Direct and Indirect Effects on Soil Nematode Communities Differ Between Facilitative and Allelopathic Plants

Plants are expected to affect soil nematode communities. However, comparative studies on the direct and indirect ways dominant plants inuence soil nematode communities are rare. In this study, we compared the effects of a dominant allelopathic plant, Ligularia virgaurea, and a dominant facilitative plant, Dasiphora fruticosa, on soil nematode richness and community composition in an alpine meadow of the Tibetan plateau. Our result indicated that 1) D. fruticosa signicantly increased nematode richness whereas L. virgaurea had no signicant effect; 2) D. fruticosa had no signicant impact on bacterial and fungal richness, but L. virgaurea increased fungal richness; 3) D. fruticosa had strong positive direct, and weak positive indirect, effects on nematode richness, mainly mediated by a marginal decrease in fungal richness. By contrast, L. virgaurea had no signicant direct effect on soil nematode richness but had strong indirect effects, mainly mediated by changes in soil pH and soil organic carbon content; 4) L. virgaurea inuenced soil nematode community composition predominantly through direct effects but also indirectly through soil organic carbon. By contrast, D. fruticosa affected nematode communities through changes in understory plant communities, soil physiochemical, and microbial communities. Both facilitative and allelopathic plants thus inuence soil nematode richness and community composition but seemingly in different ways. These highlight the importance of plants in determining soil community diversity and provide new insight to disentangle the complex above- and belowground linkages.


Introduction
Soil organisms play a major role in ecosystem functions, including plant productivity, nutrient mineralization and decomposition (Bongiorno et al. 2019;Neher 1999b). Soil nematodes are small semiaquatic multicellular animals that occupy a central position in the soil food web linking primary producers and primary consumers with higher trophic levels (Li et Yeates (1999) found that changes in the aboveground plant community alters nematode community composition, particularly plant-feeding nematodes, but also fungivores and bacterivores due to changes in microbial communities (Yeates et al. 1993). Wang et al. (2019b) demonstrated that soil nitrogen (N) content, ammonium and pH negatively affected the biomass of fungivores, while soil C and N positively and negatively affect omnivores-predators biomass, respectively. A recent global scale study found that soil characteristics explain most of the variation in nematode abundances, with greater numbers in soils with high organic C content and lesser when soil pH is low (van den Hoogen et al. 2019). Moreover, increased soil water content has a positive effect on the abundance of plant-feeding nematode (Ruan et al. 2012).
Certain plants have functional traits that increase their in uences on the surrounding environment.
Facilitative plants can promote seed germination, survival, and growth of neighboring plants by ameliorating the harsh local environment, or by promoting bene cial microbes or suppressing plant pests and pathogens (Callaway 2007 In this study, we compared the effects of a dominant allelopathic plant, Ligularia virgaurea, and a dominant facilitative plant, Dasiphora fruticosa, on the soil nematode community in an alpine meadow on the Tibetan plateau. We used high-throughput sequencing methods to identify the richness and composition of soil microbes as well as soil nematodes. Morphological identi cation of nematodes is time-consuming and requires excellent taxonomic skills, and is generally limited by incomplete description of local nematode species (Morise et al. 2012). High-throughput sequencing technology has been shown to overcome these weaknesses (Geisen et  We applied structural equation modeling (SEM) to assess the direct and indirect effects of two contrasting dominant plants on nematode community composition through changes in the understory plant community, soil properties as well as soil microbial communities. We hypothesized that: i) Dasiphora fruticosa and Ligularia virgaurea increase and reduce nematode richness, respectively, because Dasiphora fruticosa can protect understory plants thereby increasing resource availability, while allelochemicals released by Ligularia virgaurea may be poisonous to nematodes and reduce the richness of nematodes. ii) Dasiphora fruticosa and Ligularia virgaurea differ in their direct and indirect in uences on soil nematode communities, with Dasiphora fruticosa in uencing soil nematode communities mainly through biotic factors (understory composition and microbial communities), while Ligularia virgaurea increase soil nutrient and soil fungal richness, but decrease understory plant richness, and directly and indirectly affect soil nematode. Province. Azi Branch Station is located 3500 m above sea level. The annual precipitation is 620 mm, and the rain falls mainly during the short, cool summer. There are approximately 2580 h of cloud-free solar radiation annually ). The experimental plot is a typical sub-alpine meadow dominated a few shrubs, including the two focal shrubs, and annual herbaceous plants.

Experimental design
In June 2016, a grazed eld site with a healthy population of both Dasiphora fruticosa and Ligularia virgaurea was selected for the study. Within this area, we randomly allocated fteen 30 cm × 30 cm plots, 5 with D. fruticosa (abbreviation: Dasiphora), 5 with L. virgaurea (abbreviation: Ligularia) and 5 without D. fruticosa and L. virgaurea (hereafter "control"), respectively.

Soil sampling and vegetation survey
In August 2016, at the end of growing season, we collected three soil cores from the center of each plot (within a 30 cm × 30 cm quadrat) using a soil auger (4 cm diameter, 20 cm depth), picked out stones and then hand-mixed soil in plastic bags. Soils were kept at 4 ℃ until processing. A subsample of the soil from the mixed sample was transferred to a sterile 15mL centrifuge tube and stored at -80 ℃ for molecular analyses. Prior to soil sampling, we recorded the species and number of plants within the same 30 cm × 30 cm quadrate.

Soil properties
Soil water content was measured by drying 30 g soil for 72 h at 105 ℃. The remaining soil was air-dried, avoiding direct sunlight following removal of gravel and plant residues by hand, and then sieved through a 100 mesh (0.15 mm). Soil pH was measured in a 1:2.5 soil: deionized water slurry using a pH meter (PHSJ-3F, Shanghai INESA Scienti c Instrument Co., Ltd, China). Soil organic matter was measured based on dichromate oxidation procedure (Kalembasa and Jenkinson 1973). Soil total nitrogen and phosphorus were both digested by concentrated H 2 SO 4 , and measured by semi-micro Kjeldahl and Mo-Sb antispetrophotography, respectively, using an auto chemistry analyzer (SmartChem 200, AMS Alliance) (Baillie 1990;Hendershot 1985). Soil ammonium and nitrate nitrogen were measured based on heating digestion method. . Quality-ltering was consistent with nematode methods. The remaining sequences were clustered into OTUs using the uparse software (http://www.drive5.com/usearch/manual/uparseotu_algo.html) according to their similarities, and 97% similarity level was selected, which generally represent microbial taxonomy at the species level.
For bacteria, the RDP database (http://rdp.cme.msu.edu/misc/resources.jsp) was compared, and the fungal ITS area was compared with the Unite database (https://unite.ut.ee/index.php). We used the RDP classi er to compare the OTU representative sequence with the corresponding database to obtain the OTU species information, at con dence threshold of 80%.

Soil nematode identi cation
Nematode sequences were similarly ampli ed from the genomic DNA. In order to improve ampli cation e ciency, a higher proportion of nematode sequences can be obtained without nematode enrichment, we repeated DNA ampli cation before sequencing. The primers NemF and 18Sr2b (Sikder et al. 2020) were used in a pre-ampli cation step followed by ampli cation with primers NF1 and 18Sr2b in a semi-nested procedure (Sapkota and Nicolaisen 2015) (Table S1). NF1 and 18Sr2b were tag encoded using the forward primer 5 -CGTATCGCCTCCCTCGCGCCATCAG-MID-NF1-3 and the reverse primer 5 -CTATGCGCCTTGCCAGCCCGCTCAG-18Sr2b-3 (Sapkota and Nicolaisen 2015). Reactions contained 12.5µl of 2 × Taq PCR mixture with loading dye reaction buffer (GenStar), 2.5µl each of forward primer and reverse primer, 1 µl of DNA template, and 6.5µl ddH 2 O in a nal volume of 25 µl. Ampli cation with NemF and 18Sr2b used an initial DNA denaturation step of 94°C for 5 min, followed by 20 cycles at 94°C for 30 sec, 53°C for 30 sec, 72°C for 1 min and a nal elongation at 72°C for 10 min (Sapkota and Nicolaisen 2015). After ampli cation, DNA samples were subjected to 1% agarose gel electrophoresis to check whether they can be used in subsequent experiments. DNA was sequenced using Illumina Miseq PE300 High-Throughput sequencing if the DNA sample is quali ed.
We used QIIME for quality-ltering (Caporaso et al. 2010). To remove the interference sequence, we 1) split the sequence into samples according to the barcode and removed the barcode; 2) deduplicated the double-end sequences by the "Trimmomatic" software: removed bases with a tail quality value lower than 25; a 50bp sliding window was set, with a 1bp step, and the average base quality in the window was not less than 25; sequences less than 100bp were removed; 3) connected high-quality double-end sequences, with a minimum overlap region of 10bp and a maximum mismatch rate of 0.2, then removed sequences containing ambiguous base N by the " ash" software. Valid sequences without chimeras were subsequently clustered into different OTUs (Operational Taxonomic Units) by uparse (http://www.drive5.com/usearch/manual/uparseotu_algo.html) according to their similarities, and 99% similarity level was selected here. The RDP classi er was used to compare the OTU representative sequence with the Silva 18S (version 123) database to obtain OTU species information. The con dence threshold used by the RDP classi er to compare species databases was 80%.

Data analyses
We calculated OTUs richness for soil microbes and nematodes. For plants, we calculated understory species richness. We used Levene's test in the "car" package to test the homogeneity of variance and Shapiro-Wilk test to test the normality of data. If the data tted the normal distribution and homogeneity of variance, the nematode richness, soil variables, microbial richness and understory species richness were assessed for differences between treatments using one-way analysis of variance (ANOVA) followed by Tukey-test when main effects were observed. We used permutation analysis of variance tests in "lmPerm" package if the data did not conform to the assumptions of normal distribution and variance homogeneity followed by multiple comparisons by "kruskalmc" function belong to "pgirmess" package.
Non-metric multidimensional scaling (NMDS) based on the "Bray-Curtis" dissimilarity index in "vegan" package was used to visualize the spatial distance arrangement of the nematodes, plants, bacteria and fungi. Non-parametric multivariate analysis of variance (PerMANOVA) with 9999 permutations in "vegan" package was used to assess differences in community composition between plant treatments. We conducted a classi cation Random Forest analysis using the "randomForest" package to identify the relative importance of soil physicochemical properties variables in explaining nematode richness by measuring the increase in the mean square error (MSE) between the observed value and the OOB predicted value. The prediction accuracy was averaged across all trees (10000 trees) to produce the nal importance measure (Delgado-Baquerizo et al. 2016), and we then selected the two most important properties for the constructing of structural equation modeling (SEM). Structural equation modeling (SEM) in "lavaan" package was applied to explore the in uence ways of the different treatments on soil nematode community composition. The model t was evaluated through χ 2 test and root mean square error of approximation (RMSEA) test. The gures were plotted using the 'ggplot2' package. And all data were analyzed using R software, version 3.6.3 (R Core Team).
We conducted structural equation modeling (SEM) analyses according to a priori model (Fig. S1) with following these premises: (1) soil nematode richness and communities composition can be affected by soil physicochemical properties, understory plant richness and communities composition, soil microbe richness and communities composition, and dominant plant species (Wang et

Results
There was no signi cant difference in the effects of L. virgaurea and D. fruticosa on understory plant richness (Fig. 1b), but non-metric multidimensional scaling (NMDS) and non-parametric multivariate analysis of variance (PerMANOVA) results showed that plant community composition based on presence, signi cantly differed between D. fruticosa and the control treatment (Fig. 2b, P<0.01), and marginally differed between L. virgaurea and the control treatment (Fig. 2b, P<0.1).
Soil water content, soil total phosphorus, soil total nitrogen, soil organic carbon, and soil ammonium nitrogen (P<0.01) was greater in the presence of L. virgaurea relative to the control treatment, while soil pH (P<0.01) and soil nitrate nitrogen was lower. Similarly, the presence of D. fruticosa was associated with greater soil total phosphorus, soil total nitrogen (P<0.05), soil organic carbon (P<0.05), soil ammonium nitrogen (P<0.01), and lower soil pH (P<0.01). However, D. fruticosa increased soil nitrate nitrogen and decreased soil water content, which is opposite to the effect of L. virgaurea (Table 1). we retained a total of 657833 sequences after ltering and removing of chimeras. The total number of bases was 224765243, and the average sequence length 341.68 at 97% similarity. We found there was no signi cant difference in the effects of L. virgaurea and D. fruticosa on bacterial richness (Fig. 1c). L. virgaurea increased fungal richness signi cantly (Fig. 1d, P<0.01), while fungal richness was not signi cantly affected by D. fruticosa (Fig. 1d). Non-metric multidimensional scaling (NMDS) and nonparametric multivariate analysis of variance (PerMANOVA) results showed that bacterial community composition under D. fruticosa was signi cantly different from that under control (Fig. 2c, P<0.05). However, there was no signi cant difference of fungal community composition among different dominant plant types and the control treatment (Fig. 2d). The presence of D. fruticosa increased the relative abundance of Proteobacteria, Bacteroidetes and Nitrospirae, and the presence of L. virgaurea increased the relative abundance of Proteobacteria, Verrucomicrobia (Fig. S3c). L. virgaurea decreased the relative abundance of Ascomycota and D. fruticosa decreased the relative abundance of Glomeromycota (Fig. S3d).
For Eukarya, we retained a total of 4143079 sequences after ltering and removing chimeras. The total number of bases was 1536595586, and the average sequence length 370.88 at 99% similarity. Following classi cation of OTUs, we found that the average nematode content was 41.81%, and other metazoan and fungi was 58.19% (Fig. S3a). We found a signi cant positive effect of D. fruticosa on nematode richness (Fig. 1a, P<0.001), but L. virgaurea had no effect on nematode richness (Fig. 1a). Although the richness of soil nematodes was not signi cantly affected by L. virgaurea, the community composition of nematodes was signi cantly changed. Interestingly, as opposed to L. virgaurea, D. fruticosa increased the nematode richness but there was no signi cant effect of nematode communities composition (Fig. 1a,  2a). There were different responses of soil nematode species to the allelopathic L. virgaurea and the facilitative D. fruticosa. Compared with the control treatment, the presence of both L. virgaurea and D. fruticosa had a positive effect on the proportion of Diplogasterida and a negative effect the proportion of Tylenchida and Triplonchida. The proportion of Rhabditida was increased in the presence of L. virgaurea and reduced in the presence of D. fruticosa. In addition, the presence of D. fruticosa increased the proportion of Araeolaimida and Enoplida (Fig. S3b).
Random Forest results indicated soil organic carbon (34.87% IncMSE) and soil pH (22.85% IncMSE) were the two most important edaphic variables in explaining variations in nematode richness (Fig. S2). Soil pH can be considered as an indicator of soil environment and soil organic carbon can be considered as an indicator of soil nutrition.
The SEM model (n=10) assessing the effects of L. virgaurea explained 84.7% of the variation in soil nematode richness. There were highly signi cant relationships between soil organic carbon, soil pH and understory plant richness and nematode richness. Soil organic carbon and pH was positively related to nematode richness whereas understory plant richness was negatively related to nematode richness. The direct effect of L. virgaurea on nematode richness was not signi cant. However, there was a negative indirect effect of L. virgaurea on nematode richness through soil pH and a positive indirect effect through soil organic carbon. Although nematode richness was signi cantly negatively related to understory plant richness, there was no signi cant relationship between understory plant richness and L. virgaurea. This indicated that L. virgaurea affected nematode richness mainly indirectly through changes in abiotic factors (Fig. 3a).
The SEM model (n=10) assessing the effects of D. fruticosa explained 91.8% of the variation in soil nematode richness. The SEM showed not only a signi cant direct positive effect of D. fruticosa on nematode richness, but also an indirect effect of D. fruticosa on nematode richness through fungal richness. Although nematode richness was marginally negatively related to bacterial richness, there was no relationship between bacterial richness and D. fruticosa. Also, D. fruticosa signi cantly affected soil pH and soil organic carbon, but there was no correlation between soil pH and soil organic carbon and nematode richness. This indicated that D. fruticosa affected nematode richness both through direct and indirect ways, and then predominantly through the biotic factors (Fig. 3b). Both L. virgaurea and D. fruticosa can reduce soil pH and increase soil organic carbon content. These characteristics had signi cant effects on the nematode richness under L. virgaurea but not under D. fruticosa (Fig. 3).
The SEM (n=10) assessing the effects of L. virgaurea explained 83.2% of the variation in the nematode community composition. The SEM indicated that L. virgaurea had signi cant direct effects on nematode community composition and indirect effects through its in uence on soil organic carbon (Fig. 4a). Similarly, the SEM (n=10) assessing the effects of D. fruticose explained 90.6% of the variation of nematode community composition. There were highly signi cant relationships between soil pH, soil organic carbon and understory plant community composition and nematode community composition under D. fruticosa. Interestingly, with the presence of D. fruticosa, in addition to the direct effect, soil pH and organic carbon indirectly affect nematode community composition through changes in soil fungal community composition, while understory plant community composition affect soil nematode community composition indirectly through changes in soil bacterial community composition (Fig. 4b). differently to soil pH. For example, the ratio of bacterivorous to fungivorous nematodes was markedly higher after manipulation of pH by liming (Räty and Huhta 2003). Our results showed that pH signi cantly increased soil nematode richness under L. virgaurea (Fig. 3a), and signi cantly changed the composition of nematode community under D. fruticosa (Fig. 4b). Soil nematodes are particularly abundant in habitats with greater organic carbon contents and inputs (Bongers and Ferris 1999) because most feed on soil organisms that utilize organic matter (Moens et al. 2002). The negative correlation between understory plant richness and soil nematode richness may be due to the speci city of food selectivity of herbivorous nematode (Ruess and Dighton 1996). As is shown in the Figure S3b, L. virgaurea decreased the relative abundance of Tylenchida, Araeolaimida and Enoplida.

Discussion
In addition to the indirect effects associated with changes in soil organic carbon, the composition of soil nematode communities is directly affected by L. virgaurea. The root exudates of the allelopathic plant Lantana camara L. (Verbinaceae) has been shown to cause mortality of M. javanica juveniles (Shaukat et al. 2003) and marigold (Tagetes patula) can produce allelopathic compounds toxic to plant-parasitic nematodes (Marahatta et al. 2012). We speculate that L. virgaurea release chemical compounds that affect the community composition of soil nematode given that the relative abundance of all nematode species was reduced, except Diplogasterida and Other. In particular, Araeolaimida were signi cantly reduced (Table S2). While our study indicate that allelopathic plants can affect soil nematode communities further work is required to verify how leachate from L. virguarea affect nematode communities and which compounds might be involved. had no distinct effect on bacterial richness but in uenced bacterial community composition. Speci cally, the presence of R. sphaerocarpa increased the relative abundance of the gram-negative Proteobacteria and Bacteroidetes. A higher relative abundance of Proteobacteria and Bacteroidetes suggests that the soil communities are less disturbed and are considered a better resource for microbial grazers (Fierer et al. 2007).

Conclusion
In conclusion, our study discovered that dominant plants with contrasting functional characteristics have markedly different impacts on soil nematode communities. The the facilitative plant D. fruticosa affects soil nematode richness both directly and indirectly through its in uences on soil fungal richness, while the allelopathic plant L. virguarea had no overall effect on nematode richness. Moreover, both species in uence nematode community composition directly and through their in uences on edaphic and biological properties. Speci cally, L. virgaurea impacted nematode communities through its in uences on soil organic carbon, while D. fruticosa impacted communities through its in uences on soil organic carbon, pH, and understory plant commnunity, and the soil microbial communities.