A preliminary examination of the bacterial, archaeal, and fungal rhizosphere microbiome in healthy and Phellinus noxius‐infected trees

Abstract Phellinus noxius is a pathogenic fungus that causes brown root rot disease, resulting in a widespread tree and crop mortality in the tropics and subtropics. Early stages of this disease are largely asymptomatic, hindering early diagnosis and effective treatment. We hypothesized that P. noxius infection would alter the rhizosphere microbiome of infected trees, based on which diagnostic biomarkers could be developed. Here, we examined for the first time the bacterial, archaeal, and fungal rhizosphere microbiome in four species of healthy and P. noxius‐infected trees (Ficus microcarpa, Celtis sinensis, Mallotus paniculatus, and Cinnamomum camphora) using high‐throughput amplicon sequencing. Results revealed the dominance of Proteobacteria and Actinobacteria in bacteria, Crenarchaeota and Euryarchaeota in archaea, and Ascomycota and Basidiomycota in fungi. Phellinus noxius infection did not affect the alpha diversity of the bacterial rhizosphere microbiome in all four tree species but affected that of archaea and fungi in a tree species‐dependent manner. Infection with P. noxius only affected the bacterial rhizosphere composition in M. paniculatus but not the other three tree species. By contrast, P. noxius infection affected the composition of the archaeal and fungal rhizosphere microbiome in all four tree species. Collectively, these results suggest that potential diagnostic biomarkers for brown root rot disease are tree species‐specific and should be developed based on different taxonomic groups. Our study has provided insights into the rhizosphere microbiome in healthy and P. noxius‐infected trees and laid a solid foundation for future comprehensive studies.

to the plant hosts by targeting their water-transport system, culminating in root mortality and compromising stability (Hodges & Tenorio, 1984). The life cycle of P. noxius is similar to other root-rotting basidiomycetes-a new infection starts from previously infected plants or colonized wood debris, from which the mycelium of P. noxius grows to infect the lateral and taproots of the host tree (Ann et al., 2002).
Phellinus noxius is difficult to eradicate due to its ability to survive on decayed root tissue in the soil for over 10 years (Chang, 1996). Early stages of brown root rot disease are largely asymptomatic, hindering early diagnosis and effective treatment and resulting in a high mortality rate of infected plants. Visible symptoms such as chlorosis and crown dieback can only be seen at the late stages of infections, in which the majority of roots have already been destroyed (Ann et al., 2002;Sahashi, Akisa, Ishihara, Abe, & Morita, 2007). To date, there is no standard curative measure for this disease and most research has been focused on its management, such as the use of biocontrol agents or fumigants (Chang & Chang, 1999;Gohet, Van Canh, Louanchi, & Despreaux, 1991;Prasad & Naik, 2002;Schwarze, Jauss, Spencer, Hallam, & Schubert, 2012). New methods for early diagnosis of the disease are needed.
The rhizosphere microbiome is important to plant health.
Plant exudates can alter the rhizosphere microbiome composition by recruiting specific microorganisms for defense against invasive pathogens (Gu et al., 2016;Pascale, Proietti, Pantelides, & Stringlis, 2020;Wei et al., 2018;Weston et al., 2012;Zhang et al., 2011). These beneficial microbes recruited, for example, Pseudomonas, Bacillus, and Trichoderma, can produce different elicitors and trigger induced systemic resistance (ISR) of the plant hosts via a complex network of defense-related hormone signaling pathways and thereby making them resistant against pathogenic threats (Pascale et al., 2020).
To date, the majority of rhizosphere microbiome studies have focused on healthy plants and mostly on the bacterial communities (e.g. Chaparro, Badri, & Vivanco, 2013;Chapelle, Mendes, Bakker, & Raaijmakers, 2016). Rhizosphere microbiome studies on archaea and fungi are limited. These three kingdoms interact with each other and play important roles in nutrient cycling and soil upkeep, and are therefore important to be studied together (Kirk et al., 2004). Until now, there were no studies of the rhizosphere microbiome concerning P. noxius infection.
In this study, we examined the bacterial, archaeal, and fungal rhizosphere microbiome in healthy and P. noxius-infected trees of four species commonly found in Hong Kong. Our aims were (a) to characterize and compare the diversity and composition of the rhizosphere microbiome in healthy and P. noxius-infected trees and (b) to examine whether the changes in the rhizosphere microbiome due to P. noxius infections are consistent across host tree species. We hypothesized that there would be clear differences in the rhizosphere microbiome between trees with different health status.

| Soil sampling
Rhizosphere soil samples of P. noxius-infected trees belonging to Ficus microcarpa (n = 3), Celtis sinensis (n = 1), Mallotus paniculatus (n = 1), and Cinnamomum camphora (n = 1) were collected around Hong Kong (22°18′N, 114°12′E) (Table 1). These tree species, especially F. microcarpa, are commonly found in Hong Kong and are vulnerable to P. noxius infection. Phellinus noxius-infected trees were initially identified by visual symptoms of chlorosis, crown dieback, the presence of basidiocarps, and the characteristic brown webbing throughout the roots after they were cut open. Rhizosphere samples were collected in triplicate around each tree, 5 cm below the soil surface to avoid surface contamination. Large roots were exposed carefully and soil attached to the roots was sampled with a small shovel.
For each sample, a small portion of roots was also taken back to the laboratory for confirmatory tests of P. noxius infection. Rhizosphere samples were also collected from one healthy tree for each of the four tree species from proximal areas for comparison purposes. In total, 30 rhizosphere soil samples were examined in this study. In the laboratory, ~2 g of soil subsamples from each sample were ovendried at 105 °C overnight and then used for total carbon and total nitrogen content analysis in triplicate on the vario MICRO cube elemental analyzer (Elementar, Langenselbold, Hesse, Germany).

| Phellinus noxius infection confirmation
Portions of tree roots were washed in distilled water, placed on 2% malt extract agar amended with gallic acid, streptomycin, benomyl, and dichloran (Chang, 1995), and incubated in the dark at 28°C.

| DNA extraction, PCR, and amplicon sequencing
Total DNA was extracted from the rhizosphere samples using Carlsbad, CA, USA) according to the manufacturer's instructions.
The procedure was slightly modified with the additional use of TissueLyser (Qiagen, Germantown, MD, USA) at 30 Hz for 1 min per side to improve cell lysis (Cheung, Wong, Chu, & Kwan, 2018).

| Sequence analysis
Raw sequence reads were demultiplexed, filtered for quality, and analyzed using QIIME 1.9.1 (Caporaso et al., 2010) as previously described (Cheung et al., 2015). Chimeric sequences were identified and removed using USEARCH 6.1 (Edgar, 2010) against the "Gold" reference dataset for bacteria and archaea, and against the UNITE dynamic ITS1 reference dataset (2016-01-01 release) (Kõljalg et al., 2013) for fungi. Sequence reads from the same kingdom were clustered into operational taxonomic units (OTUs) at 97% similarity using uclust with the open-reference OTU picking method.
Representative OTUs were aligned to the Greengenes reference dataset (13_8 release) (DeSantis et al., 2006) for bacteria and archaea, and the UNITE dynamic reference dataset (2016-11-20 release) for fungi, and taxonomically assigned using the RDP naïve Bayesian Classifier (Wang, Garrity, Tiedje, & Cole, 2007). Sequence reads of plant origin were removed from further analysis.

| Statistical analysis
Before diversity analyses, the bacterial, archaeal, and fungal sequence datasets were rarefied to 13,055, 13,821, and 16,386 reads, respectively. Alpha diversity was estimated with the Shannon index and the number of observed OTUs. Beta diversity was estimated using principal coordinate analysis (PCoA) with the unweighted UniFrac distance for bacteria and archaea and the Bray-Curtis dissimilarity for fungi. All the above statistical tests were conducted using scripts in QIIME. Linear regression analysis was performed using GraphPad Prism 8 (GraphPad Software Inc., San Diego, CA).
Differences were considered to be statistically significant when p < 0.05.

| Physicochemical parameters
All rhizosphere samples shared a similar total carbon and total nitrogen contents among tree species and between healthy and diseased samples of the same species (Table 1).

| Beta diversity
PCoA revealed that the rhizosphere microbiome structure of samples collected around the same tree were in general similar to each other

| Taxonomic composition in Phellinus noxiusinfected trees
There was no obvious difference in the taxonomic composition of major bacterial phyla, orders, or families between healthy and P. noxius-infected trees of C. sinensis, C. camphora, and F. microcarpa TA B L E 2 Alpha diversity of the bacterial, archaeal, and fungal rhizosphere microbiome in healthy and Phellinus noxius-infected trees

| DISCUSS ION
In this study, we have examined for the first time the effects of P. noxius on the structure and diversity of the bacterial, archaeal, and fungal rhizosphere microbiome in trees. Until now, only a few rhizosphere microbiome studies have simultaneously investigated these three taxonomic groups, and the majority of them have studied only the healthy rhizosphere of trees (Uroz et al., 2016;Veach et al., 2019). By contrast, studies involving diseased samples usually focus on agricultural crop plants (Filion, Hamelin, Bernier, & St-Arnaud, 2004;Han et al., 2017;Li, Ren, Jia, & Dong, 2014;Wei et al., 2018;Zhang et al., 2011), but none of them has examined the archaeal communities. Information on the rhizosphere microbiome across multiple kingdoms is essential due to their integral roles in nutrient cycling and the ability to maintain symbiotic and antagonistic relationships with the plant hosts. Our results have demonstrated that P. noxius can alter the rhizosphere microbiome of healthy trees but the effects depend on the species of trees. However, a major limitation of this study is the lack of true biological replicates for each tree species, which has hindered statistical testing on the differences observed.

All rhizosphere samples examined in this study were abundant in
Proteobacteria. This is not unexpected as members of this bacterial phylum are fast-growing (Fierer, Bradford, & Jackson, 2007). Similar Notably, members from the phylum Chloroflexi were more abundant in diseased samples of M. paniculatus (Figure 2a). This could represent a direct response to the invasion as Chloroflexi members are often found in disease suppressive soils and have been suggested to be part of the host plant defense system due to their ability to prevent iron uptake and root colonization by fungal plant pathogens (Lemanceau & Alabouvette, 1993;Liu et al., 2016;Rodriguez & Fraga, 1999). However, it is also possible that some Chloroflexi members are opportunistic pathogens that were enriched in the diseased Infection with P. noxius did not affect the bacterial diversity in the rhizosphere of all four tree species examined here. This differs from other studies on diseased microbiomes in crops (Han et al., 2017;Li et al., 2014;Shang et al., 2016;Wei et al., 2018), tree seedlings (Filion et al., 2004), and shrubs (Zhang et al., 2011). For example, wilted Lanzhou Lily has a higher rhizosphere bacterial diversity (Shang et al., 2016), whereas diseased black spruce (Picea mariana) seedlings (Filion et al., 2004) and cotton plants (Zhang et al., 2011) have a lower rhizosphere bacterial diversity. This suggests that, in contrast to the general belief that microbial diversity can act as a biomarker for plant health (Berg et al., 2017) (Li et al., 2014;Shang et al., 2016;Wei et al., 2018). However, no obvious differences are observed for most tree species in this study. The differences in responses could be explained by species-dependent microbial recruitment (Turner et al., 2013)

CO N FLI C T O F I NTE R E S T S
None declared.

E TH I C S S TATEM ENT
None required. F I G U R E A 2 Correlation between the relative abundance of the Phellinus genus and the bacterial families Hyphomicrobiaceae (a), Gaiellaceae (b), and Rhodospirillaceae (c), the archaeal families Nitrososphaeraceae (d), DHVEG-1 (e), and SAGMA-X (f), and the fungal genera Fusarium (g), Phoma (h), and Candida (i). Samples with a relative abundance of Phellinus >3% were regarded as outliers and excluded from the analysis