Effects of Mikania Sesquiterpene Lactones on Soil Microbes


 Background and aims Allelopathy is frequently invoked as being important for successful invasion by non-native plants. Yet, the effects of specific phytochemicals of invasive plants on soil microbes remain unexplored. Methods Here we used manipulative experiments and next generation sequencing (NGS) approaches to investigate how the sesquiterpene lactones (STLs) of invasive Mikania micrantha influence soil microbial communities and nutrients.Results We found Mikania STLs to significantly increase the regulation of soil microbial activity (i.e. increased CO2 concentrations). Using the specific STL, dihydromikanolide, we found available soil nutrients to increase in the presence of this phytochemical and that bacterial richness increased while fungal richness decreased. The presence of dihydromikanolide also increased the abundance of beneficial soil bacteria and fungi associated with nutrient cycling and supply, while simultaneously lowering pathogen abundance. Clustering analysis found bacterial functional groups, such as those involved in carbon, nitrogen, phosphorus, and sulfur metabolism, to be similar in experimentally-treated dihydromikanolide soils and Mikania-invaded soils collected from the field, but significantly higher than those in uninvaded soils. This suggests that M. micrantha can enhance certain bacterial functional groups via its phytochemicals. Soil fungi, on the other hand, appeared to be less sensitive to dihydromikanolide than bacteria. Conclusions We conclude Mikania STLs, and in particular dihydromikanolide, may be key factors in determining soil microbial structure and function and may contribute to the invasion success of the species. Our findings provided a new perspective for understanding the effects of invasive plants on soil microbial communities via their impacts through phytochemicals.


Introduction
Invasive plants often change soil (a)biotic conditions, leading to altered soil microbial community diversity, structure and function (Kourtev et al. 2002;Kueffer et al. 2008;Liu et al. 2020;Prescott and Zukswert 2016;Sun et al. 2019;Zhao et al. 2019). These impacts are mediated primarily through two pathways. First, invasive plants may produce more litter which differs in chemistry from that of native species (Ehrenfeld 2003;Jo et al. 2017;Kueffer et al. 2008;Prescott and Zukswert 2016). Second, invasive species release novel phytochemicals into soils as root exudates or foliar excretions that can directly interact with soil microbes Hierro and Callaway 2003;Morris et al. 2016;Zhang et al. 2009). While the impacts of invasive plants on soil microbial community structure and function have attracted a lot of attention recently (Keet et al. 2021;Song et al. 2015;Xiao et al. 2014;Yang et al. 2020;Zhang et al. 2019), it is generally not well-understood what the contribution of these two main pathways are to changes in soil microbial communities under invasion.
Metabolites produced by some plants can also in uence how plants interact with each other. The synthesis of phytotoxic chemicals by invasive plant species can disrupt the growth of native plants (Seigler 1996). For example, some phytochemicals may have antifungal activities that inhibit plant pathogens and/or mutualists (Lankau 2012;Yang et al. 2016;Zheng et al. 2018). Changes in soil microbial communities in response to invasion may also impact soil nutrients (Becerra et al. 2018;Czaban et al. 2018). Ultimately, these changes may lead to nutrient cycles that enhance the performance of invasive plants through positive plant-soil feedbacks (Inderjit et al. 2009;Lankau 2012;Morris et al. 2016). For example, invasive spotted knapweed Centaurea maculosa in North America modi es soil microbial communities to its own bene t (Callaway et al. 2004). Similarly, we found that invasive Mikania micrantha enrich soils for microbes that participate in nitrogen cycling in China, causing positive plantsoil feedbacks by increasing soil nitrogen availability (Liu et al. 2020). Yet, few studies have documented the effects of speci c phytochemicals produced by invasive plants on soil microbial communities and on soil nutrients. M. micrantha is also known as 'mile-a-minute' weed. The species is native to the tropical regions of South America, Central America and the Caribbean, and is invasive in many tropical regions of Asia and the Paci c (Zhang et al. 2004;Manrique et al. 2011). The invasive success of M. micrantha is partly due to its strong allelopathic effects on native plants and soil microbes Kaur and Malhotra 2012;Wu et al. 2009). Various phytochemicals, such as sesquiterpene lactones (Huang et al. 2008;Piyasena and Dharmaratne 2013) and phenolic compounds (Xu et al. 2013) have been isolated from different M. micrantha tissues. Our latest work found sesquiterpene lactones (STLs) to mainly accumulate in the species' leaves and owers, and that these phytochemicals are released into the surrounding soil through litter (Liu et al. 2020). However, very little is known about the effects of these STLs on soil microbes and how they relate to the invasion success of M. micrantha. In this study, we use manipulative experiments and next generation sequencing (NGS) approaches to investigate how Mikania STLs in uence soil microbial communities and nutrients. We hypothesized that soil nutrients would increase in the presence of Mikania STLs and therefore that soil bacterial diversity would increase. We also hypothesized that fungal diversity will decrease as Mikania STLs are known to have antifungal properties.

STLs manipulation experiment
We collected soils from two sites: a M. micrantha monoculture (i.e. invaded soils, hereafter IS) and a neighboring uninvaded area (i.e. uninvaded soils, hereafter US), located in a dry riverbed of Liuxi River, Guangzhou City, Guangdong Province, China (23º28 37.99 N,113º28 41.49 E). Five independent replicates of each soil type were randomly sampled from a 30 m × 30 m eld plot using the ve-point sampling method (Jin et al. 2015). All collected samples were transported back to the laboratory as soon as possible under cooled conditions. Soils were loosened, sifted through a 1.0-mm sieve and stored at 4 o C. Soils were weighed (equivalent of 30 g dry soil) and placed into sterile 100 mL incubation bottles and left at room temperature for one week.
We added 4.5 mL of ve different Mikania STLs (20 mg L − 1 ; anhydroscandenolide, deoxymikanolide, dihydromichaelide, scandenolide and 3-epi-dihydroscandenolide) separately to US. Therefore, the nal concentration of each STL per soil was approximately 3 mg kg − 1 . We also kept untreated US and IS soils as negative and positive controls, respectively. Soil moisture was adjusted with sterilized water to 70% of soil water holding capacity and bottles sealed using rubber plugs. Each treatment had ve replicates, leading to 35 samples in total (25 US with STL addition, 5 US and 5 IS). All bottles were placed in an incubator (RXZ intelligent, Ningbo Jiangnan Instrument Factory) for 12 days (28 o C, dark condition, 65% humidity) and rubber plugs opened for 30 min every two days to balance air. Gas in incubation bottles was extracted with a 50 mL syringe and collected in 100 mL aluminum foil air sampling bags (Shanghai Haocheng Technology Co., Ltd.). At the end of the experiment, a subset of each soil sample was stored at 4 o C to determine soil nutrient contents and a subset at -80 o C for DNA extraction for NGS.

Quanti cation of Mikania STLs and CO 2 concentration
Initial Mikania STLs content of IS was immediately analyzed by UPLC-MS (Waters, Milford, MA, USA) after collection in the eld. Analyses were performed on an ACQUITY™ UHPLC system couple with a triplequadrupole Xevo TQD mass spectrometer. An ACQUITY UPLC® BEH C18 column (2.1 mm × 50 mm, 1.7 µm) was employed and the column temperature maintained at 40 o C. The gradient elution with acetonitrile containing 0.1% formic acid (A) and water containing 0.1% formic acid (B), was performed as follows: 0-1.0 min, 20% A; 1.0-3.0 min, 20-60% A; 3.0-6.0 min, 60-95% A; 6.0-8.0 min, 95% A; 8.0-8.5 min, 95 − 20% A; 8.5-10.0 min, 20% A. The ow rate was set at 0.3 mL min − 1 . The auto-sampler was conditioned at 22 o C and the injection volume of solution was 2 µL for all analyses. Mass spectrometric detection was performed on Xevo-TQD equipped with an electrospray ionization source (ESI). The capillary voltages were set to 3.0 and 2.22 kV at positive and negative modes, respectively, and the source temperature maintained at 150 o C. The collision gas was Ar, and N 2 was used for desolvation at 400 o C and cone gas at a ow rate of 700 L h − 1 , the cone gas set to 50 L h − 1 . Compounds for multiple reaction monitoring (MRM) were optimized in negative mode, the dwell time being 0.025 s (Table S1). Mass spectrometry and selected ion recording (SIR) were also used in relative quantitative analysis in both positive and negative ions measures.
After 12 days of incubation, CO 2 concentration was measured in each bottle using an Agilent 7890B Gas Chromatograph (Agilent Technology, USA), in which hydrogen ame ionization detector (FID) was used for detection, the temperature of which was set at 250 o C, the temperature of the separation column set at 55 o C, and the carrier gas was high-purity N 2 .
The in uence of dihydromikanolide on soil nutrients and soil acidity Because dihydromikanolide was the STL with the highest concentration in IS that caused the largest increase in CO 2 concentration (see Results section), we chose it to compare the effects of lactones addition (i.e. dihydromikanolide soils, hereafter referred to as DS) on soil nutrients and microbial communities with US and IS.
Subsets of fresh soil from dihydromikanolide treatments and controls were immediately extracted with 50 mL 2 mol L − 1 KCl to determine ammonium nitrogen (NH 4 + -N) and nitrate nitrogen (NO 3 − -N) concentrations by Continuous Flow Analyzer (Proxima, Alliance instruments, France). Subsets of each fresh soil sample was also air-dried at room temperature. A subset of the air-dried soil was sieved through a 0.15-mm mesh and nally used for total N measurement by combustion using a TOC analyzer (LI-8100A, Elementar company, Germany), available phosphorus (AP) by using spectrophotometer (UV-2000, Shimadzu, Kyoto, Japan). A second subset of the air-dried soils was sieved through a 2-mm mesh and used for available potassium (AK) measurements by using ame atomic absorption spectrometry (Z-5300, Polarized Zeeman Atomic, Absorption Spectrophotometer). Soil pH was measured using electrode pH meter (ST3100, Ohaus Instrument (Changzhou) Co., Ltd.), in a 1:5 (w/v) soil-water suspension.
Effects of dihydromikanolide on soil microbial community diversity and structure Total genomic DNA was extracted from dihydromikanolide-treated and control soil samples (three independent samples for each soil type; total n = 9) using the PCR products were sequenced by Pair-end 2 × 300 bp using the Illumina Miseq platform at Shanghai Personalbio Biotechnology Co., Ltd (Shanghai, China). Low-quality sequences (< 150 bp in length, average Phred scores < 20, containing mononucleotide repeats of > 8 bp) were ltered out (Chen and Jiang 2014;Gill et al. 2006). FLASH was used to assemble paired-end reads (Magoč and Salzberg 2011).
After chimera detection, the remaining high-quality sequences were clustered into operational taxonomic units (OTUs) at 97% DNA sequence similarity using the UCLUST algorithm (Edgar 2010; Navarro-Noya et al. 2013). OTUs accounting for less than 0.001% of all sequence reads were discarded. Representative sequences of all OTUs were aligned and annotated against the Greengenes database (Release 13.8) for bacterial OTUs and the UNITE database (Release 7.0) for fungal OTUs. The QIIME software (Version 1.8.0) was used to draw rarefaction curves and to calculate diversity indexes of samples, including Chao1, ACE, Shannon and Simpson's. Cluster analyses and heat maps of the top 50 genera were drawn in QIIME. Venn diagrams were drawn in the R statistical environment. Linear discriminant analysis of effect size (LEfSe) was used to identify indicator taxa (biomarkers) in different soil samples. This approach combines linear discriminant analysis with non-parametric Kruskal-Wallis and Wilcoxon rank sum tests to screen for key biomarkers (Segata et al. 2011). By submitting the relative abundance matrix (at genus level) on the galaxy online analysis platform (http://huttenhower.sph.harvard.edu/galaxy/), LEfSe can automatically analyze the composition of each classi cation level and visualize the results. We used partial least squares discriminant analyses (PLS-DA) based on species abundance matrix and sample to analyze species classi cation levels. PLS-DA is a supervised pattern recognition method based on partial least squares regression models, and it reorders samples in a new low dimensional coordinate system by searching for the maximum covariance of the species richness matrix and the given sample distribution or grouping information. Since PLS-DA can reduce the in uence of multicollinearity among variables, it is more suitable for the study of complex microbial community datasets. We used the PICRUSt software (Langille et al. 2013) to predict gene function present in our soil bacterial communities. Fungal sequencing reads were associated with putative ecological functions using FUNGuild tools (Nguyen et al. 2016).

Statistical analysis
All statistical analyses were done using SPSS 11.0 software (SPSS Inc., USA). Mikania STL concentration, CO 2 concentration, soil pH and nutrient contents and soil microbial diversity indices were individually analyzed using one-way ANOVAs, followed by Duncan-test at P < 0.05. The origin 8.5 software was used to visualize data.
CO 2 concentrations gradually increased in the all soils over the rst three days of incubation ( Fig. 1-b).
Changes in soil nutrients and soil acidity in response to dihydromikanolide Total N and NO 3 − -N contents were signi cantly reduced, and the content of NH 4 + -N signi cantly increased, in DS compared with US (Fig. 2). On average TN and NO 3 − -N content decreased by 3.69% and 46.55% respectively, while NH 4 + -N content increased by 58.79%. The addition of dihydromikanolide to US also increased AP and AK compared with US only and on average AP and AK content increased by 18.24% and 14.77% in DS, respectively. There was no signi cant difference in soil nutrients between DS and IS (Fig. 2), but there is a signi cant difference in the pH value between DS (4.78 ± 0.15) and IS (5.48 ± 0.11) (P < 0.05).
Microbial alpha-diversity responses to dihydromikanolide Rarefaction curves for bacteria and fungi plateaued in all three soils types (US, DS and IS) indicated that our sequencing depth was reasonable (Fig. S1).
Estimates of alpha diversities showed Chao1 and ACE indices for bacteria to be signi cantly higher in DS and IS compared with US, while the Chao1 index for fungi in DS was signi cantly lower than in IS and US (Table 1). Simpson and Shannon indexes of the bacterial and fungal were not signi cantly different among the three soil types (Table 1).  (Fig. 3). The fungi:bacteria ratio was 0.34, 0.28 and 0.34 in US, DS and IS, respectively (Fig. 3).
The number of soil bacterial and fungal OTUs that could be classi ed to phylum, class, order, family and genus levels is provided in Table 2. Our results showed that the addition of dihydromikanolide led to bacterial diversity across various taxonomic levels that was similar to IS but signi cantly higher than in US. However, fungal diversity in DS was not signi cantly different from that of US, and only signi cantly lower than that present in IS at the phylum-level. Eight bacterial phyla had relative abundances > 1% across soil types, namely Proteobacteria, Acidobacteria, Chloro exi, Actinobacteria, Gemmatimonadetes, Planctomycetes, Firmicutes and Verrucomicrobia, accounting for more than 95% of the total bacterial community ( Fig. 4-a). Among these, Proteobacteria had the highest relative abundance, accounting for between 23.19%-29.27% of all OTUs, followed by Acidobacteria, accounting for between 22.71%-27.61% of all OTUs. Across different soil types, the relative abundance of Proteobacteria was signi cantly higher in DS and IS compared with US, while the relative abundance of Chloro exi and Gemmatimonadetes was signi cantly lower in these two soils compared with US (P < 0.05). Compared with US, the relative abundance of Firmicutes in DS was signi cantly higher, while that of Acidobacteria was signi cantly lower (P < 0.05).
We identi ed four fungal phyla with relative abundances > 1% in all soils, namely Ascomycota, Basidiomycota, Mortierellomycota and Rozellomycota, accounting for between 55.09%-68.58% of taxa present in our fungal communities ( Fig. 4-b). Among them, Ascomycota had the highest relative abundance, followed by Basidiomycota. The relative abundance of Ascomycota and Mortierellomycota was signi cantly higher in DS and IS compared with US; while the relative abundance of Basidiomycota was slightly lower in these two soils than in US (P < 0.05). The relative abundance of Rozellomycota in IS was signi cantly lower than in US (P < 0.05).
Bacterial and fungal community structure and function in response to dihydromikanolide PLS-DA indicated that differences in the structure of bacterial and fungal communities were mainly driven by soil type (Fig. 6). In all instances, community structure between DS and US was more similar than between any one of them and IS (Fig. 6). Genus-level clustering and heat map analyses of bacteria and fungi similarly found DS and US to cluster together, and separately from IS (Fig. 7).
Our PICRUSt analysis found bacterial functional groups in DS and IS to cluster, which was apparently different from those in US (Fig. 8). Moreover, most functions in carbon, nitrogen, phosphorus, and sulfur metabolisms in DS and IS were signi cantly (P < 0.05) higher than those in US (Table 3). On the other hand, FUNGuild analysis found fungal functional groups of DS to mainly cluster with US, and to be different from IS (Fig. 8).

Discussion
Soil microbes not only play important roles in how soil ecosystems react to the abiotic changes caused by invasive plants (Rout and Callaway 2009), but also in the allelopathic effects of invasive plants on native species (Inderjit and van der Putten 2010; Uddin et al. 2017). Here we show that Mikania STLs signi cantly altered both soil CO 2 release and nutrient availability. These ndings suggest that secondary metabolites exuded by this invasive vine into soils can enhance soil microbial activity to promote soil nutrient mineralization. Our focus on the speci c Mikania STL, dihydromikanolide, provided important insights of how phytochemicals impacts these processes.
We found that dihydromikanolide to signi cantly increase soil bacterial richness, but to have insigni cant effects on bacterial Shannon or Simpson diversity. Bacterial OTU richness in dihydromikanolide-treated soils was similar to that in Mikania-invaded soils collected in the eld. On the other hand, we found richness of fungi to decrease in the presence of dihydromikanolide, compared with US and IS, indicating a negative/inhibitory effect of this phytochemical on soil fungi. There are potentially many reasons for this observation. For example, bacteria was more sensitive to soil pH than fungi (Rousk et al. 2010), which was signi cantly in uenced by dihydromikanolide addition in our study. However, the opposite has been found, where plant invasion led to decreased bacterial diversity or increased soil fungal diversity (Lorenzo et al. 2013;Si et al. 2013). Obviously, the effects of invasive plants on soil microbial diversity may depend on the species-speci c phytochemicals or the speci c environmental contexts under which they operate.
Previous studies have changed soil nutrients following plant invasion to impact soil fungal:bacterial (F:B) ratios (e.g., Perkins and Nowak 2013). Moreover, ecosystems with high F:B ratios typically have higher carbon storage and slow soil organic matter (SOM) turnover (Malik et al. 2016;Soares and Rousk 2019).
In our study, the F:B ratio in DS was lower than that in US, accompanied by higher CO 2 release in DS, implying that M. micrantha may weaken soil carbon storage and enhance SOM turnover through its phytochemicals. Surprisingly, the F:B ratio in DS was also lower than that in IS, suggesting that these changes may develop over time. Therefore, we speculate that, while M. micrantha may initially interfere with the soil micro-environment by releasing phytochemicals, more stable F:B ratios may emerge after successful colonization. Another reason for these differences may be because the micro-environment in IS has been in uenced by M. micrantha roots for longer, and by more phytochemicals, than that of DS.
The changes in bacterial and fungal communities we observed might be related to certain dominant microbial groups occupying unique niches (Fig. 4). Compared with uninvaded soils, the relative abundance of Proteobacteria in dihydromikanolide and Mikania-invaded soils was signi cantly higher, while the relative abundance of Chloro exi and Gemmatimonadetes was signi cantly lower. Moreover, Firmicutes in dihydromikanolide soils increased signi cantly, while Acidobacteria decreased ( Fig. 4-a).
Similar effects have been observed for the secondary metabolite astragalin produced by the invasive plant Flaveria bidentis, with the relative abundance of soil Proteobacteria increasing when astragalin is added exogenously to soils (Zhang et al. 2016). Acidobacteria are oligotrophic and their ability to compete with eutrophic bacteria is relatively weak in high-nutrient environments (Pascault et al. 2013).
Therefore, M. micrantha may enhance immediate nutrient availability by secreting phytochemicals that reduce oligotrophic bacterial groups during the early stages of colonization, which is consistent with the characteristics of M. micrantha growing in high-nutrient soil environments (Zhang et al. 2004).
With regard to fungi, the relative abundance of Ascomycota and Mortierellomycota in dihydromikanolide and invaded soils in this study was signi cantly higher compared with uninvaded soils, whereas the opposite was true for the relative abundance of Basidiomycota ( Fig. 4-b). Ascomycota is known to promote the transformation of soil organic matter (Egidi et al. 2019;Hanson et al. 2008).
Mortierellomycota possesses a variety of plant growth-promoting characteristics, such as phosphate solubilization, nitrogen supply and enhanced disease resistance (Liao et al. 2013;Zhang et al. 2011).
These general characteristics are consistent with our results showing that NH 4 + -N and AP contents increase after the addition of dihydromikanolide. (Fig. 2). By contrast, Basidiomycota is a major fungal phylum that includes important plant and animal pathogens (Martinez et al. 2004), some with high virulence and wide host ranges (Olson and Stenlid 2001;Zhang and Zhang 2015). Therefore, M. micrantha may increase the abundance of bene cial fungi and reduce the number of soil pathogens through some of its phytochemicals, thereby changing soil microbial communities in ways that promotes its own growth (e.g. Blumenthal et al. 2009;Lankau 2011) and negatively impacts co-occurring natives (e.g. Mangla et al. 2008;Stinson et al. 2006 (Chen et al. 2020). Previous work has also found leave extracts from M. micrantha to enhance soil nitri cation rates and higher ammonium (NH 4 + -N) and nitrate soil nitrogen (NO 3 − -N) levels ). Our previous work has also found the relative abundances of microbes carrying nitrogen cycling genes, ammonium-oxidizing bacteria, phosphorus-solubilizing, and potassiumsolubilizing bacteria to be signi cantly higher in M. micrantha-invaded soils compared with uninvaded or control soils (Liu et al. 2020). As a result, these enriched microbial functional groups promote utilization of the available nitrogen, particularly ammonium nitrogen, phosphorus, and potassium by M. micrantha.
In agreement with these ndings, we found the amount of certain bacterial functional groups, such as those involved in in carbon, nitrogen, phosphorus, and sulfur metabolism, to be similar between dihydromikanolide-treated soils and invaded soils, but signi cantly higher than in uninvaded soils, suggesting that M. micrantha can enhance bacterial functional groups via phytochemicals.

Conclusion
In general, the addition of Mikania STLs enhanced soil microbial activity and increased the availability of soil nutrients. Our DNA barcoding results indicate that soil microorganisms were stimulated by Mikania STLs, of which bacteria were more sensitive to dihydromikanolide than fungi. These changes were associated with higher bacterial richness and functional diversity that appears to accelerate the release of available nutrients. These effects may contribute the ease by which M. micrantha colonize new habitats. These ndings provide a new perspective for understanding the phytochemical effects of invasive plants on soil microbial communities and how these may impact invasiveness.

Figure 3
Venn diagrams of soil bacteria (left) and fungi (right) in three soils at the OTU level (n=3 for each soil type). IS -invaded soil; US -uninvaded soil; DS -uninvaded soil with dihydromikanolide (Dih) addition.

Figure 4
Page 21/24 Microbial composition of bacteria (a) and fungi (b) in different soil types at the Phylum level (n=3 for each soil type). IS -invaded soil; US -uninvaded soil; DS -uninvaded soil with dihydromikanolide addition.

Figure 5
Key community members (indicator taxa) of soil bacteria (left) and fungi (right) in different soil types identi ed by linear discriminant analysis effect size (LEfSe) analysis (n=3 for each soil type). IS -invaded soil; US -uninvaded soil; DS -uninvaded soil with dihydromikanolide (Dih) addition. Vertical axes display indicator taxa in each soil type (different bar colors indicate the grouping of samples with higher abundance corresponding to the taxon: blue-uninvaded soil; red-invaded soil; green-uninvaded soil with dihydromikanolide addition) and horizontal axes provide their LDA scores (higher scores correspond to more important indicator taxa).