Vanilla aerial and terrestrial roots host rich communities of orchid mycorrhizal and ectomycorrhizal fungi

Vanilla planifolia is the source of the spice vanilla. This study is part of an interna-tional initiative to study the biology, including mycorrhizal fungi and cultivation practices of vanilla to improve its production in Mexico. The study focused on documenting mycorrhizal fungal diversity in vanilla. It also provided preliminary data on differences in mycorrhizal fungal communities associated with cultivation practices. A richer mycorrhizal community was observed in vanilla growing in a wild natural farm compared with those from a highly managed influence-associated fungal communities. We hypothesized that terrestrial roots of V. planifolia have a distinct and more diverse mycorrhizal fungal community compared to its aerial roots. We also hypothesized that V. planifolia of a highly managed farm have lower species richness of mycorrhizal fungi than a wild natural farm.


Social Impact Statement
Vanilla planifolia is the source of the spice vanilla. This study is part of an international initiative to study the biology, including mycorrhizal fungi and cultivation practices of vanilla to improve its production in Mexico. The study focused on documenting mycorrhizal fungal diversity in vanilla. It also provided preliminary data on differences in mycorrhizal fungal communities associated with cultivation practices. A richer mycorrhizal community was observed in vanilla growing in a wild natural farm compared with those from a highly managed farm. Our research provides insights for sustainable vanilla production that can benefit Mexican farming communities.

Summary
• Relatively little is known regarding differences in root symbionts (i.e., mycorrhizae) between epiphytic and terrestrial orchids. We characterized the mycorrhizal fungal communities of aerial and terrestrial roots of the orchid, Vanilla planifolia, from four Mexican farms representing different management systems.
• Amplicon sequencing identified 40 putative mycorrhizal fungi based on ITS sequence data, these included traditional orchid mycorrhizal fungi such as Ceratobasidium, and Thanatephorus in the order Cantharellales as well as Serendipitaceae in the order Sebacinales, and species of several genera traditionally considered as ectomycorrhizal fungi. Mycorrhizal fungal communities were similar in aerial and terrestrial roots, but differed in read abundances.
• Plants growing in wild-natural conditions hosted a richer, but not statistically different, community of mycorrhizal fungi in comparison with plants in the highly managed farm. Soil characteristics including texture, organic matter, N, P, and K do not explain differences between fungal communities at these farms.
• This is one of the first reports of a diverse community of fungi traditionally considered to form ectomycorrhizas in association with aerial orchid roots. Further research is needed to understand the functional role of these putative mycorrhizal fungi in the ecology of V. planifolia, and if ectomycorrhizal fungi commonly occur in other hemiepiphytic and epiphytic orchids.
[Correction added on 19 December 2020, after first online publication: Corresponding author email address has been modified.]

| INTRODUC TI ON
Orchids are the largest flowering plant family in the world with an estimated 27,800 species (Chase et al., 2015;Christenhusz & Byng, 2016). The family occurs worldwide, and the majority (69%) are adapted to grow as epiphytes (Zotz, 2013). Although pollinators are essential drivers of orchid diversity (Cozzolino & Widmer, 2005), researchers have argued that fungi are also involved in fostering orchid diversity and their distribution patterns (Otero & Flanagan, 2006). All wild orchids require orchid mycorrhizal fungi (OMF) to germinate and in some cases, continue to rely on fungi as adults (Gebauer et al., 2016;Rasmussen et al., 2015;Yoder et al., 2000). Understanding the relationship between orchid diversity and OMF has received some attention (Roche et al., 2010;Shefferson et al., 2007;Taylor et al., 2004;Waterman et al., 2011), but more information is needed to elucidate fundamental aspects of OMF diversity, including potential differences in OMF composition and abundance between epiphytic and terrestrial orchids.
Several differences between the root morphology and ecology of epiphytic versus terrestrial orchids are likely to influence the mycorrhizal fungal community. For instance, roots of epiphytic orchids have a hard velamen, an external, hygroscopic tissue layer around the cortex, whereas the roots of terrestrial orchids have a spongy velamen (Stern & Judd, 1999). Epiphytic orchids are adapted to microhabitats that are water-stressed, nutrient-poor, and have a high irradiance when compared to co-occurring terrestrial orchids (Benzing, 2008). Microscopy studies of epiphytic roots have shown OMF colonization where the root adheres to the surface of the host tree bark (Porras & Bayman, 2003;Smith & Read, 2010). While a recent study from Xing et al., (2019) showed modularity in epiphytic and terrestrial OMF communities, Martos et al., (2012) showed high nestedness of mycorrhizal fungi associating with epiphytic orchids compared to terrestrial orchids. Martos et al., (2012) hypothesized that epiphytic orchids adapted to stressful environments are predisposed to associate with mycorrhizal fungi that facilitate water and nutrient access compared to terrestrial orchids. Further studies to determine these differences have yet to be investigated.
The traditional reliance on culture-based methods has limited the characterization of fungal communities of orchid roots because many fungi are either unculturable or very slow-growing (Allen et al., 2003;Arnold et al., 2007). This includes the vast majority of ECM fungi. The use of environmental sequencing (i.e., DNA isolated and sequenced directly from an environmental sample) is an improved alternative to culture-based methods for determining mycorrhizal fungi (Edwards et al., 2015;Lundberg et al., 2012;Manter et al., 2010). Recent studies of epiphytic and terrestrial orchids in temperate and tropical habitats have revealed the usefulness of environmental sequencing, specifically amplicon sequencing, to characterize OMF (Cevallos et al., 2017;Herrera et al., 2019;Oja et al., 2015).
This study was part of a broader interdisciplinary effort to understand the biotic and abiotic factors that influence the pro- We examined the diversity of mycorrhizal fungi associated with V. planifolia aerial and terrestrial roots. As the four farms available to us differed in cultivation practices, our study also served as a pilot project for investigating how cultivation practices in Mexico K E Y W O R D S ceratobasidiaceae, epiphytic roots, orchid mycorrhizal fungi, root endophytes, tropical orchids influence-associated fungal communities. We hypothesized that terrestrial roots of V. planifolia have a distinct and more diverse mycorrhizal fungal community compared to its aerial roots. We also hypothesized that V. planifolia of a highly managed farm have lower species richness of mycorrhizal fungi than a wild natural farm.

| Sampling sites and sample collections
Aerial and terrestrial roots were randomly sampled from four V. planifolia farms in Mexico during March and April 2014 (during anthesis) ( Table 1) and (c) highly managed (see Figure S1 and Table 1 One aerial and one terrestrial root each from five healthy V. planifolia plants at the four different farms were collected for a total of 40 root samples. Only aerial roots that adhered to tutors were collected (Table 1). The length of each root sample was approximately 5 mm. Each aerial and terrestrial root sample was immediately stored in cetyltrimethylammonium bromide (CTAB) buffer and then placed on ice (U' Ren et al., 2014). Roots were kept at 4°C until surface sterilization and DNA extraction. Mycorrhizal colonization was confirmed by observing the roots under a microscope.
The thermal cycler conditions were as follows: 2 min at 94°C followed by 32 cycles at 94°C for 45 s, then annealing at 59°C for 45 s, an extension for 1 min at 72°C, and a final extension at 72°C for 5 min. These PCR products were then cleaned with a concentration of 0.8x AMPure XP beads (Beckman Coulter).
Next, for the second round of PCRs, primers ITS86F-adpt and ITS4-adpt (see Table S1) were used in 25 µl reactions to amplify the PCR products generated from the first-round of PCR. This second PCR step used the same reagents and quantities as the first PCR; however, this PCR step was reduced to 25 cycles. Gel electrophoresis was used to visualize successful PCR products. These PCR were cleaned with a concentration of 0.8× AMPure XP beads.

| Data processing and statistical analyses
Paired-end reads were processed using the Pipits pipeline (version 2.2) (Gweon et al., 2015). First, forward and reverse sequences were joined (VSEARCH) (Rognes et al., 2016), and quality filtered with FASTQ_QUALITY_FILTER (FASTX-toolkit, http://hanno nlab. cshl.edu) (Gordon & Hannon, 2010) using the default settings of Pipits. During this quality filtering step global singletons were removed. Fungal ITS 2 reads were then detected and retained using ITSx (Bengtsson-Palme et al., 2013) that used HMMER3 (Mistry et al., 2013), whereas chimeras were removed with VSEARCH (Rognes et al., 2016). Lastly, reads were clustered at 95% sequence similarity into Operational Taxonomic Units (OTUs) and, taxonomy was assigned with the RDP classifier (Wang et al., 2007) that relied on the UNITE fungal database (Abarenkov et al., 2010;Nilsson et al., 2019). The OTU clustering of 95% sequence similarity is the recommended OTU clustering for reads generated with primers ITS86F and ITS4 (Waud et al., 2014). The default setting for a confidence threshold of 85% for RDP classifier was used in Pipits to assign taxonomy. Finally, to assess the functions of OTUs, OTUs were assigned to guilds using the program FUNGuild v1.0 (Nguyen et al., 2016). FUNGuild assignments did not characterize all putative OMF assignments, therefore putative OMF was determined based on the OMF taxa reported in Dearnaley et al., (2012).
Rarefaction curves were produced using iNext (Hsieh et al., 2016) an R package (R Development Core Team, 2012) that interpolated and extrapolated the data. After generating rarefaction curves, OTUs that had < 1,000 sequences and were not present in two or more samples were discarded from the overall analyses. Samples were then normalized with Cumulative Sum Scaling from the R package metagenomeSeq (Paulson et al., 2013).
Principal coordinate analysis (PCoA) with Bray Curtis distances were used to investigate the differences between fungal communities. In addition, PCoA was used to infer differences with a subset of the data that included only putative OMF sequences, that is, OMF and ECM fungal sequences. The significance of β diversity metrics was analyzed by permuting the raw data (10,000 permutations) using the function adonis in the R package vegan (the adonis function is analogous to Anderson's "permutational MANOVA") (Anderson, 2001).

| RE SULTS
FUNGuild is not effective at assigning trophic modes of genera that have multiple trophic modes (e.g., Ceratobasidium spp. are assigned as pathogens in FUNGuild but are well-documented symbionts in orchids The PCoA ordination of total fungal communities revealed distinct clusters for aerial and terrestrial roots in the ordination space ( Figure 2). These differences were further highlighted with a PERMANOVA revealing that differences were significant between aerial and terrestrial root fungal communities (F 1, 35 = 3.65, r 2 = 0.09, p < .05). Furthermore, rarefaction curves revealed that aerial roots had a greater OTU richness in comparison to terrestrial roots ( Figure S2, and also see Figure S3).
Most mycorrhizal fungi of aerial roots (86%) were present in the terrestrial roots ( Figure S4a) and, unlike the situation observed for the total fungal community, putative mycorrhizal fungi were more abundant in terrestrial roots compared to aerial roots ( Figure 1).
Mycorrhizal fungi of aerial roots comprised 2% of the total reads, whereas the terrestrial root community made up 19% of the total reads. Moreover, we detected lower read abundances for mycorrhizal fungi in most aerial root samples (Figure 1).
Ceratobasidiaceae were the dominant mycorrhizal fungi in V.
planifolia roots (both aerial and terrestrial) (Figures 1 and 3). We and Scleroderma OTUs between the two traditional farms (the intermediate cultivation practices) were also observed ( Figure S6).

Lastly, the dominant OMF, Ceratobasidiaceae OTU 88 and
Ceratobasidiaceae OTU 93, were present among all sites but were most abundant in the terrestrial roots of the two traditional farms ( Figure S6).
Soil characteristics of the highly managed and wild natural farms were similar to each other but differed from both traditional farms (Table S2). PCA results showed that pH, EC, organic matter, and Fe were significant for distinguishing farms ( Figure S7). Potential drivers

| D ISCUSS I ON
Using amplicon sequencing, we made in-depth comparisons between the fungal communities of the morphologically distinct aerial and terrestrial roots of V. planifolia. Aerial roots had greater total fungal OTU richness with 692 OTUs compared to 342 OTUs in terrestrial roots ( Figure S4a,c). However, we did not detect differences in mycorrhizal fungal communities (OMF and ECM fungi), as aerial and terrestrial roots shared 86% of their OTUs ( Figure S4b).
We detected most of the major traditional OMF genera, that is, Ceratobasidium, Thanatephorus, and Sebacina. However, we did not detect Tulasnella, which was reported from V. planifolia by Porras-Alfaro and Bayman, (2007) and many other orchids (Currah et al., 1997;Suárez et al., 2006;Zettler et al., 2017). Although the primers we employed were optimized to detect OMF including Tulasnella OTUs (Waud et al., 2014), Tedersoo et al. (2015) noted Tulasnella sequences having mismatches with these primers, and this could potentially limit their efficiency to recover Tulasnella species.
A diversity of fungal guilds in addition to potential mycorrhizal fungi were detected in V. planifolia roots. For instance, we repeatedly observed a Fusarium OTU within roots of healthy V. planifolia  Koyyappurath et al., 2016). However, a recent study has shown that Fusarium oxysporum may function as an OMF when associating with the terrestrial orchid Bletilla striata (Jiang et al., 2019). We did not resolve the species identity of the Fusarium OTUs recovered in our study, and its ecological guild remains unknown. The other principal fungal pathogen of V. planifolia, Colletotrichum, (Havkin-Frenkel & Belanger, 2018) was rarely detected in our study with < 1% of the total reads.
Similar mycorrhizal fungal communities in aerial and terrestrial roots suggest that root morphology and physiology does not constrain the fungal symbiotic diversity for either root type in V. planifolia.
Putative mycorrhizal fungi in aerial roots attached to organic tutors likely encounter more favorable conditions for growth than those at the highly managed farm (Pantepec) where the aerial roots were attached to concrete tutors and exposed to more stressful conditions.
The bark and wood of tutors at the semi natural and traditional sites may serve as sources of nutrients which are lacking at the highly managed farm (Pantepec). Fungi in aerial roots of V. planifolia may have different ecologies/play different roles depending on their microhabitat.
While we did not detect a difference in richness of mycorrhizal fungi between root types, we did identify differences in read abundance ( Figure S6) which has been used as a proxy for species  roots compared to terrestrial roots is consistent with the findings of Porras-Alfaro and Bayman (2007) who observed fewer pelotons in aerial roots than terrestrial roots of V. planifolia. Other microscopy studies also reported aerial roots sampled from epiphytic orchids being colonized at lower rates than terrestrial roots and that colonization was restricted to the root surface adjacent to their substrata (i.e., host tree bark) (Bermudes & Benzing, 1989;Hadley & Williamson, 1972;Lesica & Antibus, 1990;Porras & Bayman, 2003). Lower colonization may be due to structural barriers (e.g., passage cells) in aerial roots limiting mycorrhizal fungal colonization as proposed by Chomicki et al., (2014). The lower nutrient availability in the aerial substrate may also influence lower fungal colonization.
We observed greater diversity of fungi traditionally viewed as forming ECM (Nguyen et al., 2016) than previously reported in aerial roots. ECM fungi are rarely obtained in culture which may explain the lack of detection in traditionally culture-based studies of OMF communities (Johnston et al., 2017). The occurrence of a diverse ECM fungal community in both aerial and terrestrial roots of V. planifolia is suggestive that these fungi may be common in other photosynthetic orchids growing in tropical habitats, but little data are available. Orchids may be predisposed to forming associations with ECM fungi. For instance, mycoheterotrophic orchids are well-known examples of orchids that associate with ECM fungi (McKendrick et al., 2000;Taylor & Bruns, 1997) and traditional OMF species such as Ceratobasidiaceae and Tulasnella species can function as ECM fungi supplying nutrients to photosynthetic trees in exchange for photosynthates (Bougoure et al., 2009;Warcup, 1985;Yagame et al., 2008Yagame et al., , 2012. Although we are uncertain of the functional roles of these fungi in V. planifolia, research by González-Chávez et al. (2018) documented pelotons formed by the ECM fungus Scleroderma in terrestrial roots of V. planifolia. As pelotons are the sites of nutrient exchange in orchid mycorrhizae (Kuga et al., 2014), it is hypothesized that this Scleroderma was functioning as an orchid mycorrhiza.
The source of ECM fungi in V. planifolia aerial roots is unclear.  (2007) and González-Chávez et al., (2018). Gebauer et al. (2016) proposed that most orchids, including putative fully autotrophic orchids, are likely partial mycoheterotrophs in adult stages. Tropical orchids may benefit from forming associations with a more diverse community of ECM fungi than previously reported due to low light conditions and limited nutrient access in some substrata. Investigating the complete root mycobiota is needed for understanding the diversity and potential role fungi play in driving orchid diversity and adaptation.
This study also provided preliminary insights on possible influences of different cultivation practices of V. planifolia on mycorrhizal fungi. We observed lower OTU richness of mycorrhizal fungi at the highly managed farm (growing on concrete tutors) than at the wild natural farm (growing on living trees), and differences in fungal taxa among each of the farms. Soil characteristics at the highly managed and the wild natural farms were similar, suggesting that differences in mycorrhizal communities were not associated with differences in soil characteristics. Other aspects of vanilla farming are likely driving the differences in observed mycorrhizal communities. Studies have documented positive impacts of diverse mycorrhizal communities on plant fitness (Smith & Read, 2010). Our study provides support for the need for additional studies on the impact of cultivation practices on the mycobiome of vanilla.

ACK N OWLED G M ENTS
Thr financial support for this project was provided by funding