Fungal diversity driven by bark features affects phorophyte preference in epiphytic orchids from southern China

Study site and species

The sub-tropical forest analysed in this study is located in China, Yunnan, Xishuangbanna, Mengla county, Village Quingyanzhai (#94) N 21.802068, E 101.380214, geodetic datum WGS84 (Fig. 6). The site is characterized by a rocky outcrop rising 30–50 m over surrounding rubber plantations, harbouring about 20 ha of relict dry tropical forest. The outcrop sides are steep and mainly covered with bamboo. The top area is colonized by shrubs and 10–15 m high trees (a few trees on the slopes are much higher). The most conspicuous species is Quercus yiwuensis Y.C. Hsu & H.W. Jen. In March 2017 we selected four individual trees of Q. yiwuensis, and an equal number of Pistacia weinmannifolia Franch, and Beilschmiedia percoriacea C.K. Allen that were also numerous on the site. Plants were identified by the authors in the field and labelled. Botanical specimens were deposited in the School of Pharmaceutical Science and Technology, Tianjin University, Tianjin, China.

Q. yiwuensis was designated P-tree (P1, P2, P3, P4) because we consistently found the orchid Panisea uniflora Lindl. growing on this phorophyte species. P. weinmannifolia was designated B-tree (B1, B2, B3, B4) because it harboured the orchid species Bulbophyllum odoratissimum (Sw) Lindl. (Supplementary Fig. S2 a-d). On P. weinmannifolia trees no P. uniflora was observed, while B. odoratissimum was never found on Q. yiwuensis trees. Both tree species were richly colonized by several other orchid species. Beilschmiedia percoriacea trees were designated neutral tree, N-tree (N1, N2, N3, N4), because neither of the two target orchid species grew on them. The latter tree species carried several lichens and a single fern species (Lepisorus sp.), but only in one instance was observed to carry an orchid epiphyte (Coelogyne sp.).

GPS positions of investigated trees were obtained less than 1 m from the trunk (Fig. 6). Accuracy is about 3 m. Accuracy in altitude readings is about 100 m. Distance between degrees of latitude is 111 km. At N 21.78978 the distance between degrees of longitude is 103 km, which means that the last digit in the 5-digit decimal degrees corresponds to 1.11 m in latitude and 1.03 m in longitude.

The trees were labelled with different colours as follows:

• P-trees (carrying P. uniflora and other epiphytes, but not B. odoratissimum), identified as Q. yiwuensis, with red labels (P1 N 21.79880, E 101.37909, 1073; P2 N 21.79882, E 101.37923, 1072; P3 N 21.79878, E 101.37947, 1074; P4 N 21.79878, E 101.37904, 1073).

• B-trees (carrying B. odoratissimum and other epiphytes, but not P. uniflora), identified as P. weinmannifolia, with blue labels (B1 N 21.79878, E 101.37950, 1074; B2 N 21.79880, E 101.37938, 1078; B3 N 21.79881, E 101.37931, 1083; B4 N 21.79884, E 101.37923, 1076).

• N-trees (carrying epiphytes, but neither B. odoratissimum nor P. uniflora), identified as B. percoriacea, with yellow labels (N1 N 21.79873, E 101.37905, 1064; N2 N 21.79868, E 101.37908, 1072; N3 N 21.79883, E 101.37893, 1071; N4 N 21.79879, E 101.37895, 1071).

• The point of access to the outcrop top area was located at the Western edge (N 21.79880, E 101.37827, 1058, Fig. 6).

Sampling

For each of the twelve selected trees, breast height circumference (BH = 130 cm above ground) was measured. Approximate total height was determined by Nikon Laser Forestry Pro or estimated if sighting lines were interfered by other vegetation.

The lowermost individual of the target orchid species was recorded in relation to BH. Bark samples were collected, and bark features recorded at BH, by target orchid, and 50 cm above target orchid or BH, whichever was highest point. In N-trees, where there were no target orchids, sampling was thus at BH, BH + 50 cm, and BH + 100 cm.

Sampling on each tree involved approximately 12 cm² bark cut out with a sterile knife and rubber gloves to prevent cross-contamination, for pH-analysis, metabarcoding, fungal isolation and chemical analysis. Besides, 3 bark cores were taken by trephor sampler (16 mm, 2 mm diam., Costruzioni Meccaniche Carabin Carlo) for water holding measurement.

Roots of target orchids were sampled, from three adult individual plants on each P- and B-tree. No permissions were necessary to collect plant samples, using a protocol that avoided plant damages. All plants were left in the exact location where they were found in the sampling site, after collecting the small portions of bark and root material for the study. All experiments including the collection of plant material in this study are in compliance with relevant institutional, national, and international guidelines and legislation.

All fresh material collected from the sampling site was first kept in cool boxes, brought to the laboratory, and processed within three days.

Fungal isolation from bark

For each sample, half of the bark material and orchid roots were kept at − 80 °C for subsequent metabarcoding analysis. The rest of bark (about 2 g for each sample) was immediately processed for fungal isolation. The large bark portions were ground into powder using a sterile mortar and pestle; 5 ml were reserved for pH measurement, while the rest was suspended in a final volume 50 ml sterile water solution in a sterile centrifuge tube. The tube was shaken with Vortex vibration meter thoroughly and solution aliquots were spread homogenously onto isolation medium plates. For each bark sample, aliquots of 500, 300, 200, and 100 μl, were spread per triplicate to one plate each of PDA (Potato Dextrose Agar) medium, containing ampicillin (50 μg/mL) and streptomycin (50 μg/ml) to inhibit bacterial growth^49,50. A diluted solution was also made by mixing 1 ml of the original solution with 9 ml sterile water and plated. Petri dishes were incubated at room temperature (23–25 °C) in the dark for up to 2 months to allow the development of slow-growing mycelia. Fast growing fungal strains started to grow after about two days. Colonies showing different morphology and appearance were transferred to fresh plates to obtain pure cultures. In the following days, other slower growing mycelia were available in the Petri dishes and were also regularly picked up and isolated onto new PDA plates every 2 days. All isolated fungal strains were stored at 4 °C for further analysis. All strains were deposited in the LP Culture Collection (personal culture collection held in the laboratory of Prof. Lorenzo Pecoraro), at the School of Pharmaceutical Science and Technology, Tianjin University, Tianjin, China.

Molecular and morphological analysis of bark culturable fungi

The identification of isolated fungal colonies was performed using DNA sequencing combined with microscopy. Total genomic DNA from isolated fungi was extracted following the cetyltrimethyl ammonium bromide (CTAB) method modified from Doyle and Doyle⁵¹. Fungal ITS regions were PCR-amplified using the primer pair ITS1F/ITS4⁵² following the procedure described in Pecoraro et al.³⁷ for PCR reaction, thermal cycling, and purification of PCR products. Controls with no DNA were included in every amplification experiment in order to test for the presence of laboratory contamination from reagents and reaction buffers. Purified DNA amplicons were sequenced with the same primer pair used for amplification. DNA sequencing was performed at the GENEWIZ Company, Tianjin, China.

Sequences were edited to remove vector sequences and to ensure correct orientation and assembled using Sequencher 4.1 for MacOsX (Genes Codes, Ann Arbor, MI). Sequence analysis was conducted with BLAST searches against the National Center for Biotechnology Information (NCBI) sequence database (GenBank; http://www.ncbi.nlm.nih. gov/BLAST/index.html) to determine the closest sequence matches that enabled taxonomic identification. DNA sequences were deposited in GenBank (Accession Nos. MW603206 – MW603451). Fungal morphological characters (hyphae, pseudohyphae, conidiophores, conidia, poroconidia, arthroconidia, etc.) were examined using a Nikon ECLIPSE Ci microscope for the identification of isolates following the standard taxonomic keys^{53,54,55,56,57}.

Assessment of bark and orchid associated fungal community using Illumina sequencing

Bark and orchid root samples were pulverized in a sterile mortar, and genomic DNA was extracted using the FastDNA® Spin Kit as described by the manufacturer (MP Biomedicals, Solon, OH, USA)^58,59. In total, this resulted in 60 DNA samples, including 36 from bark (3 sampling points for each tree × 12 trees) and 24 from orchid roots (3 orchid individuals sampled on each P- and B-tree × 8 trees; the 4 individual N-trees were not used for orchid sampling because they did not carry the study orchid species). Subsequently, amplicon libraries were created using two primer combinations targeting the internal transcribed spacer 2 (ITS-2): ITS7F and ITS4R⁶⁰ was used as universal fungal primer pair to target nearly all fungal species, while ITS3⁶¹ and ITS4OF⁶² was used to more specifically target orchid mycorrhizal fungi. Previous research has shown that most universal fungal primers have multiple mismatches to many species of the orchid-associating basidiomycetes, in particular in Tulasnellaceae family^46,58,63. Since the goal of the present work was to analyse the total fungal community in the orchid-phorophyte environment (bark and orchid roots), as well as more specifically detect the orchid mycorrhizal fungi in the studied samples, it was necessary to combine two different primer pairs to characterise the whole investigated fungal diversity^47,64,65,66. Polymerase chain reaction (PCR) amplification was performed in 50 μl reaction volume, containing 38 μl steril distilled water, 5 μl 10 × buffer (100 mM Tris–HCl pH 8.3, 500 mM KCl, 11 mM MgCl₂, 0.1% gelatin), 1 μl of dNTP mixture of 10 mM concentration, 0.25 μM of each primer, 1.5 U of RED TaqTM DNA polymerase (Sigma) and approximately 10 μg of extracted genomic DNA. PCR conditions were as follows: 1 cycle of 95 °C for 5 min initial denaturation before thermocycling, 30 cycles of 94 °C for 40 s denaturation, 45 s annealing at various temperatures following Taylor and McCormick⁶², 72 °C for 40 s elongation, followed by 1 cycle of 72 °C for 7 min extension. To minimize PCR bias, three PCRs were pooled for each sample. The resulting PCR products were electrophoresed in 1% agarose gel with ethidium bromide and purified with the QIAEX II Gel Extraction Kit (QIAGEN). Amplicon libraries were generated using the NEB Next Ultra DNA Library Prep Kit for Illumina (New England Biolabs, USA) following the manufacturer’s instructions to add index codes. Samples were sequenced using the Illumina MiSeq PE 250 sequencing platform (Illumina Inc., San Diego, CA) at Shanghai Majorbio Bio‐Pharm Technology Co., Ltd. (Shanghai, China).

Bioinformatics of fungal sequences

Sequences originated from the total (ITS7F and ITS4R primers) and orchid-associated (ITS3 and ITS4OF primers) fungi datasets were processed separately. Raw reads were merged with a minimum overlap of 30 nucleotides, and the primer sequences were trimmed off. Subsequently, reads were filtered by discarding all sequences with expected error > 1. The quality-filtered reads were denoised using the UNOISE3 algorithm⁶⁷ to create zero-radius operational taxonomic units (zOTUs), with chimera removal. All the steps were performed using USEARCH v.11⁶⁸. Raw sequences have been deposited in the Sequences Read Archive (SRA) of NCBI as BioProject ID PRJNA702612. The fungal zOTUs were assigned to taxonomic groups using the Blast algorithm by querying against the UNITE + INSD fungal ITS database (version 7.2, released on 10 October 2017)⁶⁹ using the sintax algorithm with 0.8 cutoff⁷⁰. The zOTUs originated with the orchid-associated fungal primers were manually screened for possible orchid-associated mycorrhizal families based on the information provided in Table 12.1 in Dearnaley et al.⁷¹, and only these were retained for further analysis in this dataset.

To attempt removing spurious counts due to cross-talk (assignment of reads to a wrong sample) we removed all the zOTUs represented by less than 0.02% of reads in each sample, which is more conservative than previous error estimates⁷². The datasets were rarefied to the minimum sequencing depth (23,419 for total fungi and 13,074 for orchid-associated fungi), zOTUs present in less than three samples and low abundant zOTUs (with relative abundance < 0.1%) were removed from further analyses. The filtering steps resulted in removing two and six samples from the total and orchid-associated fungi datasets, respectively.

Fungal communities

We tested for differences in richness between orchid and bark samples (host tree), and for the interaction between host tree and orchid species using ANOVA. We performed pairwise comparisons using t-test with Bonferroni corrections to assess differences in richness between the two orchid species (B. odoratissumum and P. uniflora) and the barks from the three tree species (P. weinmannifolia, Q. yiwuensis and B. percoriacea).

We tested for the effect of orchid species and host tree on the total and orchid-associated fungal community structure with two-way PERMANOVA analyses with 999 permutations using the function adonis2 in vegan R package⁷³ on the Bray–Curtis dissimilarity matrix calculated on the Hellinger transformed count data. Pairwise comparisons were performed using the RVAideMemoire R package with Benjamini–Hochberg corrections⁷⁴. Variance explained by orchid species and host tree was investigated with variation partitioning using the vegan R package. We also investigated whether fungal communities varied according to the height on the tree where they were collected. In addition, to understand whether each tree had specific fungal communities on the bark, we tested for the effect of tree species, individual tree and height in tree on the total and orchid-associated fungal communities only between bark samples using one-way PERMANOVAs.

To test whether the geographic location of the sampled trees had an influence on the fungal community composition of the orchids and bark, we computed Mantel test correlation tests between the distance among sampled trees, calculated with the geosphere R package⁷⁵ on the geographic coordinates, and the Bray–Curtis dissimilarity matrix of fungal communities. Because we collected samples at three heights in each tree, we computed a separated Mantel test for each height.

Bark chemical analysis

In the laboratory, small chips were taken from selected bark samples for pilot chemical analysis, comprising one pooled sample from each tree species (P-, B-, N-tree). They were carefully cleaned from every lichen and moss residual, then dried overnight at 50 °C and ground in a mortar. Bark content was subsequently extracted in methanol (50 ml); the ground material was soaked, vortexed and sonicated for 30 min in an ultrasonic bath. The mixture was centrifuged at 10,000 rpm for 15 min. The supernatants were collected and filtered. The extraction was repeated three times on the precipitate. The three supernatants for each bark sample were finally merged in a single tube and vacuum-dried to afford the crude extracts. The extract was dissolved in 2 mL methanol, and centrifuged. Analytical HPLC was conducted with a Waters system (Waters e2695-2998, Photodiode Array Detector, Milford, MA, USA) using a C18 column (Hypersil GOLD C18 column; 5 μm; 4.6 × 250 mm, with a flow rate of 1.0 ml/min; from 20% MeOH in H2O with 0.1% HCOOH to 100% MeOH for 30 min, followed 100% MeOH for 10 min; 1.0 ml/min; The UV detection wavelengths were set at 230 and 254 nm, respectively).

Bark pH

The ground bark (particle size < 2 mm) was weighed into portions of 150, 100, 50 mg, or less, as availability allowed. They were subsequently immersed in milliq water, amounting to 0.04 ml per mg bark, shaken 60 rounds/min for 60 min, kept stationary to settle for 60 min and measured with an Unisense pH microelectrode system.

Water holding capacity of bark

One or two cores from each tree and height with intact inner and outer bark were, after removal of any xylem parts, selected for water holding capacity measurement. Immediate weight, wet weight after soaking in dd water for 24 h, and dry weight determined after desiccation at 50 °C in oven overnight. While in soaked condition, the core lengths were measured in dissection microscope at 12 × magnification to obtain a measure of bark thickness and core volume. Water holding capacity was calculated as (WW-DW)*100/WW.

Assessment of bark roughness

Bark roughness of each analysed tree was recorded at the three bark sampling points described above, by determining a vertical bark surface profile, using a Yamano 150 mm contour gauge (Yamano Seisakusho, Japan) held parallel to the length axis of the tree trunk, the shape of which was subsequently photographed. Simple surface roughness parameters usable at all scales were measured for numerical assessment: Contour length, ratio contour length/evaluation length, maximum height of profile (Rt) and maximum heights (Rti) within 5 sampling intervals; average maximum height of the profile Rz(Din) was calculated (Supplementary Figs. S3 and S4) (DIN 4768 1990 Deutsches Institut fur Normung E.V.)⁷⁶. Contour length was measured with the image processing program ImageJ 1.50i (2016).

Source: Ecology - nature.com