https://orcid.org/0000-0003-3914-5824
Abstract
The effects of fertilization on fungal plant pathogens in agricultural soils have been studied extensively. However, we know little about how fertilization affects the relative abundance and richness of soil fungal plant pathogens in natural ecosystems, either through altering the soil properties or plant community composition.
Here, we used data from a 7-year nitrogen (N) addition experiment in an alpine meadow on the Qinghai-Tibetan Plateau to test how N addition affects the relative abundance and richness of soil fungal plant pathogens, as determined using Miseq sequencing of ITS1 gene biomarkers. We also evaluated the relative importance of changes in soil properties versus plant species diversity under N addition.
Using general linear model selection and a piecewise structural equation model, we found that N addition increased the relative abundance of soil fungal plant pathogens by significantly altering soil properties. However, higher host plant species richness led to higher soil fungal plant pathogen richness, even after excluding the effects of N addition. We conclude that the relative abundance and richness of soil fungal plant pathogens are regulated by different mechanisms in the alpine meadow. Continuous worldwide N inputs (through both fertilizer use and nitrogen deposition) not only cause species losses via altered plant species interactions, but also produce changes in soil properties that result in more abundant soil fungal plant pathogens. This increase in pathogen relative abundance may seriously threaten ecosystem health, thus interrupting important ecosystem functions and services.
摘要
在农业生态系统中，氮添加对植物病原真菌丰富度和相对多度的影响已被基本阐明，然而在自然生态系统中，氮添加如何影响土壤中的植物病原真菌(通过影响植物群落结构或土壤理化指标)仍知之甚少。本研究以青藏高原东部高寒草甸为研究对象，基于7年的氮添加梯度野外实验，使用Miseq平台，针对土壤真菌的ITS1基因进行测序，以评估氮添加对高寒草甸土壤中的植物病原菌丰富度和相对多度的影响，并阐明氮添加通过不同途径(即植物群落结构和土壤理化指标)影响病原菌的潜在机制。基于模型筛选和结构方程模型等统计方法，本研究发现，氮添加通过改变土壤理化指标影响土壤中的植物病原菌相对多度。但是，在排除掉氮添加对土壤中植物病原菌丰富度的影响后，地上植物物种丰富度与土壤中植物病原菌丰富度仍存在显著的正相关性。因此，我们认为高寒草甸土壤中植物病原菌丰富度和相对多度受到不同的机制调控。世界范围内，自然生态系统中氮素输入量的加剧(包括氮沉降和氮肥施用)所引起的植物物种丧失引起了人类的较大关注。除此之外，氮素输入所引起的植物病原菌丰富度和相对多度的变化也值得我们警醒，因为植物病原菌群落结构的改变可能会对生态系统功能和服务产生重要影响。
INTRODUCTION
Plant pathogens are ubiquitous, found in a diverse array of ecosystems (Makiola et al. 2019) and across all known plant lineages (Alexander 2010). The pathogens of economically important plants (e.g. Triticum aestivum L. and Oryza sativa L.) in highly managed agro-ecosystems (e.g. Zhu et al. 2000) are the best characterized to date. However, pathogens in natural plant-pathogen systems are subject to more complex interactions and regulatory mechanisms (Mitchell et al. 2002; Parker et al. 2015), as they occur in complex spatial ecosystems with dozens of host plant and pathogen species (Liu et al. 2019). Despite this complexity, natural systems are gradually getting more research attention (Gilbert and Parker 2016).
In fact, plant pathogens (especially fungal pathogens) play a dual but important role in natural ecosystems (Chen et al. 2019; Mordecai 2011). On the one hand, pathogens cause disease, leading to decreases in host plant photosynthetic capacity and nutritional status (Ghelardini et al. 2016), seedling deaths (Jia et al. 2020) and biomass reductions (Seabloom et al. 2017). Furthermore, some invasive pathogens may cause ecological collapse in grasslands and forested ecosystems, leading to environmental disaster (Blumenthal et al. 2009; Fisher et al. 2012). On the other hand, pathogens can promote coexistence among plant species; e.g. when fitness costs (of infection) increase with a plant species relative abundance (e.g. Janzen–Connell effect, Janzen 1970; or negative plant–soil feedback, Bever 2003). Pathogen infection might also prevent competitive exclusion among plant species via competition/growth-defense trade-offs (Cappelli et al. 2020; Lind et al. 2013), ultimately affecting ecosystem function (Chen et al. 2020). However, conclusive insights into how environmental factors and plant community characteristics shape pathogen diversity and relative abundance are lacking at the local scale.
A range of abiotic and biotic factors may affect plant pathogen abundance (Liu et al. 2020a; Mitchell et al. 2003). Plant pathogens are likely to benefit from increased temperatures (Launay et al. 2014), moisture (Strengbom et al. 2006), water availability (Hong and Moorman 2005) and levels of disturbance (Makiola et al. 2019). Nutrients inputs are thought to increase pathogen abundance in agro-ecosystems (Anderson 2002; Veresoglou et al. 2014). The ‘nitrogen disease hypothesis’ states that high nitrogen availability increases plant palatability and susceptibility to pathogens due to increased metabolic rate, tissue nitrogen concentrations and specific leaf area (Mitchell et al. 2003; Semchenko et al. 2019). In a meta-analysis of 47 studies, nitrogen additions were found to increase plant fungal disease severity (i.e. pathogen abundance; Nguyen et al. 2017), providing extensive evidence for the nitrogen disease hypothesis (Veresoglou et al. 2014). However, most studies of how nutrient inputs affect pathogen abundance have been performed in agro-ecosystems. In natural ecosystems, the potential mechanisms driving changes in pathogen relative abundance under fertilization are likely more complicated (Halliday et al. 2018; Liu et al. 2017).
Natural ecosystems usually contain more than one host plant species. In these diverse systems, as total plant species richness increases, there tend to compensatory declines in the abundance of individual plant species (e.g. Mitchell et al. 2002; Rottstock et al. 2014). Many plant pathogens (e.g. rusts) show density-dependent transmission (Collins et al. 2020; Liu et al. 2020a), given their relatively narrow host range (Gilbert et al. 2012; Gilbert and Webb 2007). In communities with relatively high host plant diversity, there may be lower pathogen abundance as a direct result of the lower density of available host plants (Liu et al. 2016; Mitchell et al. 2002; Rottstock et al. 2014). The ‘dilution effect hypothesis’ posits that disease incidence (or pathogen abundance) decreases with increasing host species richness (Ostfeld and Keesing 2012). Based on a meta-analysis of 145 cases, Liu et al. (2020a) found that there was a significant overall dilution effect in natural grassland ecosystems, especially for biotroph pathogens. In addition, plant–soil feedbacks also provide strong evidence for the dilution effect (Bever 2003; Collins et al. 2020). Nitrogen addition is driving dramatic plant species losses on a local scale, as well as functional composition shifts in many ecosystems, including alpine meadows (Liu et al. 2017; Suding et al. 2005). Therefore, nitrogen additions may affect soil plant pathogen relative abundance via either changes to soil properties or host plant species richness (Cassman et al. 2016).
Apart from pathogen abundance, the richness of soilborne plant pathogens also represents another important community characteristic, which may be influenced by a series of abiotic (Makiola et al. 2019) and biotic factors (Kamiya et al. 2014). For instance, nitrogen additions raise nitrate and ammonia nitrogen concentrations in the soil, but lower soil pH (Li et al. 2014; Liu et al. 2017). Hence, as with other types of organisms (e.g. Li et al. 2015), soil pathogen richness under nitrogen addition depends directly on the relative strength of local extinction and colonization (Cassman et al. 2016). Several previous studies have also shown that higher plant species richness consistently leads to higher foliar fungal pathogen richness, even under variable conditions such as climate warming and nitrogen addition (Liu et al. 2016; Rottstock et al. 2014). As the richness of higher trophic levels is largely determined by their hosts, so plant species richness may directly translate to pathogen species richness (Kamiya et al. 2014). However, to date, the relative importance of soil properties versus plant community factors in driving pathogen relative abundance and richness remains poorly understood, especially in alpine meadows under conditions of nitrogen addition.
To address this knowledge gap, we performed a 7-year nitrogen addition experiment in an alpine meadow on the Qinghai-Tibetan Plateau to test how long-term nitrogen addition affects soil fungal plant pathogen relative abundance and richness, and to evaluate the relative importance of changes in soil properties versus plant community characteristics for pathogen relative abundance and richness. We focused on soil fungal plant pathogens, given their critical role in ecosystem health and functioning (Chen et al. 2019; Jia et al. 2020). In addition, previous studies at this site found that foliar fungal plant pathogen load, soil properties and plant community characteristics (including plant species richness) were all relatively sensitive to long-term nitrogen addition (Liu et al. 2017, 2020b); thus, this particular ecosystem represents an ideal system to separate the effects of soil versus plant community characteristics on pathogen communities. Specifically, we tested the hypotheses that: (i) nitrogen addition will increase the relative abundance of soil fungal plant pathogens via changes to soil properties; and (ii) higher host plant species richness will lead to higher soil fungal plant pathogen richness.
MATERIALS AND METHODS
Study site
The field experiment was conducted in the eastern part of the Qinghai-Tibetan Plateau, at the Research Station of Alpine Meadow and Wetland Ecosystems belonging to Lanzhou University in Maqu County, Gansu Province, the People’s Republic of China (101°53′ E, 35°58′ N; 3500 m a.s.l.). The mean annual temperature is 1.2 °C (minimum −10.7 °C in January and maximum 11.7 °C in July) with >270 days per year of frost, and mean annual precipitation is 620 mm, 85% of which falls during the short growing season (May–September). The nitrogen-limited soils in the region are classified as ‘chestnut’ soils, or ‘sub-alpine meadow soils’ according to the Chinese soil classification system, with a mean thickness of 80 cm (Liu et al. 2015). The grassland vegetation is a typical alpine meadow, and the plant community is dominated by perennial herbaceous species of Poaceae, Asteraceae and Ranunculaceae, such as Anemone trullifolia, Elymus nutans, Ligularia virgaurea and Saussurea stella. The dominant herbivores include yaks, horses, sheep, marmots (Marmota himalayana) and zokor (Myospalax spp.).
Experimental design
In June 2011, we established a nitrogen addition experiment on a southeast-facing meadow. The study site was fenced, with grazing (mainly by yaks) only permitted in winter. We regularly arranged sixty 5 m × 5 m plots with roughly the same plant community composition within the meadow, and included a 1 m buffer zone between adjacent plot edges. Among the 60 plots, 24 plots were randomly assigned one of four concentrations of supplemental nitrogen (supplied as ammonium nitrate NH4NO3, which provides a short-term release of nitrogen into the soil): 0 (control), 5, 10 or 15 g/m2. There were 6 replicates of each treatment, while the remaining 36 plots were used for other purposes (Liu et al. 2016). The experiment began in 2011, and we fertilized (NH4NO3) the treatment plots annually. The fertilizer was applied each year in mid-June on a cloudy day (a c. 2 h process) to avoid exposure to direct sunlight, which could lead to the quick degradation of surface nitrogen. Follow-up tests, completed after the 7-year experimental period, confirmed that the nitrogen addition treatments significantly altered soil properties, plant species richness and plant community composition (Supplementary Table S1). In a previous study we performed, nitrogen additions led to changes in plant community composition, which were now dominated by taller species with higher leaf phosphorus content, lower leaf nitrogen content, greater seed production and lower N:P ratios (Liu et al. 2017).
Sampling
In August 2017, for each plot in the nitrogen addition experiment, we randomly arranged a single 0.5 m × 0.5 m subplot parallel to one edge of the plot, yet at least 1 m away from any plot edge. We sampled the plant community composition annually beginning in 2011, but sampled soils only in 2017, given that soil collection negatively impacts the study site. We harvested all the stems in each subplot at ground level, then dried the samples at 70 °C and weighed them (to 0.1 mg) to determine the aboveground biomass (Babove). We assessed the abundance (i.e. number of individuals) of each species in the subplots, and also calculated total abundance (Abundance) as the sum of all individuals in each subplot. Next, we calculated the plant species richness (S), Shannon’s diversity index (H′), Pielou’s evenness index (J′) and Simpson’s diversity index (SDI) using the diversity function in the vegan package in R (Oksanen et al. 2013). We also performed a principal component analysis (PCA) of the plant community using the rda function in vegan and retained the first and second axis values (PCA1 and PCA2). In addition, we collected four soil cores (5 cm in diameter, 10 cm in depth) from each plot, and pooled them as one sample in the field. Roots were collected and then dried at 70 °C for belowground biomass (Bbelow). These variables (S, H′, J′, SDI, Abundance, Babove, Bbelow, PCA1 and PCA2) are referred to as the ‘plant community variables’ hereafter, given that they reflect plant community attributes.
Soil properties
Total soil carbon (C; g/kg) and nitrogen (Nitrogen; g/kg) were measured by combustion on an elemental analyzer (a vario EL III, CHNOS Elemental Analyzer, Elementar Analysensysteme GmbH, Germany). A pH analyzer was used to measure the supernatant of a 1:5 dry soil-to-water mixture for soil pH. Four grams of fresh soil were extracted with 20 mL 0.2 mol L−1 KCl for 1 h at 70 rev/s using a shaker, before measuring soil ammoniacal nitrogen (NH4+; mg/kg) and nitrate nitrogen (NO3−; mg/kg) in extracts using an auto-analyzer (AA3, Bran-Luebbe, Germany). Soil moisture content (Water; %) was measured gravimetrically after 72 h of desiccation at 70 °C. These variables (Water, NH4+, NO3−, Nitrogen and C) are referred to as ‘soil variables’ hereafter, given that they reflect soil attributes.
Molecular analyses and bioinformatics
Mortars and pestles were used to grind soil samples with liquid nitrogen. We then extracted total genomic DNA from each sample using MOBIO DNeasy PowerSoil Kits (Mo Bio Laboratories Inc., Carlsbad, CA, USA) according to the manufacturer’s protocol. The resultant DNA was evaluated using gel electrophoresis (1% agarose gel). The internal transcribed spacer 1 (ITS1) region was amplified using polymerase chain reactions (PCRs) with the forward primer ITS1F (5′-GGAAGTAAAAGTCGTAACAAGG-3′) and the reverse primer ITS1R (5′-GCTGCGTTCTTCATCGATGC-3′) (Gardes and Bruns 1993). The ITS1 region is commonly used for fungal pathogen identification (Lentendu et al. 2011), but is not specific for oomycetes. The 18.75 µL PCR reaction mixture included: 9.3 µL of Sigma REDtaq ReadyMix, 6 µL of DNA template, 1.5 µL of 16 mg/mL bovine serum albumin, 0.75 µL of 10 mmol L-1 MgCl2, 0.6 µL of 20 µmol L-1 forward primer and 0.6 µL of 20 µmol L-1 reverse primer. The PCR conditions were as follows: a hot start at 94 °C for 2 min; then 25 cycles of 94 °C for 30 s, 55 °C for 30 s and 72 °C for 1 min (to minimize PCR bias); reaction termination at 72 °C for 10 min and a hold at 4 °C. The PCR products were amplified in triplicate to minimize PCR bias, and PCR products were pooled for each sample.
Amplicon sequencing was performed on a paired-end 2 × 250 bp Illumina MiSeq Benchtop Sequencer (Illumina, San Diego, CA, USA) at MAGIGENE, Inc. (Guangzhou, China), and all paired sequences (24 samples) were successfully assembled. Quality filtering of sequences was performed using the FASTX-Toolkit v. 0.0.13 (Hannon Lab). We identified and removed chimeric sequences and singletons using UPARSE (Edgar 2013), before clustering the high-quality ITS1 sequences into operational taxonomic units (OTUs) at a 97% similarity threshold. For each sample, we rarefied to 40 000 sequences to eliminate the impact of sequencing depth for downstream analysis, enabling comparison among samples. Taxonomic affiliations for each OTU were assigned using the RDP classifier with a representative sequence (Wang et al. 2007) against the UNITE ITS sequence database (Abarenkov et al. 2010).
We employed the FunGuild algorithm (Nguyen et al. 2016) to tentatively assign functional groups of sequences to: Pathotrophs, Saprotrophs, Symbiotrophs and Others. The FunGuild algorithm relies on OTU taxonomic assignments rather than genetic loci (Nguyen et al. 2016). Given the importance of plant pathogens in maintaining plant biodiversity and ecosystem functioning, we also separated Pathotroph plant pathogens for subsequent analysis (Chen et al. 2019; Liang et al. 2016). In addition, we confirmed the identity of plant pathogens using information from the United States Department of Agriculture (https://nt.ars-grin.gov/fungaldatabases/). A fungal species was assigned as a plant pathogen if it was a member of a genus that consists of many plant pathogens that are known to cause plant disease symptoms (Chen et al. 2019; Liang et al. 2016). Lastly, the relative abundance of each functional group was calculated as the percentage of OTUs belonging to the group out of the total fungal OTUs for each sample.
Statistical analyses
Factors influencing soil fungal guilds
We employed one-way analyses of variance to test whether nitrogen additions (Treatment) affected the soil and plant community variables. The variables that we considered included: soil pH (pH), soil moisture content (Water; %), ammoniacal nitrogen content (NH4+; mg/kg), nitrate nitrogen content (NO3−; mg/kg), soil nitrogen content (Nitrogen; g/kg), soil carbon content (C; g/kg), plant species richness (S), Shannon’s diversity index (H′), Pielou’s evenness index (J′), SDI, total abundance (Abundance), aboveground biomass (Babove), belowground biomass (Bbelow) and the first and second axis values from a PCA of the plant community (PCA1 and PCA2). Tukey’s HSD test was used to evaluate how variables differed among treatments, with significance assigned at P < 0.05.
To assess how the nitrogen addition treatment affected plant species and soil fungal OTU richness, we set either plant species richness or soil fungal OTU richness as the response variable, and the amount of added nitrogen as the independent variable across the 24 study plots. To determine the shape of these relationships, we compared the AICc (Akaike’s information criterion corrected for small sample size) values for first- and second-order polynomials, and also an intercept-only model (i.e. null model) (Supplementary Table S2). We used the information-theoretic evidence ratio (ER, wAICc [slope model]: wAICc [intercept-only model]) as an index of relative support for the linear slope model versus the intercept-only (null) model; when ER >1.5, there was sufficient evidence to support the slope model (Burnham et al. 2011). We also calculated the percent deviance explained in the response variable (De) as an index of each model’s goodness-of-fit (Burnham et al. 2011). For each fungal OTU, we fit a linear model to test the relationship between the amount of nitrogen added (NH4NO3; g/m2) and the relative abundance of each fungal pathogen across the 24 plots.
Several additional relationships were evaluated: (i) the relative abundance/number of fungal OTUs versus the community-level variables; (ii) the relative abundance/number of pathotrophic fungal OTUs versus the community-level variables; (iii) the relative abundance/number of symbiotrophic fungal OTUs versus the community-level variables and (iv) the relative abundance/number of symbiotrophic fungal OTUs versus the community-level variables. To evaluate the best fit model, we constructed a series of general linear models with a single variable (i.e. Water, NH4+, NO3−, Nitrogen, C, S, H′, J′, SDI, Abundance, Babove, Bbelow, PCA1 or PCA2) as the independent variable using the lm function. We calculated the Spearman rank-order correlation between variables using the cor.test function. We calculated AICc, wAICc, ER and De as a measure of each model’s goodness-of-fit.
Structural equation model
In order to reduce the number of variables to a manageable level for the structural equation model (SEM) described below, we summarized soil properties by conducting a PCA using the vegan package in R (Oksanen et al. 2013); the soil variables tended to be highly correlated (Supplementary Fig. S1). The first principal component, named Soil PCA1, explained 92.24% of the total variance, and was negatively related to soil pH (pH), soil moisture content (Water; %), soil nitrogen content (Nitrogen; g/kg) and soil carbon content (C; g/kg), and positively related to the ammoniacal nitrogen content (NH4+; mg/kg) and nitrate nitrogen content (NO3−; mg/kg) (Supplementary Table S3).
We constructed a piecewise SEM (Lefcheck 2016) to determine how nitrogen addition (NH4NO3; g/m2) (Treatment) affected the relative abundance and number of fungal OTUs through a number of plausible pathways, such as effects on soil properties (Soil PCA1), total plant abundance (Abundance) and plant species richness (S) (see hypothetical causal piecewise SEM in Supplementary Fig. S2). To simplify the piecewise SEM, aboveground biomass (Babove), belowground biomass (Bbelow) and the first and second PCA axes (PCA1 and PCA2) were dropped, given their poor predictive power (Table 1). Shannon’s diversity index (H′), Pielou’s evenness index (J′) and SDI were also removed from the piecewise SEM, because: (i) these variables were highly correlated with plant species richness (S); and (ii) these variables were calculated based on plant species richness (S) and total abundance (Abundance).
General linear model results for soil fungal plant pathogen (a) relative abundance and (b) number of OTUs as a function of the nitrogen addition rate (NH4NO3; g/m2) (Treatment), soil pH (pH), soil moisture content (Water; %), ammoniacal nitrogen content (NH4+; mg/kg, log transformed), nitrate nitrogen content (NO3−; mg/kg), soil nitrogen content (Nitrogen; g/kg), soil carbon content (C; g/kg), plant species richness (S), Shannon’s diversity index (H′), Pielou’s evenness index (J′), Simpson’s diversity index (SDI), total plant abundance (Abundance), plant aboveground biomass (Babove), plant belowground biomass (Bbelow), first axis values from a principal component analysis (PCA) of plant community structure (PCA1) and second axis values from the same PCA (PCA2)
| Model | k | LL | AICc | ΔAICc | wAICc | De |
|---|---|---|---|---|---|---|
| (a) Soil fungal plant pathogen relative abundance | ||||||
| NH4+ | 2 | −45.986 | 99.172 | 0 | 0.998 | 66.97 |
| NO3− | 2 | −52.137 | 111.474 | 12.302 | 0.002 | 44.84 |
| Treatment | 2 | −55.344 | 117.888 | 18.716 | <0.001 | 27.95 |
| S | 2 | −56.371 | 119.943 | 20.771 | <0.001 | 21.51 |
| Bbelow | 2 | −57.756 | 122.712 | 23.540 | <0.001 | 11.91 |
| 1 | 1 | −59.277 | 123.126 | 23.954 | <0.001 | 0.00 |
| Abundance | 2 | −58.400 | 124.000 | 24.828 | <0.001 | 7.05 |
| pH | 2 | −58.429 | 124.057 | 24.886 | <0.001 | 6.83 |
| PCA2 | 2 | −58.667 | 124.535 | 25.363 | <0.001 | 4.96 |
| Water | 2 | −58.743 | 124.686 | 25.514 | <0.001 | 4.35 |
| C | 2 | −58.856 | 124.911 | 25.739 | <0.001 | 3.45 |
| H′ | 2 | −58.989 | 125.178 | 26.006 | <0.001 | 2.37 |
| J′ | 2 | −59.062 | 125.325 | 26.153 | <0.001 | 1.78 |
| PCA1 | 2 | −59.115 | 125.430 | 26.258 | <0.001 | 1.35 |
| Nitrogen | 2 | −59.160 | 125.521 | 26.349 | <0.001 | 0.97 |
| SDI | 2 | −59.258 | 125.715 | 26.544 | <0.001 | 0.16 |
| Babove | 2 | −59.262 | 125.725 | 26.553 | <0.001 | 0.12 |
| (b) Number of OTUs for soil fungal plant pathogen | ||||||
| SDI | 2 | −77.459 | 162.118 | 0 | 0.226 | 17.85 |
| H′ | 2 | −77.889 | 162.978 | 0.860 | 0.147 | 14.86 |
| S | 2 | −78.080 | 163.361 | 1.243 | 0.121 | 13.49 |
| Babove | 2 | −78.270 | 163.740 | 1.622 | 0.100 | 12.11 |
| 1 | 1 | −79.819 | 164.209 | 2.091 | 0.079 | 0.00 |
| J′ | 2 | −78.782 | 164.763 | 2.645 | 0.060 | 8.28 |
| Water | 2 | −79.420 | 166.040 | 3.922 | 0.032 | 3.27 |
| Abundance | 2 | −79.444 | 166.089 | 3.971 | 0.031 | 3.07 |
| PCA2 | 2 | −79.566 | 166.332 | 4.214 | 0.027 | 2.09 |
| NO3− | 2 | −79.741 | 166.682 | 4.564 | 0.023 | 0.65 |
| Treatment | 2 | −79.761 | 166.723 | 4.605 | 0.023 | 0.48 |
| PCA1 | 2 | −79.771 | 166.741 | 4.623 | 0.022 | 0.40 |
| Bbelow | 2 | −79.776 | 166.751 | 4.633 | 0.022 | 0.36 |
| pH | 2 | −79.808 | 166.816 | 4.698 | 0.022 | 0.09 |
| NH4+ | 2 | −79.810 | 166.821 | 4.703 | 0.022 | 0.07 |
| C | 2 | −79.818 | 166.837 | 4.719 | 0.021 | 0.00 |
| Nitrogen | 2 | −79.819 | 166.838 | 4.720 | 0.021 | 0.00 |
| Model | k | LL | AICc | ΔAICc | wAICc | De |
|---|---|---|---|---|---|---|
| (a) Soil fungal plant pathogen relative abundance | ||||||
| NH4+ | 2 | −45.986 | 99.172 | 0 | 0.998 | 66.97 |
| NO3− | 2 | −52.137 | 111.474 | 12.302 | 0.002 | 44.84 |
| Treatment | 2 | −55.344 | 117.888 | 18.716 | <0.001 | 27.95 |
| S | 2 | −56.371 | 119.943 | 20.771 | <0.001 | 21.51 |
| Bbelow | 2 | −57.756 | 122.712 | 23.540 | <0.001 | 11.91 |
| 1 | 1 | −59.277 | 123.126 | 23.954 | <0.001 | 0.00 |
| Abundance | 2 | −58.400 | 124.000 | 24.828 | <0.001 | 7.05 |
| pH | 2 | −58.429 | 124.057 | 24.886 | <0.001 | 6.83 |
| PCA2 | 2 | −58.667 | 124.535 | 25.363 | <0.001 | 4.96 |
| Water | 2 | −58.743 | 124.686 | 25.514 | <0.001 | 4.35 |
| C | 2 | −58.856 | 124.911 | 25.739 | <0.001 | 3.45 |
| H′ | 2 | −58.989 | 125.178 | 26.006 | <0.001 | 2.37 |
| J′ | 2 | −59.062 | 125.325 | 26.153 | <0.001 | 1.78 |
| PCA1 | 2 | −59.115 | 125.430 | 26.258 | <0.001 | 1.35 |
| Nitrogen | 2 | −59.160 | 125.521 | 26.349 | <0.001 | 0.97 |
| SDI | 2 | −59.258 | 125.715 | 26.544 | <0.001 | 0.16 |
| Babove | 2 | −59.262 | 125.725 | 26.553 | <0.001 | 0.12 |
| (b) Number of OTUs for soil fungal plant pathogen | ||||||
| SDI | 2 | −77.459 | 162.118 | 0 | 0.226 | 17.85 |
| H′ | 2 | −77.889 | 162.978 | 0.860 | 0.147 | 14.86 |
| S | 2 | −78.080 | 163.361 | 1.243 | 0.121 | 13.49 |
| Babove | 2 | −78.270 | 163.740 | 1.622 | 0.100 | 12.11 |
| 1 | 1 | −79.819 | 164.209 | 2.091 | 0.079 | 0.00 |
| J′ | 2 | −78.782 | 164.763 | 2.645 | 0.060 | 8.28 |
| Water | 2 | −79.420 | 166.040 | 3.922 | 0.032 | 3.27 |
| Abundance | 2 | −79.444 | 166.089 | 3.971 | 0.031 | 3.07 |
| PCA2 | 2 | −79.566 | 166.332 | 4.214 | 0.027 | 2.09 |
| NO3− | 2 | −79.741 | 166.682 | 4.564 | 0.023 | 0.65 |
| Treatment | 2 | −79.761 | 166.723 | 4.605 | 0.023 | 0.48 |
| PCA1 | 2 | −79.771 | 166.741 | 4.623 | 0.022 | 0.40 |
| Bbelow | 2 | −79.776 | 166.751 | 4.633 | 0.022 | 0.36 |
| pH | 2 | −79.808 | 166.816 | 4.698 | 0.022 | 0.09 |
| NH4+ | 2 | −79.810 | 166.821 | 4.703 | 0.022 | 0.07 |
| C | 2 | −79.818 | 166.837 | 4.719 | 0.021 | 0.00 |
| Nitrogen | 2 | −79.819 | 166.838 | 4.720 | 0.021 | 0.00 |
We included all single-predictor models and the intercept-only (null) model. Shown are the estimated number of model parameters (k), maximum log-likelihood (LL), the information-theoretic Akaike’s information criterion corrected for small samples (AICc), the change in AICc relative to the top-ranked model (ΔAICc), AICc weight (wAICc = model probability) and the percent deviance explained (De) as a measure of the model’s goodness-of-fit. Gray shading indicates models with strong support compared with the intercept-only (null) model (ER >1.5; where ER = wAICc [slope model]: wAICc [intercept-only model]).
General linear model results for soil fungal plant pathogen (a) relative abundance and (b) number of OTUs as a function of the nitrogen addition rate (NH4NO3; g/m2) (Treatment), soil pH (pH), soil moisture content (Water; %), ammoniacal nitrogen content (NH4+; mg/kg, log transformed), nitrate nitrogen content (NO3−; mg/kg), soil nitrogen content (Nitrogen; g/kg), soil carbon content (C; g/kg), plant species richness (S), Shannon’s diversity index (H′), Pielou’s evenness index (J′), Simpson’s diversity index (SDI), total plant abundance (Abundance), plant aboveground biomass (Babove), plant belowground biomass (Bbelow), first axis values from a principal component analysis (PCA) of plant community structure (PCA1) and second axis values from the same PCA (PCA2)
| Model | k | LL | AICc | ΔAICc | wAICc | De |
|---|---|---|---|---|---|---|
| (a) Soil fungal plant pathogen relative abundance | ||||||
| NH4+ | 2 | −45.986 | 99.172 | 0 | 0.998 | 66.97 |
| NO3− | 2 | −52.137 | 111.474 | 12.302 | 0.002 | 44.84 |
| Treatment | 2 | −55.344 | 117.888 | 18.716 | <0.001 | 27.95 |
| S | 2 | −56.371 | 119.943 | 20.771 | <0.001 | 21.51 |
| Bbelow | 2 | −57.756 | 122.712 | 23.540 | <0.001 | 11.91 |
| 1 | 1 | −59.277 | 123.126 | 23.954 | <0.001 | 0.00 |
| Abundance | 2 | −58.400 | 124.000 | 24.828 | <0.001 | 7.05 |
| pH | 2 | −58.429 | 124.057 | 24.886 | <0.001 | 6.83 |
| PCA2 | 2 | −58.667 | 124.535 | 25.363 | <0.001 | 4.96 |
| Water | 2 | −58.743 | 124.686 | 25.514 | <0.001 | 4.35 |
| C | 2 | −58.856 | 124.911 | 25.739 | <0.001 | 3.45 |
| H′ | 2 | −58.989 | 125.178 | 26.006 | <0.001 | 2.37 |
| J′ | 2 | −59.062 | 125.325 | 26.153 | <0.001 | 1.78 |
| PCA1 | 2 | −59.115 | 125.430 | 26.258 | <0.001 | 1.35 |
| Nitrogen | 2 | −59.160 | 125.521 | 26.349 | <0.001 | 0.97 |
| SDI | 2 | −59.258 | 125.715 | 26.544 | <0.001 | 0.16 |
| Babove | 2 | −59.262 | 125.725 | 26.553 | <0.001 | 0.12 |
| (b) Number of OTUs for soil fungal plant pathogen | ||||||
| SDI | 2 | −77.459 | 162.118 | 0 | 0.226 | 17.85 |
| H′ | 2 | −77.889 | 162.978 | 0.860 | 0.147 | 14.86 |
| S | 2 | −78.080 | 163.361 | 1.243 | 0.121 | 13.49 |
| Babove | 2 | −78.270 | 163.740 | 1.622 | 0.100 | 12.11 |
| 1 | 1 | −79.819 | 164.209 | 2.091 | 0.079 | 0.00 |
| J′ | 2 | −78.782 | 164.763 | 2.645 | 0.060 | 8.28 |
| Water | 2 | −79.420 | 166.040 | 3.922 | 0.032 | 3.27 |
| Abundance | 2 | −79.444 | 166.089 | 3.971 | 0.031 | 3.07 |
| PCA2 | 2 | −79.566 | 166.332 | 4.214 | 0.027 | 2.09 |
| NO3− | 2 | −79.741 | 166.682 | 4.564 | 0.023 | 0.65 |
| Treatment | 2 | −79.761 | 166.723 | 4.605 | 0.023 | 0.48 |
| PCA1 | 2 | −79.771 | 166.741 | 4.623 | 0.022 | 0.40 |
| Bbelow | 2 | −79.776 | 166.751 | 4.633 | 0.022 | 0.36 |
| pH | 2 | −79.808 | 166.816 | 4.698 | 0.022 | 0.09 |
| NH4+ | 2 | −79.810 | 166.821 | 4.703 | 0.022 | 0.07 |
| C | 2 | −79.818 | 166.837 | 4.719 | 0.021 | 0.00 |
| Nitrogen | 2 | −79.819 | 166.838 | 4.720 | 0.021 | 0.00 |
| Model | k | LL | AICc | ΔAICc | wAICc | De |
|---|---|---|---|---|---|---|
| (a) Soil fungal plant pathogen relative abundance | ||||||
| NH4+ | 2 | −45.986 | 99.172 | 0 | 0.998 | 66.97 |
| NO3− | 2 | −52.137 | 111.474 | 12.302 | 0.002 | 44.84 |
| Treatment | 2 | −55.344 | 117.888 | 18.716 | <0.001 | 27.95 |
| S | 2 | −56.371 | 119.943 | 20.771 | <0.001 | 21.51 |
| Bbelow | 2 | −57.756 | 122.712 | 23.540 | <0.001 | 11.91 |
| 1 | 1 | −59.277 | 123.126 | 23.954 | <0.001 | 0.00 |
| Abundance | 2 | −58.400 | 124.000 | 24.828 | <0.001 | 7.05 |
| pH | 2 | −58.429 | 124.057 | 24.886 | <0.001 | 6.83 |
| PCA2 | 2 | −58.667 | 124.535 | 25.363 | <0.001 | 4.96 |
| Water | 2 | −58.743 | 124.686 | 25.514 | <0.001 | 4.35 |
| C | 2 | −58.856 | 124.911 | 25.739 | <0.001 | 3.45 |
| H′ | 2 | −58.989 | 125.178 | 26.006 | <0.001 | 2.37 |
| J′ | 2 | −59.062 | 125.325 | 26.153 | <0.001 | 1.78 |
| PCA1 | 2 | −59.115 | 125.430 | 26.258 | <0.001 | 1.35 |
| Nitrogen | 2 | −59.160 | 125.521 | 26.349 | <0.001 | 0.97 |
| SDI | 2 | −59.258 | 125.715 | 26.544 | <0.001 | 0.16 |
| Babove | 2 | −59.262 | 125.725 | 26.553 | <0.001 | 0.12 |
| (b) Number of OTUs for soil fungal plant pathogen | ||||||
| SDI | 2 | −77.459 | 162.118 | 0 | 0.226 | 17.85 |
| H′ | 2 | −77.889 | 162.978 | 0.860 | 0.147 | 14.86 |
| S | 2 | −78.080 | 163.361 | 1.243 | 0.121 | 13.49 |
| Babove | 2 | −78.270 | 163.740 | 1.622 | 0.100 | 12.11 |
| 1 | 1 | −79.819 | 164.209 | 2.091 | 0.079 | 0.00 |
| J′ | 2 | −78.782 | 164.763 | 2.645 | 0.060 | 8.28 |
| Water | 2 | −79.420 | 166.040 | 3.922 | 0.032 | 3.27 |
| Abundance | 2 | −79.444 | 166.089 | 3.971 | 0.031 | 3.07 |
| PCA2 | 2 | −79.566 | 166.332 | 4.214 | 0.027 | 2.09 |
| NO3− | 2 | −79.741 | 166.682 | 4.564 | 0.023 | 0.65 |
| Treatment | 2 | −79.761 | 166.723 | 4.605 | 0.023 | 0.48 |
| PCA1 | 2 | −79.771 | 166.741 | 4.623 | 0.022 | 0.40 |
| Bbelow | 2 | −79.776 | 166.751 | 4.633 | 0.022 | 0.36 |
| pH | 2 | −79.808 | 166.816 | 4.698 | 0.022 | 0.09 |
| NH4+ | 2 | −79.810 | 166.821 | 4.703 | 0.022 | 0.07 |
| C | 2 | −79.818 | 166.837 | 4.719 | 0.021 | 0.00 |
| Nitrogen | 2 | −79.819 | 166.838 | 4.720 | 0.021 | 0.00 |
We included all single-predictor models and the intercept-only (null) model. Shown are the estimated number of model parameters (k), maximum log-likelihood (LL), the information-theoretic Akaike’s information criterion corrected for small samples (AICc), the change in AICc relative to the top-ranked model (ΔAICc), AICc weight (wAICc = model probability) and the percent deviance explained (De) as a measure of the model’s goodness-of-fit. Gray shading indicates models with strong support compared with the intercept-only (null) model (ER >1.5; where ER = wAICc [slope model]: wAICc [intercept-only model]).
Before fitting the piecewise SEM, we checked the normality of the variables, confirming the use of a ‘gaussian’ family with a series of component linear models (Laforest-Lapointe et al. 2017). We calculated the standardized path coefficients (scaled by their mean and standard deviation) and corresponding significance (P values) for each path in the piecewise SEM. We evaluated the overall fit of both full and final models using Fisher’s C statistic and AICc in the R package piecewiseSEM. In addition, we made partial regression plots to test the relationships between the residuals of the relative abundance/number of fungal OTUs and the residuals of Soil PCA1/S. All statistical analyses were performed using R V2.15.1 (R Development Core Team 2015).
RESULTS
Factors influencing soil fungal guilds
The concave relationship between the amount of nitrogen added and plant species richness was best described by a second-order polynomial model (AICc = 142.784, ER = 64.09, De = 43.86) rather than a simple linear model (AICc = 144.410) (Fig. 1a; Supplementary Table S2). Plant species richness decreased as supplemental nitrogen increased from 0 to 10 g/m2, then slightly increased at a concentration of 15 g/m2 (Fig. 1a). The number of fungal OTUs (across all plots) increased linearly with the nitrogen addition rate (ER = 2.56, De = 17.12) (Fig. 1b).
The relationship between the nitrogen addition rate (NH4NO3; g/m2) (Treatment) and (a) plant species richness (information-theoretic ER = 64.09; percent De = 43.86) and (b) the number of soil fungal OTUs (ER = 2.56; De = 17.12).
Among the 151 fungal OTUs, 61 showed a negative relationship between the nitrogen addition rate and their relative abundance, while 80 showed a positive relationship and ten could not be tested (Supplementary Table S4). Nitrogen addition (Treatment) increased the relative abundance of fungal plant pathogens in the soil (ER = 13.72, De = 27.95) (Fig. 2a), while a negative relationship was detected between plant species richness (S) and the relative abundance of pathogens (ER = 4.91, De = 21.51) (Fig. 2c). Among the tested general linear models for predicting pathogen relative abundance, the model with soil ammoniacal nitrogen (NH4+) was the best predictor (AICc = 99.172, wAICc = 0.998, De = 66.97), explaining up to 66.96 % variation in relative pathogen abundance. The second best model included nitrate nitrogen content (NO3−) (AICc = 111.474, wAICc = 0.002, De = 44.84) (Table 1a), indicating that the effects of nitrogen addition on soil properties drove changes in the relative abundance of plant pathogens under treatment.
Linear relationships between measures of soil fungal plant pathogen relative abundance and richness, and explanatory variables such as the treatment rate (NH4NO3; g/m2), ammoniacal nitrogen content (NH4+; mg/kg) and plant species richness (S) in a nitrogen addition experiment (n = 24 plots). (a) Soil fungal plant pathogen relative abundance versus treatment rate (g/m2; information-theoretic ER = 13.72; percent De = 27.95); (b) soil fungal plant pathogen relative abundance versus ammoniacal nitrogen content (mg/kg, log transformed; ER = 1.59 × 105; De = 66.97); (c) soil fungal plant pathogen relative abundance versus plant species richness (ER = 4.91; De = 21.51); (d) soil fungal plant pathogen OTU number versus treatment rate (g/m2; ER = 0.28; De = 0.48); (e) soil fungal plant pathogen OTU number versus ammoniacal nitrogen content (mg/kg, log transformed; ER = 0.95; De = 0.07); (f) soil fungal plant pathogen OTU number versus plant species richness (ER = 1.53; De = 13.49).
Although the nitrogen addition treatment had no effect on the number of fungal OTUs (Fig. 2d), there was a positive relationship between the number of OTUs and SDI (AICc = 162.118, wAICc = 0.226, De = 17.85), Shannon’s diversity index (H′) (AICc = 162.978, wAICc = 0.147, De = 14.86) and plant species richness (S) (AICc = 163.361, wAICc = 0.121, De = 13.49) (Table 1b). Overall, in contrast to the situation with pathogen relative abundance, plant community variables rather than soil variables explained more of the variance in the number of fungal pathogen OTUs (Table 1b; Fig. 2). In addition, the relative abundance and number of pathotrophic fungal OTUs consistently showed the same patterns as did the fungal plant pathogens (Supplementary Table S5 and Fig. S3), as plant pathogens accounted for the majority of soil pathotrophic fungi.
Nitrogen addition did not affect the relative abundance or OTU number of the soil saprotrophic fungi (Supplementary Fig. S4 and Table S6a). Only total plant abundance (Abundance) (AICc = 135.597, wAICc = 0.205, De = 15.14) provided a better fit than the intercept-only (null) model (AICc = 136.909, wAICc = 0.107, De = 0.00) in explaining the relative abundance of soil saprotrophic fungi (Supplementary Table S6a). However, both SDI (AICc = 259.156, wAICc = 0.285, De = 20.52) and Shannon’s diversity index (H′) (AICc = 259.954, wAICc = 0.191, De = 17.83) were related to the number of soil saprotrophic fungal OTUs (Supplementary Table S6b), indicating that plant diversity might play an important role in driving soil fungal diversity.
Nitrogen addition led to a decrease in the relative abundance of soil symbiotrophic fungi (ER = 9.50, De = 13.28) (Supplementary Fig. S5a), while a positive relationship was detected between plant species richness (S) and the relative abundance of plant pathogens (ER = 4.91, De = 25.67) (Supplementary Fig. S5c). Moreover, plant species richness (S) was selected as the best single predictor for the relative abundance of symbiotrophic fungi among the general linear models (Supplementary Table S7a). However, the number of symbiotrophic fungal OTUs was not affected by nitrogen addition and only weakly affected by plant species richness (S) (AICc = 185.053, wAICc = 0.133, De = 12.06) (Supplementary Fig. S5 and Table S7b).
Structural equation model
The final piecewise SEM (standardized path coefficients are given in Supplementary Table S8), which adequately fitted the data (Fisher’s C = 7.69, df = 6, P = 0.262; AICc = 1063.69), explained ~49% of the variance in the relative abundance of fungal plant pathogens (De = 0.487) and ~44% of the variance in the number of fungal pathogen OTUs (De = 0.439) (Fig. 3). In the final piecewise SEM, Soil PCA1 increased with the nitrogen addition rate (standardized path coefficient β = 0.837, P < 0.001), while plant species richness decreased (S) (β = −0.567, P = 0.004); total plant abundance was not affected. Although nitrogen addition affected both Soil PCA1 and S, Soil PCA1 (β = 0.799, P = 0.024), rather than S, was positively correlated with the relative abundance of fungal plant pathogens. In contrast, S (β = 0.517, P = 0.041), but not Soil PCA1, drove the number of fungal plant pathogen OTUs, as confirmed by the partial regression plots (Fig. 4). The results of the final piecewise SEM and the general linear model selection were qualitatively similar, further confirming that nitrogen addition increased the relative abundance of soil fungal plant pathogens by altering soil properties, while decreasing soil pathogen richness by decreasing plant species diversity.
Results of a final piecewise SEM. The final model adequately fit the data: Fisher’s C = 7.69 (df = 6, P = 0.262) and AICc = 1063.69. Numbers on arrows are standardized path coefficients (scaled by their mean and standard deviation), while asterisks indicate statistical significance (***P < 0.001; **P < 0.01; *P < 0.05). Solid red arrows indicate evidence for positive relationships, solid blue arrows evidence for negative relationships and gray arrows insufficient statistical evidence for a path coefficient (P > 0.05). Arrow width is indicative of the strength of the causal relationship, and De stands for the percent deviance explained (i.e. how much of the variance is explained by predictive variables in the model).
Partial regression plots visualizing piecewise SEM output from the final SEM for the effects of PCA1 (from a PCA of soil properties, Soil PCA1) and plant species richness on the relative abundance of soil fungal plant pathogens (a, b) and the number of fungal pathogen OTUs (c, d).
DISCUSSION
Here, we integrate information on soil properties, the plant community and the relative abundance and richness of soil fungal plant pathogens (i.e. number of OTUs) in a 7-year nitrogen addition experiment in an alpine meadow on the Qinghai-Tibetan Plateau. Our study provides critical empirical evidence that soil fungal plant pathogen communities are structured by various soil properties and characteristics of the plant community under nitrogen addition. Specifically, we conclude that nitrogen addition treatments led to shifts in soil properties, altering the relative abundance of soil fungal plant pathogens, and plant species losses that reduced their richness.
The relative abundance of soil fungal plant pathogens in response to nitrogen addition
We attribute the increase in soil fungal plant pathogen relative abundance (under nitrogen addition) to the following three plausible mechanisms: (i) the soil microenvironment became more suitable for pathogens but not for soil symbiotrophic fungi (Jiang et al. 2018); (ii) plant tissues became more palatable, thus harboring more pathogens (Mitchell et al. 2003) and (iii) the plant community composition shifted toward fast-growing species, ultimately increasing pathogen relative abundance. Strong growth-defense trade-offs are seen across plant species, not only on the Qinghai-Tibetan Plateau where our study site was located (Liu et al. 2017), but also in other ecosystems such as old fields (Lind et al. 2013) and Swiss meadows (Cappelli et al. 2020).
The tight link between plant disease susceptibility and tissue nutrition suggests that fungal plant pathogens may be sensitive to changes in soil properties (Anderson 2002; Veresoglou et al. 2014). As the ‘nitrogen disease hypothesis’ suggests, nitrogen additions can increase susceptibility to pathogens, because greater soil nitrogen availability leads to higher plant tissue nitrogen content, and nitrogen is a limiting element for fungal pathogens (Mitchell et al. 2003). When nitrogen becomes abundant, the fungal mycelium grows and fungal fitness should increase (Huber and Watson 1974). In contrast, many studies from both natural (Liu et al. 2012) and agro-ecosystems (Zhu et al. 2018) have found that increasing soil nitrogen availability reduces the abundance and root colonization rate of symbiotrophic fungi (especially Glomeromycota). In our study, there was a negative relationship between soil plant fungal pathogens and symbiotrophic fungi; therefore, the empty niche left by symbiotrophic fungi following nitrogen addition may be occupied by soil plant fungal pathogens, eventually leading to an increase in pathogen relative abundance.
Shifts in plant community composition in response to nitrogen addition may indirectly contribute to changes in the relative abundance of soil plant fungal pathogens. Plant species vary in their ability to harbor pathogens (Fisher et al. 2012). Fast-growing plant species with resource-acquisitive traits, such as high plant tissue nitrogen, metabolic rate and specific leaf area, usually attract more fungal pathogens (Semchenko et al. 2019). Many empirical studies (e.g. Blumenthal et al. 2009; Coley 1987; Liu et al. 2017), and also a meta-analysis (Endara and Coley 2011), support the resource availability hypothesis, which predicts that fast-growing plant species (i.e. weaker nutrient competitors that benefit most from nutrient addition; sensu Endara and Coley 2011) will invest less in defense to herbivory and/or diseases. Such a growth-defense trade-off is expected to maintain plant diversity by preventing competitive exclusion (Mordecai 2011), and is thought to be the main diver of foliar fungal pathogen relative abundance in a Swiss meadow (Cappelli et al. 2020). In a previous study, we found that plant species that were less susceptible to foliar fungal pathogens were most likely to be lost in fertilized plots; thus, the community weighted mean of disease susceptibility increased following nitrogen addition (Liu et al. 2017), suggesting a growth-defense trade-off among plant species coexisting in the alpine meadow. Such a change in plant community composition toward fast-growing plant species with resource-acquisitive traits may increase the relative abundance of fungal pathogens indirectly, independent of any direct effects of nitrogen addition through soil nitrogen availability.
Plant species richness mediated soil fungal plant pathogen relative abundance and richness
There was a slightly negative relationship between plant species richness and soil fungal pathogen relative abundance, in line with accumulating evidence for dilution effects of host plant diversity on disease (Liu et al. 2020a; Ostfeld and Keesing 2012) and negative plant–soil feedback theory (Bever 2003; Collins et al. 2020). A dilution effect can arise by susceptible-host regulation in speciose communities with dozens of host plant species (Liu et al. 2016). Following nitrogen addition, plant species richness may be lost and the remaining plant species show compensatory increases in abundance (Liu et al. 2017); this would increase search costs for pathogens to find a suitable host individual, leading to a decrease in their abundance. Overall, these findings suggest that more diverse plant communities have lower fungal pathogen relative abundance, which highlights the important role of soil pathogens in maintaining plant species coexistence (Chen et al. 2020; Mordecai 2011) and biodiversity–ecosystem functioning relationships (Chen et al. 2019).
Plant species richness promotes fungal pathogen diversity. Broad evidence from both plants and animals suggests that the diversity of higher trophic levels (e.g. soil plant fungal pathogens) strongly depends on the species richness of lower trophic levels (Kamiya et al. 2014); host availability and evolutionary conservation (e.g. coevolution between hosts and parasites) are thought to be the main factors shaping this relationship (Gilbert and Parker 2016). Host plant species richness generally translates into habitat diversity, thus supporting diverse pathogens. Previous studies from both artificial plant communities (The Jena Experiment, Germany) (Rottstock et al. 2014) and our study site (Liu et al. 2016) found that foliar fungal pathogen richness increases linearly with plant species richness. Foliar fungal pathogens and soil fungal plant pathogens are similar in terms of phylogeny (mainly Ascomycota and Basidiomycota), indicating that the ‘host–parasite diversity’ relationship for foliar fungal pathogens likely also holds for soil fungal plant pathogens.
There are several limitations to our study: (i) we only sampled soils in 2017, but a time series of soil samples might better elucidate the causal relationships between soil fungal pathogen relative abundance/richness and soil properties/plant community characteristics. For the fertilization treatments, NH4NO3 addition significantly altered overall soil properties; however, some treatment levels did not differ significantly in NH4+/NO3− (e.g. for NO3−, there was no significant difference between the 0 and 5 g/m2 treatments), owing to spatial heterogeneity within the alpine meadow. (ii) Many soil plant pathogens belong to the oomycetes (Makiola et al. 2019), but the primers we used (ITS1F and ITS1R) are not specific for oomycetes. Therefore, this study only shows the relationship between soil properties/plant community characteristics and fungal pathogens, rather than total soil pathogens (including pathogenic oomycetes/bacteria). (iii) Methodologically, our results were based on the relative abundance of fungal plant pathogens rather than the absolute abundance (e.g. qPCR). Additionally, we determined the functional group of each fungal OTU at the genus level using only the FunGuild algorithm. Addressing these limitations in future studies will allow for a more complete picture of soil fungal plant pathogen dynamics.
CONCLUSIONS
Our study demonstrates that nitrogen addition increases soil fungal plant pathogen relative abundance mainly by altering soil properties, while decreasing fungal pathogen richness as a result of plant species losses in an alpine meadow. Most pathogens show host specificity (Gilbert and Webb 2007; Gilbert et al. 2012). Nitrogen addition alters the phylogenetic structure of host plant communities (Yang et al. 2018). Hence, soil fungal pathogens may accumulate at a higher rate under nitrogen addition as compared with natural conditions, an example of the ‘host–parasite diversity’ relationship (Kamiya et al. 2014). The increase in soil fungal plant pathogen relative abundance may translate into higher colonization rates (of plant tissues from pathogens in the soil), influencing plant growth and survival. These results suggest that continuous nitrogen inputs (both from fertilizer use and nitrogen deposition) may seriously threaten ecosystem health, and hence the provision of ecosystem services, by increasing the relative abundance of fungal pathogens. We also provide new empirical evidence for a positive ‘host–parasite diversity’ relationship for soil fungal pathogens under nitrogen addition. This study expands our knowledge of several key topics in community and disease ecology, including dilution effects, the nitrogen disease hypothesis, plant–soil feedbacks and host–parasite diversity relationships. Our results help form a unified framework to understand host–pathogen interactions in a community context.
Supplementary Material
Supplementary material is available at Journal of Plant Ecology online.
Table S1: The differences in variables (mean ± standard error) between treatments.
Table S2: Model selection results for predictive relationships between the nitrogen addition rate (NH4NO3; g/m2) (Treatment) and (a) plant species richness; (b) soil fungal OTUs.
Table S3: Principal component analysis (PCA) characterizing soil properties in a nitrogen addition experiment with 24 plots.
Table S4: Linear models for the relationship between the nitrogen addition rate (NH4NO3; g/m2) (Treatment) and the relative abundance of each soil fungal pathogen across the 24 study plots.
Table S5: General linear model results for (a) relative abundance and (b) the number of OTUs of soil pathotrophic fungi.
Table S6: General linear model results for (a) relative abundance and (b) the number of OTUs of soil saprotrophic fungi.
Table S7: General linear model results for (a) relative abundance and (b) the number of OTUs of soil symbiotrophic fungi.
Table S8: Coefficient estimates from the final piecewise structural equation model relating how the nitrogen addition rate (NH4NO3; g/m2) (Treatment) affects the relative abundance of soil fungal pathogens and the number of fungal OTUs.
Figure S1: Correlation matrix of variables.
Figure S2: Hypothesized effects of the nitrogen addition rate (NH4NO3; g/m2) (Treatment) on the relative abundance of soil fungal plant pathogens and the number of fungal OTUs.
Figure S3: Linear relationships between the relative abundance or number of OTUs of soil pathotrophic fungi and the nitrogen treatment level (NH4NO3; g/m2), ammoniacal nitrogen content (NH4+; mg/kg) and plant species richness (S) in a nitrogen addition experiment (n = 24 plots).
Figure S4: Linear relationships between the relative abundance and number of OTUs of soil saprotrophic fungi and the nitrogen treatment level (NH4NO3; g/m2), ammoniacal nitrogen content (NH4+; mg/kg) and plant species richness (S) in a nitrogen addition experiment (n = 24 plots).
Figure S5: Linear relationships between the relative abundance and number of OTUs of soil symbiotrophic fungi and the nitrogen treatment level (NH4NO3; g/m2), ammoniacal nitrogen content (NH4+; mg/kg) and plant species richness (S) in a nitrogen addition experiment (n = 24 plots).
Funding
This study was supported by the National Natural Science Foundation of China (31830009 and 31770518 to S.Z., 32001116 to X.L.), a Fundamental Research Fund for Central Universities (lzujbky-2020-cd01 to X.L.) and start-up funds for Introduced Talent at Lanzhou University (561119211 to X.L.).
Acknowledgements
This work was conducted at the Research Station of Alpine Meadow and Wetland Ecosystems of Lanzhou University. We thank Shengman Lyu, Dexin Sun and Fei Chen from Fudan University for assistance in the field, and Dr Emily Drummond from the University of British Columbia for English language editing.
Conflict of interest statement. The authors declare that they have no conflict of interest.
