM fungi was performed following Murata et al. [12]. Briefly, each ECM tip sample was placed in a 2.0-mL tube containing a zirconia ball and then pulverized using a bead beater. Total DNA was extracted using a modified cetyl trimethylammonium bromide method [35]. Internal transcribed spacer regions (ITS1-5.8S-ITS2) of ribosomal DNA were amplified using the ITS1F or ITS0F-T forward primer and several reverse primers (ITS4, ITS4Cg, LB-W [37][38][39][40] depending on amplification success. Platinum1 Multiplex PCR Master Mix (Applied Biosystems, Foster City, CA, USA) was used for polymerase chain reaction (PCR).
The amplified products were purified and subjected to direct sequencing (3730xl DNA Analyzer; Applied Biosystems) using ITS1 as the sequencing primer. For poorly sequenced samples, ITS4 was also used as a sequencing primer. The obtained high-quality sequences were grouped into molecular operational taxonomic units (MOTUs) with ! 97% ITS sequence similarity (including 5.8S regions) using ATGC software (ver. 7.0; GENETYX Corp., Tokyo, Japan). Species identity was assigned based on the results of BLAST searches against known taxa in international sequence databases (DNA Data Bank of Japan [DDBJ]/EMBL/GenBank). Potential PCR chimeras were removed manually before further analyses, based on inconsistent BLAST results between segments within each MOTU sequence, e.g., between the initial and last parts (about 100 bp each) of the sequence. If the ITS similarity with a described species from herbarium specimens was > 97%, we used that species name for the MOTU. When the similarity to known species was 90-97% or < 90%, we identified MOTUs at the genus and family level, respectively. One exception was C. geophilum, which was identified by amplifying the Cenococcum-specific primer (ITS4Cg) and confirmed by sequencing randomly selected samples. Although several different MOTUs of C. geophilum were found (< 97% ITS sequence similarity), they were treated as one MOTU in this study, because no taxonomic agreement was reached among the many previous studies of ECM fungal communities (e.g., [36,41]). Representative sequences of individual MOTUs were deposited in the DDBJ under accession numbers LC315810-LC315919.
To confirm the identity of host species in all molecular samples, the plastid trnL or rbcL region of plant DNA was amplified using primers trnC (5 0 -cgaaatcggtagacgctacg-3 0 ) or rbcLa-F (5 0 -atgtcaccacaaacagagactaaagc-3 0 ) in combination with trnD (5 0ggggatagagggacttgaac-3 0 ) or rbcLa-R (5 0 -gtaaaatcaagtccaccrcg-3 0 ). All the amplicons were purified and subjected to direct sequencing using trnC or rbcLa-F as the sequencing primer. As with species identification of ECM fungi, the identity of host species was assigned based on the results of BLAST searches against known taxa in international sequence databases.

Statistical analyses
The frequency of an ECM fungal species was defined as the number of bioassay seedlings colonized by each ECM fungal species or, for resident trees, the number of soil samples containing each ECM fungal species. The frequencies of major fungal species were compared between different bioassay host species using Fisher's exact test implemented in SPSS (ver. 11.5; SPSS Japan Inc., Tokyo, Japan). Statistical significance was set at α = 0.05. Estimate S software (ver. 8.2; [42]) was used to calculate the minimum number of total species richness indicators (Jackknife 2).
A fungal community matrix was built for the major host groups (P. amaiana, T. sieboldii and Fagaceae) at each site. Community compositional similarity was visualized using nonmetric dimensional scaling (NMDS), implemented in the 'vegan' package of R software [43], using the Bray-Curtis distance (Manhattan distance for NMDS based on the unit of soil samples for resident ECM roots) and 999 permutations. Differences in community composition between resident ECM roots and assayed spore banks were determined using Adonis (permutation multivariate analysis of variance; [44]) in 'vegan', again using the Bray-Curtis distance and 999 permutations [45]. Data dispersions of the communities among groups were analyzed by the Betadisper test (Permutational analysis of multivariate dispersions; [44]) in 'vegan'.

ECM fungal communities of Pinus amamiana forests
Of the 104 soil samples collected from the four study sites, 44 contained P. amamiana ECM roots. Other dominant host taxa in the belowground ECM tips were T. sieboldii and Fagaceae, which were found in 9 and 58 soil samples, respectively ( Table 2). As a minor host, the subgenus Pinus was confirmed in one soil sample. No ECM roots were found in 15 soil samples.
Of the 101 putative ECM fungi that were identified in the four sites, 42 species were found on P. amamiana roots ( Table 2, S3 Appendix). Thirteen and sixty-one ECM fungi were found on T. sieboldii and Fagaceae roots, respectively. The rarefaction curves indicated that neither the observed ECM fungal richness on P. amamiana nor richness on all hosts approached an asymptote with the maximum sample size (Fig 1), indicating that additional fungal species would be found with additional sampling effort. The richness estimator Jackkife2 showed that at least 229 ECM fungal species should inhabit these forests (Fig 1). The Jackkife2 richness estimator for ECM fungi on P. amamiana was 102 (Fig 1). The observed ECM fungal richness Jackkife2 estimators on T. sieboldii and Fagaceae trees were 31 and 141, respectively.
The total ECM fungal community was composed of a few common species and a large number of rare species. Only eight ECM fungal taxa appeared in five or more soil samples. In contrast, 73 taxa were found only once (i.e., singletons). Cenococcum geophilum was the most frequent taxon, found in 39% of the soil samples. The occurrence frequencies of Russulaceae (48%), Boletaceae (36%), Clavulinaceae (17%), Rhizopogonaceae (14%) and Thelephoraceae (13%) were also high at the family level. Russulaceae (30 spp.) was the most species-rich ECM fungal lineage, followed by Boletaceae (19 spp.), Clavulinaceae (eight spp.) and Thelephoraceae (six spp.), while only two species of Rhizopogonaceae were found ( Table 2).
Only Rhizopogon sp.1 was found at all sites (Table 2). In addition, Boletaceae sp.1, Clavulinaceae sp.3, Phylloporus sp.1 and Thelephoraceae sp.1, among the major ECM fungal species with >5% relative frequencies, were only found on Yakushima Island (sites 1 and 2). In contrast, none of the major ECM fungi were exclusive to Tanegashima Island (sites 3 and 4).

Bioassay experiments
By the end of the growth period, three, fifteen, two and seven seedling mortalities occurred among the P. amamiana, P. parviflora, P. densiflora and C. sieboldii soil samples, respectively (S5 Appendix). ECM formation was observed in 100%, 81.3%, 75.3% and 80.6% of seedlings in P. amamiana, P. parviflora, P. densiflora and C. sieboldii, respectively. Five, four, nine and   (Table 3 and S3 and S5 Appendices). None of the control seedlings had ECM roots. In bioassays of the subgenus Strobus, six ECM fungal species were found, of which three were shared between P. amamiana and P. parviflora (Table 3, Fig 3). Nine and seven ECM fungal species were detected in the P. densiflora and C. sieboldii bioassays, respectively. The estimated species richness (Jackknife2) was 7.9, 7.9, 20.7 and 13.8 in P. amamiana, P. parviflora, P. densiflora and C. sieboldii, respectively.
Rhizopogon sp.1 was the most dominant ECM fungal species detected in the P. amamiana bioassay, found in 40 of 43 (93% frequency) soil samples (Fig 3). This fungus was also found in P. parviflora and P. densiflora samples but at lower frequencies, 67% and 31%, respectively ( Fig  3). Rhizopogon sp.1 was not found in any of the C. sieboldii bioassays. Cenococcum geophilum was the most common fungal species in the P. densiflora and C. sieboldii bioassays (Fig 3). Among the 16 ECM fungal species found in bioassays, only C. geophilum was shared among the four host species (Fig 3).

Discussion
We found Rhizopogon sp.1 in all remaining P. amamiana forests. Host identification of the resident ECM roots confirmed that Rhizopogon sp.1 was associated only with the endangered species P. amamiana. In soil spore banks, Rhizopgon sp.1 was far more dominant than in resident ECM roots and was identified in nearly all soil samples examined (93%) among the major remaining P. amamiana populations. These findings generally support both our hypotheses, confirming the existence of an ECM fungus specific to the endangered pine species in its natural settings and the predominance of this fungus in soil spore banks [12]. Rhizopogon sp.1 is very likely to have a close ecological relationship with the endangered P. amamiana, probably due to long co-evolutionary periods on isolated islands. Dominant ECM fungi in soil spore banks generally contribute to establishment of tree seedlings, especially for pine trees regenerating after a disturbance [46][47][48]. Thus, the Rhizopogon sp.1 found in this study likely also contributes to seedling establishment of the endangered P. amamiana after a disturbance [49,50]. Although no other native pine species are distributed in the remaining forests of P. amamiana, some broadleaf trees could compete with P. amamiana during regeneration. It should be noted here that the dominant broadleaf tree C. sieboldii was not compatible with Rhizopogon sp.1 in the bioassay experiments. Alternatively, C. sieboldii bioassay seedlings were dominated by the true generalist C. geophilum, sclerotia of which Ectomycorrhizal fungal communities in endangered Pinus amamiana forests exist in almost all forests investigated (e.g. [8,11,12,15]). Murata & Nara [51] suggested that C. geophilum competes with other ECM fungi, thus affecting the infection rates of other soil spore bank fungi. While ECM symbiosis is a prerequisite for trees to grow and survive in nature [3], controlling ECM fungal communities may be difficult, especially after forest development. Eliminating other ECM fungi from soil spore banks in disturbed sites via treatments, such as heating [12,[52][53][54] or filtering [51], may allow an increase in the relative frequency of Rhizopogon sp.1 and eventually increase regeneration of the endangered P. amamiana preferentially.
We found no sequences that match Rhizopogon sp.1 in an international nucleotide sequence database (DNA Data Bank of Japan [DDBJ]/EMBL/GenBank), potentially indicating its endemism to these islands. In fact, based on the morphological characteristics of the sporocarps and phylogenetic relationships with other described Rhizopogon species, we are now describing it as a new species, which does not belong to any subgenus proposed by Grubisha et al. [55] (Sugiyama et al. under revision). The existence of a Rhizopogon fungus specific to an endangered tree was documented in our recent study of Pseudotsuga species in Japan and Ectomycorrhizal fungal communities in endangered Pinus amamiana forests China; i.e., P. japonica is associated with R. togasawariana [12,32] and P. sinensis with an undescribed Rhizopogon species [56]. Here, we first confirmed the existence of a Rhizopogon fungus specifically associated with endangered Pinus, which is a far larger genus than Pseudotsuga and includes far more endangered species. Given the predominance of the host-specific fungus in soil spore banks and its potential roles in seedling regeneration, such fungi should be explored further in other endangered Pinus species that have been geographically isolated for long periods. Without identifying such key ECM fungi, conservation of endangered pine species will be difficult.
Rhizopogon sp.1 was associated solely with the endangered P. amamiana under natural settings. This fungus was also compatible with P. parviflora and P. densiflora in bioassays, but the colonization frequency was reduced significantly with increasing phylogenetic distance. P. amamiana is closely related to P. armandii var. armandii and P. armandii var. mastersiana [57,58], belonging to the same clade of subsection Strobus found in East Asian subtropical areas [29,57], as the divergence of P. amamiana from P. parviflora or P. densiflora dates approximately 6.4 Mya or 85 Mya, respectively [29]. The partial compatibility with these distantly related pine species may indicate that the isolation period has not been long enough for compatibility to be lost completely. In previous studies of R. togasawariana associated with Pseudotsuga japonica, colonization in P. densiflora was totally absent in a similar bioassay, but it was compatible with North American Pseudotsuga menziesii [12]. R. togasawariana belongs to the subgenus Villosuli, which is specific to Pseudotsuga, and its origin precedes the migration of the host to Asia approximately 34 Mya. Although Rhizopogon is regarded as a host-specific ECM fungal lineage [59], its compatibility is not strict within the same host genus. Yet, the compatibility with non-native hosts may not be fully functional in terms of nutrient transfer and host performance.  Another interesting finding of this study is that P. amamiana was not associated with any Suillus species, either in ECM roots of resident trees or in soil spore banks. Suillus produces epigeous sporocarps for spore dispersal by wind but is a phylogenetic sister to Rhizopogon [55]. While the life span of Suillus spores is shorter than that of Rhizopogon, it is frequently recorded in soil spore banks [20], probably because of its strong spore dispersal abilities [60,61] and good germination rates [62]. Some Suillus species are specific to, or prefer, the subgenus Strobus within the genus Pinus [63][64][65], and these fungi have wide intercontinental distributions. For example, Suillus spraguei is found in Strobus pine forests from North America to East Asia [63]. In Japan, S. spraguei is associated with P. parviflora, P. pumila and P. koraiensis [64,66], all of which belong to the subgenus Strobus. In addition, S. spraguei and some closely related species are associated with P. armandii, which is closely related to P. amamiana [63,   67]. The nearest pine population belonging to Strobus, P. parviflora, is located in the Takakumayama mountains on the main island of Kyushu, which is its southern population limit [68], and this population is located >100 km away from the remaining P. amamiana forests by sea. Therefore, the absence of Suillus species in the remaining P. amamiana forests may indicate the difficulty of airborne spore dispersal from distant Strobus forests together with local extinction on Yakushima and Tanegashima Islands.
In the bioassay experiment, it was difficult to characterize the effect of ECM infection on the initial growth of seedlings, because the experimental period was too short and the bioassay tubes were too small to evaluate the growth of P. amamiana. Therefore, we did not compare the growth of seedlings infected with ECM fungi with that of control plants in the bioassays. Instead, seedlings infected with Rhizopogon sp.1 and uninfected seedlings were raised for 1 year in a separate bioassay experiment, after which growth was compared. Seedlings infected with Rhizopogon sp.1 exhibited increased growth compared with control seedlings, with marginal statistical significance (S7 Appendix) and the difference would become much larger with the growth period. Moreover, P. amamiana seedlings colonized by Rhizopogon sp.1 are much more tolerant to transplantation than are uncolonized seedlings (S8 Appendix). Thus, even in a nursery setting, the application of Rhizopogon sp.1 is a promising approach to producing good seedlings for transplantation to conserved areas. S7 Appendix. In the bioassay experiment different from this study (unpublished data), the number of leaves and tree height of seedlings infected with Rhizopogon sp.1 and seedlings not infected with ECM fungi (control) grown for 1 year. Significant probability between Rhizopogon sp.1 and Control of leaf number and tree height by T test was p = 0.08 and p = 0.06, respectively. The bar shows the standard error. (PDF) S8 Appendix. From a bioassay experiment separate from this study (unpublished data), transplantation of seedlings infected with Rhizopogon sp.1 (upper) and uninfected seedlings (lower) to new pots after 1 year of growth. (PDF)