Community composition of arbuscular mycorrhizal fungi associated with native plants growing in a petroleum‐polluted soil of the Amazon region of Ecuador

Abstract Arbuscular mycorrhizal fungi (AMF) are worldwide distributed plant symbionts. However, their occurrence in hydrocarbon‐polluted environments is less investigated, although specific communities may be present with possible interest for remediation strategies. Here, we investigated the AMF community composition associated with the roots of diverse plant species naturally recolonizing a weathered crude oil pond in the Amazon region of Ecuador. Next generation 454 GS‐Junior sequencing of an 800 bp LSU rRNA gene PCR amplicon was used. PCR amplicons were affiliated to a maximum‐likelihood phylogenetic tree computed from 1.5 kb AMF reference sequences. A high throughput phylogenetic annotation approach, using an evolutionary placement algorithm (EPA) allowed the characterization of sequences to the species level. Fifteen species were detected. Acaulospora species were identified as dominant colonizers, with 73% of relative read abundance, Archaeospora (19.6%) and several genera from the Glomeraceae (Rhizophagus, Glomus macrocarpum‐like, Sclerocystis, Dominikia and Kamienskia) were also detected. Although, a diverse community belonging to Glomeraceae was revealed, they represented <10% of the relative abundance in the Pond. Seventy five % of the species could not be identified, suggesting possible new species associated with roots of plants under highly hydrocarbon‐polluted conditions.

able to degrade a variety of petroleum molecules (Sahoo, Ramesh, & Pakshirajan, 2012). Due to the fast-growing demand for hydrocarbon derivatives all over the world (Lee, 2015), it is expected that environmental pollution will increase in the coming years (Öztürk et al., 2015). It is thus of the highest priority either to apply physicochemical or to develop biological-friendly remediation strategies.
The identification of native microbial communities well-adapted to polluted conditions and their further isolation, mass production and application to polluted soils are part of this strategy and may represent an interesting approach for handling oil-contaminated sites.
The Amazonian region of Ecuador is a hotspot of biodiversity, which unfortunately is also a major reservoir of hydrocarbons (Ministerio del Ambiente de Ecuador, 2016). Therefore, the effects of petroleum pollutants on fauna and flora regularly reported in the literature (see above) are also significant here. Nevertheless, in hydrocarbon-polluted sites of the Charapa field in the Amazonian region, a natural recolonization of the abandoned weathered oil ponds was observed through the years (Garcés-Ruiz, Senés-Guerrero, Declerck, & Cranenbrouck, 2017). This suggested that plant roots and microbial communities associated with the rhizosphere were able to establish, probably enhancing the degradation of petroleum compounds, and thus representing a potentially important approach for the in situ treatment of hydrocarbon-polluted soils (Öztürk et al., 2015).
Among the rhizosphere microbial communities, one of the most important is the arbuscular mycorrhizal fungi (AMF). These obligate root symbionts contribute to the formation and stability of soil aggregates, and to the transport of nutrients and water to the plants (Smith & Read, 2008). Thus, phytoremediation assisted by AMF has been suggested for hydrocarbon-polluted environments (Lenoir, Lounes-Hadj Sahraoui, & Fontaine, 2016). The application of AMF may enhance plant growth and nutrient uptake. Several studies, in controlled conditions, reported an increased plant biomass, root and shoot length, P and N uptake and chlorophyll content (Liu & Dalpé, 2009;Tang, Chen, Huang, & Tian, 2009;Wu, Yu, Wu, Lin, & Wong, 2011). An increase in biodegradation activity of roots and rhizosphere microorganisms was also demonstrated, as well as an improved absorption and bioaccumulation of hydrocarbons by roots (see review Rajtor & Piotrowska-Seget, 2016). A recent study, conducted in a hydrocarbonpolluted soil from a natural environment in the Amazonian region of Ecuador (i.e., the Charapa field, Garcés-Ruiz et al. (2017)) recorded the presence of a relatively diverse community of AMF associated with various herbaceous plants. A high root colonization was noticed in all the plants sampled and a molecular diversity analysis, using a clone library and Sanger sequencing approach of a 1.5 kb fragment defined as the DNA barcode for AMF (Stockinger, Krüger, & Schüßler, 2010) allowed the identification of four AMF genera (i.e., Glomus, Rhizophagus, Acaulospora and Archaeospora) associated with three plant species (Euterpe precatoria, Carludovica palmata and Costus scaber) (Garcés-Ruiz et al., 2017). However, more than 74% of the species could not be ascribed to an identified AMF, suggesting the presence of numerous unidentified taxa.
The objective of this study was to explore in-deep the community composition of AMF associated with roots of a diverse assemblage of plants present in a weathered crude oil Pond, from the Charapa field (see Garcés-Ruiz et al., 2017), which could possibly be used to assist in phytoremediation efforts. High-throughput 454-pyrosequencing of an ~800 bp rDNA fragment was conducted and analyzed, using a reference "phylogenetic backbone" based on long AMF sequences (i.e., SSU-ITS-LSU 1.5 kb fragment) (Krüger, Krüger, Walker, Stockinger, & Schüßler, 2012;Stockinger et al., 2010) and an evolutionary placement algorithm (EPA), which allows the phylogenetic annotation of sequences to the species level (Senés-Guerrero & Schüßler, 2016a). The AMF community composition in plant roots was evaluated according to the site of collection within the Pond by comparing the relative read abundance (RA) of AMF species (Loján et al., 2017;Senés-Guerrero & Schüßler, 2016a).

| Sampling location and experimental design
The sampling was done on December 2013 in a weathered crude oil Pond of 450 m 2 in the Charapa field (76°48′54″ W, 00°11′46″ S) located in the province of Sucumbíos in the Amazonian region of Ecuador. More details and description of this site can be found in Garcés-Ruiz et al. (2017). The Pond has an irregular shape; the west, north, and south sides measured around ~23 m while the east side was only ~15 m. The perimeter of the Pond was marked every  Table 1). The number of plants sampled at each point varied from 1 to 3 according to their abundance and in a few cases, no plant was present.

| Root processing
After sampling, the shoots were used for plant identification and separated from roots that were kept within the soil at 4°C. Roots were cleaned from the soil particles with tap water. They were further washed for 10 min with Tween 80 (Panreac, Spain) diluted in sterilized distilled water (3%), to detach the crude oil (from Gordillo and Decock, personal communication). Roots were finally rinsed with sterilized distilled water. The cleaned roots were divided for evaluation of AMF colonization and DNA extraction.

| Physicochemical soil analysis
The Pond consisted of a layer of organic matter and soil above the weathered crude oil known as "tar". Soil was sampled below the or-

| AMF Root colonization
After cleaning, the roots were stained in acidic-blue ink (Walker, 2005) and the percentage of total colonization (%TC), arbuscular (%AC), and spores/vesicles (%VC) colonization were estimated under a dissecting microscope (Olympus BH2-RFCA, Japan) at 10× TA B L E 1 Plant species and number of individuals collected outside, inside and in the center of the pond Chemical and physical analysis from three independent samples and two composite samples: (1) a surface sample and (2) a sample collected at 30 cm depth, close to the center of the pond, (3) a composite surface sample and (4) a composite sample collected at 30 cm depth, made of soil cores collected in the 4 corners inside the pond, and (5) a sample at 30 cm depth collected 70 m outside the Pond (i.e., nonpolluted soil -as control).

Plant species 3 m outside 3 m inside Center
magnification following the method of McGonigle, Miller, Evans, Fairchild, and Swan (1990). An approximate of 100 intersections was observed per sample.

| DNA extraction
DNA was extracted from the 40 root samples according to Garcés-Ruiz et al. (2017). In brief, ~70 mg of dried roots from each sample were ground and material was transferred into the Lysing Matrix E tube from the FastDNA SPIN Kit for Soil (MP Biomedicals, USA).
DNA was extracted following the manufacturer's protocol. The DNA integrity was visualized in 1% electrophoresis gel, and 5 μl of the product were stained with 100 × GelRed ™ (Nucleic Acid Gel Stain, Biotium, Belgium). Samples were run at 100 V for 18 min in 0.5× TAE buffer and stored at −20°C until further use.

| PCR conditions and 454-pyrosequencing
Two PCRs were performed. The first PCR was developed according to Krüger, Stockinger, Krüger, and Schüßler (2009). The amplification of the partial SSU, the complete ITS region and partial LSU rRNA gene, using the SSUmAf-LSUmAr or SSUmCf-LSUmBr primers pairs was done. The primers targeted a 1.8 or 1.5 kb region, respectively. The nested PCR was performed as described by In the nested PCR, 1 μl of the first PCR product was used in the final reaction (20 μl

| Bioinformatic analyses
Analyses were performed according to Senés-Guerrero and Schüßler (2016a,b). In an initial step, sequences were quality-filtered and clustered at 98% to obtain one representative sequence (RS) per cluster.
The next step involved the phylogenetic placement by EPA of the RS into a reference phylogenetic tree ( Figure S1). The QIIME pipeline (Caporaso et al., 2010)  The sequences were deposited at NCBI with accession numbers MH503958 to MH504107.

| Statistical and data analysis
Data for AMF root colonization percentage were analyzed by one way ANOVA. Normal distribution was checked and nonnormal data were normalized by arcsine transformation before analysis. One way ANOVA was used to determine significant difference between sites AMF root colonization.
Statistical analyses were performed, using the IBM SPSS statistic 25 software.
Data for NMDS were square root normalized and analyzed, using The Shannon diversity index (H′) was calculated using the formula: where pi is the proportional abundance of AMF species according to the site of sampling.

| Physicochemical soil properties
The level of total petroleum hydrocarbon (TPH) in the Pond was >5000 mg Kg −1 (limit of quantification LOQ), while in the control (sample 5) it was 1188.8 mg Kg −1 ( Table 1). The pH was alkaline in the control and in the superficial sample from the middle of the pond (sample 1) ( Table 2). The pH was neutral in samples 2, 3, and 4 ( Table 2). The mineral nutrient content (P, N and K) was higher in the superficial samples (1 and 3), although the sampling was performed below the organic matter layer. Conversely, sample 5 (control) had a low amount of P compared with the other samples (Table 2) while N was similar or higher compared to the others as well as K ( Table 2).
The analysis of organic matter was higher than the LOQ in all the samples with the exception of the control soil (Table 2).

| AMF root colonization
The roots of all the plants sampled contained AMF structures.
Colonization percentages (i.e., total (%TC), arbuscular (%AC) and spores/vesicles (%VC) colonization) were analyzed according to the sampling place (outside, inside and in the center of the Pond). The %TC was 62.5% ± 4.3, 51.2% ± 5.1 and 43.8% ± 13.3, outside, inside and in the center of the Pond, respectively, without any significant difference (p = 0.138). The %VC outside the Pond was 4.7% ± 0.9 while it was higher inside and in the center of the Pond (i.e., 10.2% ± 1.9 and 11.5% ± 6.2, respectively). However, no significant difference was noticed (p = 0. 058). The %AC did not differ between the sampling places (p = 0.054). The values were low and varied from 1.05% ± 0.46 to 2.3% ± 1.2, outside and inside the Pond, respectively. No arbuscules were observed in plant roots from the center of the Pond.

| AMF community composition of roots
From a total of 40 root samples collected in the Pond (Table 1) for the EPA approach and the placement of the query sequences ( Figure S1 and Table S1), revealing 150 well-defined AMF representative sequences from 2557 reads (i.e., from approx. 800 bp).
The 150 AMF representative sequences were annotated as 15 species belonging to 7 genera (Acaulospora, Archaeospora, Rhizophagus, Glomus, Sclerocystis, Dominikia and Kamienskia) (Figure 2a,b). Some sequences could not be classified at the family or genus level. For instance, 5 OTUs were related either to Rhizophagus sp. or Dominikia sp. but could not be defined at the genus level and 1 OTU was in an undefined AMF branch from the reference phylogenetic tree (Figure 2a,b).

| D ISCUSS I ON
Arbuscular mycorrhizal fungi are obligate roots symbionts associated with an approximate of 72% of vascular plants (Brundrett & Tedersoo, 2018) and are occurring in almost every ecosystem (Cabello, 2001). However, their presence in hydrocarbon-polluted soils is poorly reported, though they may be of interest for remediation strategies (De la Providencia and was very similar to a previous study (Garcés-Ruiz et al., 2017).
High colonization was also reported by Huang, Tang, Niu, and Zhang (2007)  The study performed by Iffis, St-Arnaud, and Hijri (2016) also identified Acaulospora as a dominant genus in hydrocarbon-polluted soils. Our study as well as the previous one (Garcés-Ruiz et al., 2017) demonstrated thus the prevalence of this genus at different abundances. Acaulospora has been classified as stress-tolerant (Chagnon, Bradley, Maherali, & Klironomos, 2013). Indeed, it was frequently reported under harsh climatic conditions in acidic soils (i.e., pH 3.6-4.20 (Morton, 2017)) as well as under high elevation sites such as the Andes region 2500 m asl (Loján et al., 2017). In our study, this genus was observed in alkaline soils (pH 8) as well as at lower pH (5-6) at 300 m asl, in highly hydrocarbon-polluted condi- of shared identity (data not shown). Thus, we may hypothesize that the same unidentified species was detected in both studies.
Finally, the family Glomeraceae was represented by several species, although their relative abundance only represented 7% of the total AMF community. This could possibly be attributed to a lower affinity with the plant species sampled from the Pond or to an AMF species competition within the hydrocarbon-polluted soil.
The AMF species that were not detected by Sanger but by pyrosequencing were Glomus macrocarpum-like, Sclerocystis sp., Kamienskia sp., and Dominikia sp. The last two genera were only recently described (Błaszkowski, Chwat, Góralska, Ryszka, & Kovács, 2015). These two genera were identified by their spores, and it is thus suggested that their low occurrence may be attributed to a rare or seasonally sporulation or to their delicate spores easily decomposed by other organisms (Błaszkowski, Tadych, & Madej, 2000;Błaszkowski et al., 2015;Stutz & Morton, 1996).
Sclerocystis sp. and G. macrocarpum-like were found at low oc- to hydrocarbon pollutants could help in the selection of AMF species, production of inoculum, and its possible application for phytoremediation strategies. with all aspects of the work.

CO N FLI C T O F I NTE R E S T
We have no conflicts of interest to declare. Permits were given by the Ministry of environment of Ecuador and public enterprise PetroAmazonas EP for sampling and field study. The field study did not involve endangered or protected species.

DATA ACCE SS I B I LIT Y
The data used in this manuscript are available at request from