Soil Bacterial Community Response and Nitrogen Cycling Variations Associated with Subalpine Meadow Degradation on the Loess Plateau, China

Soil microbes play a crucial role in the biogeochemical cycles of grassland ecosystems, yet information on how their community structure and functional characteristics change with subalpine meadow degradation is scarce. In this study, we evaluated the changes in bacterial community structure and nitrogen functional genes in degraded meadow soils. Meadow degradation had a significant effect on bacterial community composition. Soil total nitrogen was the best predictor of bacterial community structure. The beta diversities of the plant and soil bacterial communities were significantly correlated, while their alpha diversities were only weakly correlated. Meadow degradation decreased the potential for nitrogen fixation and denitrification but increased the potential for nitrification. These results have implications for the restoration and reconstruction of subalpine meadow ecosystem on the Loess Plateau.

ABSTRACT

Grassland degradation is an ecological problem worldwide. This study aimed to reveal the patterns of the variations in bacterial diversity and community structure and in nitrogen cycling functional genes along a subalpine meadow degradation gradient on the Loess Plateau, China. Meadow degradation had a significant effect on the beta diversity of soil bacterial communities (P < 0.05) but not on the alpha diversity (P > 0.05). Nonmetric multidimensional scaling (NMDS) and analysis of similarity (ANOSIM) indicated that the compositions of bacterial and plant communities changed remarkably with increasing meadow degradation (all P < 0.05). The beta diversities of the plant and soil bacterial communities were significantly correlated (P < 0.05), while their alpha diversities were weakly correlated (P > 0.05) along the meadow degradation gradient. Redundancy analysis (RDA) showed that the structure of the bacterial community was strongly correlated with total nitrogen (TN), nitrate nitrogen (NO₃⁻-N), plant Shannon diversity, plant coverage, and soil bulk density (all P < 0.05). Moreover, the abundances of N fixation and denitrification genes of the bacterial community decreased along the degradation gradient, but the abundance of nitrification genes increased along the gradient. The structure of the set of N cycling genes present at each site was more sensitive to subalpine meadow degradation than the structure of the total bacterial community. Our findings revealed compositional shifts in the plant and bacterial communities and in the abundances of key N cycling genes as well as the potential drivers of these shifts under different degrees of subalpine meadow degradation.

IMPORTANCE Soil microbes play a crucial role in the biogeochemical cycles of grassland ecosystems, yet information on how their community structure and functional characteristics change with subalpine meadow degradation is scarce. In this study, we evaluated the changes in bacterial community structure and nitrogen functional genes in degraded meadow soils. Meadow degradation had a significant effect on bacterial community composition. Soil total nitrogen was the best predictor of bacterial community structure. The beta diversities of the plant and soil bacterial communities were significantly correlated, while their alpha diversities were only weakly correlated. Meadow degradation decreased the potential for nitrogen fixation and denitrification but increased the potential for nitrification. These results have implications for the restoration and reconstruction of subalpine meadow ecosystem on the Loess Plateau.

INTRODUCTION

The Loess Plateau in China experiences very serious soil erosion and is one of the most fragile ecological environments in the world. More than 60% of the land is eroded by natural (topography, climate, soil, and vegetation) and human factors (1). Soil erosion, mainly caused by grassland degradation, has seriously damaged the land resources and ecosystem of the Loess Plateau. At present, the grassland degradation area has reached 14 million hm², accounting for approximately 22% of the total area of this region, and the grassland quality and function have gradually decreased (1). Grassland degradation seriously threatens the local ecological environment and biodiversity and hinders the sustainable development of animal husbandry in this region. Subalpine meadow is a kind of grassland that is widely distributed in the high-altitude regions of the Loess Plateau (2). As an important part of this terrestrial ecosystem, subalpine meadows provide important ecosystem services, such as regulating the climate, maintaining biodiversity, preventing soil erosion, providing forage for grazing livestock, and regulating the nutrient cycles (2–5). However, the subalpine meadow grasslands on the Loess Plateau in Shanxi, China, are undergoing severe degradation mainly due to climate change and long-term unsustainable human activities (1, 5).

Previous studies have mainly focused on the impacts of grassland degradation on soil fertility (4, 6), plant composition, biodiversity (7), and above- and belowground productivity (4). Researchers tend to pay more attention to the succession dynamics of plant communities during grassland ecosystem degradation (6, 7), and the change patterns of plant communities are often used as indicators of grassland degradation (8). Soil microbes may also be affected by grassland degradation. The turnover of microbial communities is similar to that of plants and may be restricted by available resources, abiotic environmental conditions, biological interactions, and historical contingencies (9). However, the differences in key biological attributes between plants and microorganisms may lead to their different responses to grassland degradation. The influence of vegetation litter deposition and root exudates on microbial communities indicates that plant attributes may be the key drivers of microbial turnover during grassland degradation (10), and the diversity of aboveground plants may also be closely related to the diversity of soil microbial communities (11). Therefore, one of our concerns is how soil microbial diversity, structure, and function respond to meadow degradation in subalpine meadows where soil properties and plant community structure have been significantly changed.

Soil microbial communities, as the core of the soil biogeochemical cycle, are involved in the regulation of ecosystem processes, and their diversity, structure, and function are very sensitive to disturbance (12, 13). Soil substrate availability (14), enzyme activity (15), plant attributes (16), and environmental heterogeneity (17) are important factors affecting the structure and function of soil microbial communities. At present, there are relatively few studies on the impacts of grassland degradation on soil microbial communities. Those that exist mainly focus on the alpine meadow ecosystems of the Qinghai-Tibet Plateau (11, 18), while little is known about the effects of subalpine meadow degradation on the soil microbial community on the Loess Plateau. These studies on the Qinghai-Tibet Plateau mainly focused on the taxonomic diversity of microbial communities, which cannot accurately predict the functional characteristics of microorganisms. Studies on the changes in microbial communities in disturbed ecosystems have shown that microbial communities change gradually with changes in environmental conditions that are caused by disturbances, which may also affect ecosystem functions (12, 19). An increasing number of studies have shown that microbial functional traits and taxonomy respond differently to environmental disturbance (20, 21). The results from Louca and Doebeli (22) showed that microbial communities with phage infection exhibited remarkably stable functional community structure despite large variation in taxonomic composition. In contrast, Ma et al. (20) reported a high level of sensitivity with respect to changes in functional traits (functionally categorized taxa or genes) in response to mild environmental disturbance.

Nitrogen (N) is widely recognized as a limiting factor for plant growth that affects microbial communities in terrestrial ecosystems, and increasing attention is being paid to the impacts of climate change and human disturbance on N cycling in grasslands (19, 22, 23). Soil microorganisms play a vital role in soil N cycling and participate in important ecological processes such as N fixation, ammonification, nitrification, and denitrification (12, 24). Some key nitrogen functional genes (NFGs), such as nifH for N fixation, ammonia-oxidizing archaeon (AOA) and ammonia-oxidizing bacterium (AOB) genes (amoA) for nitrification, and narG, nirS, nirK, and nosZ genes for denitrification, have been widely used to evaluate microbial contributions to different stages of the N cycle (23, 25, 26). Measurements of NFGs using quantitative PCR (qPCR) provide an effective tool for assessing the contribution of microorganisms to different stages of the N cycle. Studying the changes in the abundance of NFGs during meadow degradation can provide direct insights into the biological potential for N cycling to occur, including process rates, substrate availability, and the population density of functional groups of microorganisms (27). Therefore, it is important to evaluate how bacterial community structure and NFGs change, to compare taxonomic and N cycling functional traits along the meadow degradation gradient, and to link these results to abiotic and biotic properties.

Our study site, Mount Wutai, located in the northeastern margin of the Loess Plateau, has a relatively intact alpine and subalpine natural meadow ecosystem. The subalpine meadow grassland at Mount Wutai is one of the largest alpine summer pastures in north China, with an area of 106,993 ha. The annual grazing period is concentrated in 70 to 90 days from June to August. However, in recent years, the subalpine meadows of Mount Wutai have suffered from severe degradation, mainly caused by unsustainable grazing systems. In the summer grazing period, to facilitate stocking management, a large number of livestock are concentrated together at high density, resulting in the overloading of local meadows, serious vegetation destruction, and the formation of degraded patches and even bare land in some areas (3, 28, 29). It is estimated that the meadows with various degrees of degradation occupy approximately three-fifths of the area of all meadows of Mount Wutai (5). Subalpine meadows with different degrees of degradation, i.e., nondegraded (ND), lightly degraded (LD), moderately degraded (MD), and heavily degraded (HD) meadows, of Mount Wutai (Fig. 1) were selected to examine changes in soil bacterial community structure and function by using Illumina MiSeq sequencing of the 16S rRNA gene and qPCR measuring of NFGs. We aimed to investigate (i) the responses of plant and microbial communities to the degradation of subalpine meadows and whether the changes in the structure and biodiversity of the plant and microorganism communities would be consistent, (ii) the responses and sensitivity of the N cycling functional genes of the microbial community to meadow degradation, as well as microbial taxonomic structure and biodiversity, and (iii) the drivers of the responses of the plant and microbial communities and functional genes to meadow degradation.

FIG 1.

Map showing the location of Mount Wutai and the sampling region. ND, nondegraded meadow; LD, lightly degraded meadow; MD, moderately degraded meadow; HD, heavily degraded meadow.

RESULTS

Soil and plant variables along the meadow degradation gradient.

The soil properties and plant variables in the degraded meadows are presented in Table 1. Ammonium nitrogen (NH₄⁺-N), pH, soil bulk density, and sand content increased with the increasing degradation level, while total carbon (TC), soil water content (SWC), soil organic matter (SOM), plant coverage, plant height, aboveground plant biomass (AGB), and soil clay and silt contents decreased as the degradation level increased (Table 1). SWC, TC, total nitrogen (TN), and soil clay and silt contents significantly decreased (P < 0.05) in the MD and HD meadows compared to those in ND meadow, and no significant differences (P > 0.05) were observed between the TC and TN contents of the LD and ND meadows. Plant coverage, plant height, and AGB decreased significantly (P < 0.05) with the increasing degree of degradation (Table 1). C/N, SWC, soil clay, and silt in the LD, MD, and HD meadows were significantly lower (P < 0.05) than those in the ND meadow. Electrical conductivity (EC), nitrite nitrogen (NO₂⁻-N), and available phosphorus (AP) were not significantly different (P > 0.05) among the different meadow degradation levels (Table 1). Along the meadow degradation gradient, the plant species richness and plant Shannon diversity in the LD and MD meadows increased significantly (P < 0.05), and those in the HD meadow decreased significantly (P < 0.05) (Table 1). The nonmetric multidimensional scaling (NMDS) and analysis of similarity (ANOSIM) results indicated that meadow degradation significantly changed the overall plant compositions (see Fig. S1 in the supplemental material) (P < 0.05). Soil organic matter (SOM), pH, and soil bulk density were confirmed as the drivers for the structural variation in the plant communities (see Fig. S2) (P < 0.05).

TABLE 1.

Plant and soil physicochemical properties in the sampling sites

Parametera	Value (mean ± SE)b
	ND	LD	MD	HD
SWC (%)	39.17 ± 0.76 A	31.46 ± 0.90 B	21.59 ± 0.91 C	20.79 ± 2.49 C
Bulk density (g/cm3)	1.08 ± 0.03 D	1.15 ± 0.06 C	1.22 ± 0.06 B	1.35 ± 0.07 A
Content (%)
Clay (<2 mm)	17.22 ± 0.92 A	13.20 ± 1.19 B	11.38 ± 0.86 B	9.03 ± 0.39 C
Silt (2–50 mm)	49.34 ± 1.30 A	43.37 ± 1.69 B	41.45 ± 1.20 B	36.90 ± 0.70 C
Sand (>50 mm)	33.43 ± 1.30 C	43.43 ± 1.77 B	47.17 ± 1.68 B	53.07 ± 0.38 A
EC (μS/cm)	95.0 ± 3.21 A	127.8 ± 6.77 A	120.0 ± 16.23 A	104.0 ± 20.46 A
pH	6.87 ± 0.12 C	7.12 ± 0.12 B	7.32 ± 0.04 AB	7.46 ± 0.03 A
TN (%)	0.46 ± 0.02 A	0.56 ± 0.01 A	0.31 ± 0.06 B	0.30 ± 0.03 B
TC (%)	6.69 ± 0.24 A	5.72 ± 0.33 A	3.58 ± 0.58 B	3.08 ± 0.28 B
C/N	14.55 ± 0.63 A	10.21 ± 0.78 B	11.76 ± 0.75 B	10.50 ± 0.41 B
SOM (g/kg)	155.96 ± 10.04 A	114.20 ± 2.48 B	81.38 ± 14.94 C	43.67 ± 8.19 D
NO3−-N (mg/kg)	34.22 ± 4.38 B	21.63 ± 1.79 B	77.68 ± 19.03 A	52.46 ± 12.74 AB
NO2−-N (mg/kg)	14.76 ± 1.66 A	12.63 ± 1.79 A	10.92 ± 2.09 A	10.48 ± 1.31 A
NH4+-N (mg/kg)	2.06 ± 0.19 C	2.80 ± 0.17 B	2.85 ± 0.13 B	3.55 ± 0.19 A
AP (mg/kg)	13.16 ± 0.69 A	9.80 ± 0.90 A	11.28 ± 1.45 A	11.32 ± 1.92 A
AK (mg/kg)	326.60 ± 75.99 A	158.60 ± 11.59 B	267.00 ± 16.15 AB	328.84 ± 31.54 A
Plant coverage (%)	98.80 ± 0.49 A	87.20 ± 1.00 B	65.00 ± 2.24 C	47.00 ± 2.00 D
Plant ht (cm)	36.40 ± 1.33 A	21.60 ± 1.47 B	13.20 ± 0.73 C	5.2 ± 0.73 D
AGB (g/m2)	389.63 ± 23.16 A	275.49 ± 16.54 B	142.30 ± 6.74 C	80.12 ± 3.95 D
Plant richness	18.60 ± 1.60 B	22.40 ± 0.81 A	21.80 ± 0.58 A	12.80 ± 0.37 C
Plant diversity	1.72 ± 0.06 B	1.97 ± 0.06 A	1.99 ± 0.04 A	1.59 ± 0.06 B
Main plant species	Kobresia pygmaea, Thalictrum alpinum, Polygonum viviparum	Polygonum viviparum, Deschampsia cespitosa, Kobresia pygmaea	Plantago depressa, Puccinellia distans, Taraxacum platypecidium	Potentilla anserina, Taraxacum platypecidium, Plantago depressa

^aSWC, soil water content; TC, total carbon; SOM, soil organic matter; TN, total nitrogen; AP, available phosphate; AK, available potassium; EC, electrical conductivity; AGB, aboveground biomass.

^bValues are the means (± standard errors). Different uppercase letters indicate significant differences with a P value of <0.05 based on the analysis of variance.

Responses of bacterial composition and diversity to subalpine meadow degradation.

Forty-three phyla, 102 classes, 193 orders, 366 families, and 672 genera were identified in the survey area. The dominant bacterial phyla (relative abundance of >1%), in descending order of relative abundance, were Proteobacteria (27.76%), Actinobacteria (24.06%), Acidobacteria (19.14%), Chloroflexi (12.71%), Bacteroidetes (3.39%), Firmicutes (3.32%), Gemmatimonadetes (2.57%), Nitrospirae (2.05%), Verrucomicrobia (1.58%), and Parcubacteria (1.05%). These dominant phyla composed more than 97% of the bacterial sequences at each meadow degradation level (Fig. 2). Nitrospirae, Actinobacteria, Gemmatimonadetes, and Parcubacteria exhibited significantly different relative abundances among the four degradation levels (all P < 0.05) (see Fig. S3). The relative abundances of Nitrospirae (4.12%) and Parcubacteria (1.06%) in the LD meadow were the highest; Gemmatimonadetes (3.45%) was the most abundant in the MD meadow, and Actinobacteria (27.62%) was the most abundant in the ND meadow (Fig. S3). At the class, family, and genus levels, the comparisons of the relative abundances of the 12 most abundant bacterial classes, families, or genera showed differences in community composition among the four degradation levels (see Fig. S4 to S6). There were significant shifts in the relative abundances of seven classes (Nitrospirae, Chloroflexia, Betaproteobacteria, Gammaproteobacteria, Sphingobacteriia, Actinobacteria, and Gemmatimonadetes), four families (norank_c_Nitrospira, Sphingomonadaceae, Nitrosomonadaceae, and Gemmatimonadaceae), and five genera (Nitrospira, Sphingomonas, Nitrosospira, norank_c_Actinobacteria, and Bacillus) along the degradation gradient (all P < 0.05) (Fig. S3 to S6).

FIG 2.

Relative abundances of the dominant bacterial phyla (with average relative abundance of >1%) in nondegraded (ND), lightly degraded (LD), moderately degraded (MD), and heavily degraded (HD) meadows.

Meadow degradation had no significant effect on the alpha diversity index or the species richness of the soil bacterial communities (all P > 0.05) (Fig. 3), but it had a significant impact on beta diversity (Fig. 4A). The ANOSIM test indicated that the overall compositions of bacterial communities were significantly (P < 0.05) separated across this degradation gradient (Fig. 4A and Table S2). Except MD and HD meadows (P > 0.05), the bacterial community compositions of any two degraded levels of meadows were significantly separated (P < 0.05) (Table S2). Furthermore, we estimated the differences in the beta diversity of bacterial communities among the different degraded meadows based on Bray-Curtis distance (Fig. 4B). Compared with that in the ND meadow, the LD meadow exhibited no significant change in bacterial beta diversity (P > 0.05), whereas MD and HD meadows had significantly increased bacterial beta diversity (P < 0.05) (Fig. 4B).

FIG 3.

The species richness (A) and Shannon-Wiener indices (B) in nondegraded (ND), lightly degraded (LD), moderately degraded (MD), and heavily degraded (HD) meadows. Data that do not share a lowercase letter are significantly different (P < 0.05). Error bars represent standard errors (n = 5).

FIG 4.

General patterns of bacterial beta diversity along the degradation gradient. (A) NMDS showed the structure of soil bacteria. Similarity values among different meadows were examined via the ANOSIM test, and these are shown in the plot (r = 0.421, P < 0.001). (B) Differences in bacterial beta diversity among the treatments were estimated based on a Bray-Curtis distance matrix of all 20 soil samples. Data are means ± standard errors (SEs). Data that do not share a lowercase letter are significantly different (P < 0.05). ND, nondegraded meadow; LD, lightly degraded meadow; MD, moderately degraded meadow; HD, heavily degraded meadow.

Relationship between shifts in bacterial community composition and environmental variables.

Redundancy analysis (RDA) was conducted to identify the drivers of the response of the bacterial community to meadow degradation (Fig. 5). The first two axes of the RDA axis explained 43.92% of the total variation. Among the environmental variables, TN, NO₃⁻-N, plant Shannon diversity, plant coverage, and soil bulk density were confirmed as significant predictors of the structural variation in bacterial communities (P < 0.05) (Fig. 5A). Among all the correlated environmental factors, TN accounted for the largest proportion (27.7%) of the total variation in bacterial communities, followed by NO₃⁻-N (8.1%) and plant Shannon diversity (8.0%) (see Table S3). The variation partition analysis (VPA) results showed that the selected environmental variables explained 54.3% of the bacterial community composition variations (Fig. 5B). Among them, soil chemical properties independently explained the largest fraction of the variation (20.7%). Plant variables and soil physical properties independently explained 15.5% and 2.8% of the variation, respectively (Fig. 5B). Linear regression analysis revealed that the bacterial alpha diversity was independent of the plant alpha diversity (all P > 0.05) (Fig. 6), while the composition of plant and bacterial communities was significantly correlated (P < 0.05) (Fig. 7).

FIG 5.

RDA and VPA of the soil bacterial community structure and environmental variables. (A) RDA of bacterial communities along the degradation meadow gradient on Mount Wutai. (B) VPA of the effects of soil chemical properties (soil chem), soil physical properties (soil phy), and plant variables (plant) on the bacterial community composition. The data present percentages of variation explained by the variables.

FIG 6.

Relationships between plant diversity and soil bacterial diversity in different degradation sites. (A) Species richness; (B) Shannon-Wiener indices.

FIG 7.

Relationship between plant community Bray-Curtis distances and soil bacterial community Bray-Curtis distances, measured at plot scale across different degradation sites.

The variation in N cycling functional characteristics in the degraded subalpine meadow.

Similar to that for the soil properties and plant variables, the abundance of NFGs measured in the four degraded levels of meadows varied significantly with degradation (Fig. 8). With increasing degradation intensity, the numbers of AOA amoA and AOB amoA copies increased significantly (P < 0.05) (Fig. 8B and C), whereas the numbers of nifH, nirS, nirK, and nosZ copies decreased significantly (P < 0.05) (Fig. 8). In this study, nifH gene abundance varied from 3.33 × 10⁵ to 3.73 × 10⁵ copies g⁻¹ of dry soil, and there were no statistically significant differences (all P > 0.05) between the ND meadow and the other three degraded meadows. However, the nifH gene abundance significantly decreased in the HD meadow compared with that in the ND meadow (P < 0.05) (Fig. 8A). The abundance of the nitrifier groups (AOA and AOB) was estimated by quantifying their respective amoA gene copy numbers (Fig. 8B and C). The AOA amoA gene copy numbers increased with increasing degradation, ranging from 4.31 × 10⁶ to 5.74 × 10⁶ copies g⁻¹ of dry soil. The abundance of AOA amoA genes in the HD meadow was significantly higher than that in the ND meadow (P < 0.05), and there was no significant change in the abundance of AOA amoA genes in the LD and MD meadows compared with that in the ND meadow (both P > 0.05) (Fig. 8B). The abundance of the AOB amoA gene ranged from 3.14 × 10⁵ to 4.22 × 10⁵ copies g⁻¹ of dry soil and was significantly higher in the HD meadow than in the ND and MD meadows (all P < 0.05) (Fig. 8C). Although the abundances of AOA and AOB increased with increasing meadow degradation, the AOB/AOA ratio remained constant (7.27%, 8.47%,7.55%, and 7.35% in ND, LD, MD, and HD meadows, respectively). Three functional genes associated with denitrification (nirS, nirK, and nosZ) were quantified in this study (Fig. 8D to F). The abundance of the nirS gene (1.17 × 10⁶ to 6.23 × 10⁶ copies g⁻¹ of dry soil) predominated over nirK gene abundance (6.26 × 10⁴ to 1.79 × 10⁵ copies g⁻¹ of dry soil) independent of meadow type, even though their functions are the same (Fig. 8D and E). Compared to that in the ND meadow, the nirS gene abundance first significantly increased in the LD meadow (P < 0.05) and then significantly decreased in the MD and HD meadows (all P < 0.05) (Fig. 8D). The nirK gene abundance was significantly different among the four degradation levels (all P < 0.05) (Fig. 8E). However, the abundance of the nosZ gene was significantly lower in the MD and HD meadows (all P < 0.05) and unchanged in the LD meadow (P > 0.05) (Fig. 8F) compared with that in the ND meadow. The NMDS and ANOSIM results indicated that meadow degradation clearly changed the overall composition of the N cycling functional genes (Fig. 9) (see Table S4). Pairwise ANOSIM analysis revealed significant (all P < 0.05) separation of the N cycling functional taxon compositions between any two degradation sites (Table S4).

FIG 8.

Gene copy numbers of functional genes involved in major steps of the nitrogen cycle (nifH, AOA amoA, AOB amoA, nirK, nirS, and nosZ). Some key N-related functional genes (NFGs) including nifH for N fixation, ammonia-oxidizing archaeon (AOA), and ammonia-oxidizing bacterium (AOB) genes (amoA) for nitrification, and nirS, nirK, and nosZ genes for denitrification. Data that do not share a lowercase letter are significantly different (P < 0.05). Error bars represent standard errors (n = 5). R values represent Spearman’s correlation coefficients between nitrogen cycle gene copy numbers and meadows with different degrees of degradation. ND, nondegraded meadow; LD, lightly degraded meadow; MD, moderately degraded meadow; HD, heavily degraded meadow.

FIG 9.

Distribution of nitrogen cycle gene abundance based on nonmetric multidimensional scaling (NMDS). Similarity values among different meadows were examined via the ANOSIM test, and these are shown in the plot (r = 0.905, P < 0.001). ND, nondegraded meadow; LD, lightly degraded meadow; MD, moderately degraded meadow; HD, heavily degraded meadow.

Stepwise regression analysis showed that there was a significant correlation between environmental variables and the abundances of NFGs (all P < 0.05) (Table 2). The abundance of the nifH gene was significantly positively correlated with soil pH and SWC, which together accounted for 56.7% of the variation (Table 2). Ammonium nitrogen (NH₄⁺-N), plant richness, and C/N together accounted for 87.7% of the variation in the abundance of the AOA amoA gene, and SOM accounted for 59.9% of the variation in the abundance of the AOB amoA gene (Table 2). SOM and AGB together explained 89.8% of the total variation in the abundance of the nirK gene, while TN, AP, and available potassium (AK) explained 86.3% of the variation of in the nirS group (Table 2). In addition, the abundance of nosZ genes was significantly correlated with TC and pH, which explained 69.4% of the variation (Table 2).

TABLE 2.

Environmental variables that are significantly correlated with the abundance of N cycling functional genes

Modela	P value	R2
nifH = −6.38E5 × pH + 1.60E2 × SWC − 1.45E5	0.001	0.567
AOA amoA = 2.38E6 + 9.06E5 × NH4+-N − 6.70E4 × plant richness + 9.11E4 × C/N	<0.0001	0.877
AOB amoA = 4.43E5 − 8.82E2 × SOM	0.001	0.599
nirS = 2.06E6 + 1.21E7 × TN − 2.26E5 × AP − 5.57E3 × AK	<0.0001	0.863
nirK = 5.61E2 × SOM + 1.69E2 × AGB + 2.86E4	<0.0001	0.898
nosZ =6.00E4 × TC + 1.19E5 × pH − 5.16E5	<0.0001	0.694

^aSOM, soil organic matter; SWC, soil water content; TN, total nitrogen; TC, total carbon; AP, available phosphate; AK, available potassium; AGB, aboveground biomass.

DISCUSSION

Bacterial taxonomic shifts along the meadow degradation gradient.

Changes in the soil bacterial communities across the different meadow degradation levels were apparent at both low and high taxonomic levels, from phylum to genus (see Fig. S3 to S6 in the supplemental material), suggesting that important taxonomic changes occurred throughout the bacterial community and supporting the theory that at least some degree of ecological consistency existed across broad phylogenetic groups (30, 31). The relative abundances of four phyla, seven classes, four families, and five genera were significantly different among the four degradation levels (Fig. S3 to S6). Nitrospirae (phylum), Nitrosomonadaceae (family), Nitrospira (genus), and Nitrosospira (genus) were more abundant in the LD meadow than in the other meadows (Fig. S3 to S6). All known AOB belong to the Betaproteobacteria and Gammaproteobacteria, and they are mostly represented by the genera Nitrosomonas, Nitrosococcus, and Nitrosospira (32). Nitrospira spp. are chemolithotrophic nitrifying bacteria that oxidize nitrite to nitrate. Nitrospira are perceived as the main participants in nitrite-oxidizing bacterial nitrification (oxidizing nitrite to nitrate) in terrestrial ecosystems (19). Recently, the ability to oxidize ammonia has also been found in members of the genus Nitrospira, which can oxidize ammonia all the way to nitrate and contain ammonia monooxygenase and hydroxylamine oxidoreductase (33, 34). Nitrosomonadaceae (family) comprise a monophyletic phylogenetic group within the Betaproteobacteria, include two genera (Nitrosomonas and Nitrosospira), and generally exert control over nitrification by oxidizing ammonia to nitrite (35). We also found that the relative abundances of the genera Sphingomonas and Bacillus in the degenerated MD and HD meadows were significantly higher than in the ND and LD meadows (Fig. S6). Sphingomonas has been isolated from many different terrestrial and aquatic habitats, can survive at low concentrations of nutrients, and can metabolize various carbon sources (36). Numerous strains of Sphingomonas have nitrogen fixation and denitrification abilities (36), and these strains play an important role in maintaining the nitrogen balance in nature. The genus Bacillus has demonstrated the potential for dissimilatory reduction of nitrogen compounds in 45 of 87 investigated species, with 19 species showing denitrifying abilities (37). This indicates that denitrification is a common characteristic of Bacillus. Therefore, the enrichment of these genera (Nitrospira, Nitrosospira, Sphingomonas, Bacillus, etc.) in some degraded meadows further suggested that there were significant changes in soil bacterial communities under degradation.

The alpha diversity patterns differ between plant and soil bacterial communities during meadow degradation.

The present study indicated that the vegetation composition, the plant species richness, the plant diversity, and the aboveground biomass showed significant changes along the meadow degradation gradient. With the increase in grassland degradation, aboveground biomass and plant coverage decreased, and primary plant communities were gradually replaced by secondary plant communities dominated by forbs. It was also found that the richness and diversity of the plant community were the highest in the stages of light and moderate degradation (Table 1) and showed a humped variation pattern along the degradation gradient. These results are similar to those of previous studies on degraded alpine grasslands and meadows (6, 10) and are in agreement with the intermediate disturbance hypothesis. This hypothesis suggests that moderate environmental disturbance or an adverse environment is beneficial for increasing biodiversity (38). Moderate disturbances increase environmental heterogeneity and lead to a greater diversity of plant species (10). In this study, the soil characteristics in the LD and MD meadows were changed due to the destruction of the original vegetation. This may have provided favorable conditions for exotic species to invade and grow, eventually increasing the plant diversity.

In contrast to the effects on plant communities, our results suggested that meadow degradation had no significant effect on the alpha diversity of the soil bacterial communities. This is consistent with the result of Che et al. (18), who found that the formation of degraded patches on the Qinghai-Tibet Plateau had no significant effect on prokaryotic alpha diversity. However, Li et al. (11) reported that, compared with that in ND meadows, severely degraded (SD) meadows had significantly increased soil bacterial Shannon-Wiener index and species richness in the alpine meadows on the Qinghai-Tibet Plateau, while the alpha diversity of the MD meadow showed no significant difference. Zhang et al. (39) found that MD significantly increased, but SD significantly decreased, the soil bacterial Shannon-Wiener index and richness. This inconsistency might be related to the disruption to soil conditions, plant species, and other environmental factors caused by degradation in these areas. Our results indicated that the alpha diversity of soil bacterial and plant communities showed distinct patterns during meadow degradation, and these correlations were not significant (Fig. 6). Thus, our results most strongly support Wardle’s conclusion (40) that the alpha diversities of soil and plants are largely uncoupled. This lack of a significant relationship may be due to differences in the biogeographic factors driving soil bacterial and plant alpha diversities, which mask potential effects on plant richness under degradation disturbance. Bardgett et al. (41) also found that disturbance events may lead to changes in plant alpha diversity that are not synchronous with changes in microbial alpha diversity. In our analysis, meadow degradation only significantly affected plant species richness and alpha diversity but did not significantly affect bacterial taxonomic richness or alpha diversity, indicating that plant communities and bacterial communities change at different rates during meadow degradation.

Plant beta diversity is correlated with the beta diversity of the soil bacterial community during meadow degradation.

Although the soil bacterial alpha diversity indices did not change, the compositions of the soil bacterial community changed as expected. In this study, bacterial community structure and composition, which were closely related to soil and vegetation properties (Fig. 5; Table S3), were distinctly different among the four meadows with different degrees of degradation (Fig. 4; Table S2). This indicated that the structure of the bacterial community is sensitive to the degradation of subalpine meadows. Additionally, we found greater variation in the beta diversity of the soil bacterial communities in the MD and HD meadows than in the ND meadow, which could be related to the fact that meadow degradation enhanced variations in soil and vegetation properties. The results of the NMDS and ANOSIM also confirmed that meadow degradation significantly influenced the structures of the plant communities (Fig. S1). Remarkably, across the meadow degradation gradient, the plant alpha diversity patterns were independent of those of the soil bacterial communities (P > 0.05), but there was a significant correlation between plant beta diversity (composition difference between sites) and bacterial beta diversity (P < 0.05) (Fig. 7). This is consistent with the research of Prober et al. (42), that is, plant diversity can predict the beta diversity of soil microorganisms but not the alpha diversity. The strong coupling of plant-bacterial beta diversity indicated that the composition of plant communities has an important influence on the soil bacterial communities of degraded meadows in Mount Wutai. This finding confirms previous work showing that the members of plant and soil microbial communities can interact with each other (42). Soil microorganisms are the key factor affecting the decomposition of plant litter, and their community composition depends to a certain extent on the properties of the vegetation. Different vegetation properties produce distinct quantities and qualities of plant litter, which mainly contribute to the changes in soil carbon and nitrogen and eventually lead to changes in the bacterial community composition (43). In addition, aboveground vegetation appears to play an important role in shaping the microbial community, since plants can select certain soil microorganisms to become associated with (44, 45).

Accordingly, our RDA results also showed significant relationships between bacterial community composition and plant coverage and plant Shannon diversity (Fig. 5). Furthermore, as a shared environmental driver of plant and bacterial beta diversity variables, bulk density contributed to the relationship between soil bacterial and plant beta diversities (Table S3; Fig. S2). A significant increase in bulk density leads to a change in soil porosity, which affects the soil permeability to air (46) and thus affects the distributions of both plants and soil bacteria. However, not all environmental factors associated with beta diversity were shared between plant and bacterial communities. TN, NO₃⁻-N, plant Shannon diversity, plant coverage, and soil bulk density shaped the bacterial community composition, whereas soil bulk density, pH, and SOM were the best predictors for the plant community. This suggests that the relationship between soil microbial and plant beta diversities may be weakened by the different responses of soil microbial and plant communities to environmental drivers during meadow degradation (42).

Meadow degradation decreased the potential for nitrogen fixation and denitrification but increased the potential for nitrification.

The absolute measurements of nitrogen cycle-related functional genes are reported per gram of soil, similar to the measurements of specific metabolic activity (i.e., activity per unit protein content) that are commonly used when reporting enzyme activities (15). As meadow degradation increases, if the abundance of a specific nitrogen cycle-related functional gene decreases, this may indicate a general decrease in metabolic activity as available substrates and energy become increasingly scarce or vice versa (47, 48). By measuring the abundance of NFGs, we were able to analyze the trend of bacterial community investment in major nitrogen cycle processes. Nitrogen enters terrestrial ecosystems largely through biological fixation. This process is performed by bacteria that are estimated to be responsible for more than 95% of the N input in natural ecosystems (32, 49, 50). The presence of the nifH genes of Alphaproteobacteria (Bradyrhizobium, Azospirillum, and Gluconacetobacter), Betaproteobacteria (Burkholderia), and Deltaproteobacteria (Anaeromyxobacter and Geobacter) was observed in different degraded levels of soil (24, 50). In this study, the abundance of the nifH gene substantially declined with meadow degradation (Fig. 8A), which is consistent with the results from Ding et al. (23) in Inner Mongolia grasslands but contrary to the findings of Yang et al. (12) in Tibetan alpine grasslands. This may be due to the similar climatic conditions of the Loess Plateau and Inner Mongolia Plateau, which have temperate arid and semiarid climates. Moreover, we observed that nifH gene abundance was significantly correlated with soil pH during meadow degradation. In addition, soil pH explained 50.3% of the variance in nifH gene abundance (Table 2), suggesting that soil pH played a key role in the nitrogen fixation process in the degraded meadows. We speculated that this was because acidic soil would inhibit dinitrogenase reductase, thereby reducing the abundance of the nifH gene (51). Our study was supported by the results from Ding et al. (23), showing that the abundance of the nifH gene was directly affected by soil pH in grasslands degraded by grazing.

Our results revealed that the abundances of N nitrification genes (AOA amoA and AOB amoA) increased with increasing degrees of degradation and that the AOB/AOA ratio remained constant (Fig. 8B and C). This was consistent with the results of other grassland soil studies (19, 52), suggesting that meadow degradation increased the abundances of AOA and AOB to similar extents (Fig. 8B and C) and enhanced the functional potential of the nitrifier groups. The abundance of AOA amoA was 1 order of magnitude higher than that of AOB amoA in all meadows, and only the abundance of AOA amoA was significantly positively correlated with the ammonium content, which indicated that AOA amoA was a more important nitrifier than AOB amoA in this study. This was in line with the results of previous studies that showed that AOA amoA had a broader range of habitats and might be the predominant ammonia-oxidizing population in most natural environments (47, 53). Our results also indicated that environmental variables might play a key role in affecting the relative abundances of AOA amoA and AOB amoA. NH₄⁺-N, plant richness, and C/N together explained 87.7% of the variance in AOA amoA gene abundance (Table 2). Pan et al. (19) found that NH₄⁺-N contents had the best individual correlation with the composition of nitrifier distribution in a grassland degraded by grazing. AOA amoA genes were greatly affected by plant richness, and such a pattern indicates that AOA might be more susceptible to the different root exudates from microbe-specific plant defense and growth responses (54). Our results indicated that SOM was the environmental factor that explained the largest proportion of the variation in AOB amoA gene abundance. The mineralization of nitrogen from the pool of SOM supplied NH₄⁺-N, a kind of substrate used for nitrification by ammonia-oxidizing bacteria (55).

In the denitrification process, the abundances of nirK, nirS, and nosZ decreased with increasing degrees of degradation (Fig. 8D to F). The abundance of nirS was 1 to 2 orders of magnitude greater than that of nirK in the degraded subalpine meadow, which suggests that nirS-harboring denitrifiers were more important than nirK-harboring denitrifiers in the degraded subalpine meadow soil. A dominance of nirS over nirK in different soils has also been reported in previous studies (23, 56). In general, the abundances of denitrification genes tended to decrease with increasing degrees of degradation, suggesting that the functional potentials of relevant bioprocesses were reduced in degraded subalpine meadows. Commonly, nirS and nirK are used as gene markers for denitrifiers; however, these genes are present in many other microorganisms, including anaerobic ammonium-oxidizing bacteria, nitrite- and methane-oxidizing bacteria, and ammonia-oxidizing bacteria and archaea (24). Denitrifying bacteria are abundant and widespread in grassland soils, and denitrification genes have been found in bacterial strains belonging to Acidobacteria, Proteobacteria, and Firmicutes, as well as in other bacterial phyla (24, 32).

Stepwise regression analysis illustrated that total C, AP, and AK were significantly correlated with nirS abundances (P < 0.05), and SOM and AGB had the strongest correlations with nirK gene abundance (P < 0.05). TC and pH were thought to be the most important factors affecting the abundance of nosZ denitrifier genes. These results indicated that the variations observed in denitrification functional genes in the process of meadow degradation are likely to be due to the effects of soil properties, and the functional changes might influence nutrient cycle processes (57). Consistent with the findings of Huhe et al. (54), we found that the nirK gene was also affected by plant parameters, indicating that nirK might be more readily affected by root exudate amounts and composition. Therefore, we speculated that resource availability (e.g., SOM, TC, and TN), pH, and vegetation parameters changed significantly along the meadow degradation gradient, which might have an important effect on the distribution of denitrifiers.

The increase in nitrification potential and decrease in denitrification potential were interesting, which might influence the balance between NH₄⁺-N and NO₃⁻-N and, consequently, the N conservation at the sites. This shift could also affect N₂O emissions. Therefore, we infer that in degraded subalpine meadows, soil nutrition and vegetation gradually change with degradation intensity; thus, different requirements develop for growth-limiting factors such as the availability of nitrogen. In turn, these fluctuations in demand affect the microbial community, particularly those involved in N cycling.

Nitrogen cycle-related functional traits are sensitive indicators of subalpine meadow degradation.

The results indicated that meadow degradation clearly changed the overall composition of the bacterial community and the abundances of N cycling functional genes (all P < 0.001). Similar to that in previous studies showing the coupling of microbial phylogeny and functional compositions (58, 59), we found an association between them (r = 0.219, P = 0.023). The change in the functional genes profile (r = 0.905, P < 0.001) was more obvious than in the taxonomic structure of the bacterial community (r = 0.421, P < 0.001) along the meadow degradation gradient (r = 0.905, P < 0.001). The pairwise ANOSIM analysis revealed significant (all P < 0.05) separation in the N cycling functional gene profile between any two degradation gradient sites (Table S4), while the compositions of the bacterial communities in any two sites were significantly separated, except those of the MD and HD sites (P > 0.05) (Table S2). These results indicated that community functional gene profiles were more sensitive to meadow degradation than bacterial taxonomy. As a taxonomic marker, the 16S rRNA gene is ubiquitous in bacteria, and its relatively high conservatism makes it difficult to detect small variations in the sequences; in contrast, most functional genes are not as highly conserved as the 16S rRNA gene. In addition, environmental disturbances tend to have a greater impact on functional traits and a weaker impact on microbial taxonomy (21, 60), which was also observed in our study. By associating the functional traits of microorganisms with the environment, the predictability of ecological studies can be improved, as taxonomically similar communities could be functionally distinct due to functionally divergent evolution.

Quantitative PCR used to examine the abundances of N cycling functional genes may reflect the N metabolism potential of the microbial community (61, 62) but does not necessarily reflect the N metabolism of the actual microbial populations. Furthermore, the results of qPCR depend on the effectiveness of the primer set used to generate sequence data to analyze the characteristics of microbes. Even though it may be sufficient to observe the relative changes in the taxa targeted by a given primer set, it is possible to miss significant portions of all taxa (62). Therefore, further analysis of the relationship between rates of ecosystem processes and functional microorganisms using different approaches, such as omics (metagenomics, metatranscriptomics, and metabolomics) and stable isotope probing, is needed (63). This will provide a deeper understanding of the relationship between the functional microorganisms and N transformation processes during the process of subalpine meadow degradation in the future.

Conclusions.

In this study, we demonstrated that subalpine meadow degradation affected soil properties and vegetation parameters and led to significant changes in the bacterial community and the abundances of N cycling genes. Moreover, plant diversity can predict patterns in the composition of soil bacterial communities but cannot predict patterns in alpha diversity. It is indicated that the plant and bacterial communities respond to the degradation of subalpine meadows to different degrees. The abundance of nitrogen fixation and denitrification genes decreased, while that of nitrification genes increased, indicating that meadow degradation altered microbe-mediated N cycling. Meadow degradation also clearly changed the overall composition of the N cycling functional gene profile, but the structure of the N cycling genes was more sensitive to subalpine meadow degradation than that of the bacterial community. These results provide solid and important information for a better understanding of the responses of soil microbial communities and nitrogen cycling potential to grassland degradation in subalpine meadows.

MATERIALS AND METHODS

Study site description.

This study was conducted at Mount Wutai (112°48′ to 113°55′E; 38°27′ to 39°15′N) located on the northeastern margin of the Loess Plateau, Shanxi Province, China (Fig. 1). The local climate is characterized by strong solar radiation with long cold winters and short cool summers. The annual average temperature is 4.2°C, and the annual average precipitation is 650 mm. The precipitation occurs mainly from June to September.

Four subalpine meadow sites with different degrees of degradation (nondegraded [ND], lightly degraded [LD], moderately degraded [MD], and heavily degraded [HD]) were selected. The meadow degradation was classified based on plant coverage, species dominance, aboveground biomass, and the proportion of edible plants (64) (see Table S1 in the supplemental material). The area of each degraded meadow site was 100 m by 100 m, and the maximum distance between sites was not more than 500 m. The soil types were all subalpine meadow soil. The ND meadow is dominated by Kobresia pygmaea (Carex parvula). Other prevalent grasses include Thalictrum alpinum and Polygonum viviparum (Bistorta vivipara). The dominant vegetation is gradually replaced by Plantago depressa and Potentilla anserina during degradation (Table S1).

Soil sampling and vegetation survey.

On 3 to 6 August 2018, we adopted a random sampling method to ensure representative sampling from each meadow degradation site. Five sampling plots (1 m by 1 m) with a distance of more than 50 m between plots were randomly selected from the patches of meadow at each degradation level. Therefore, a total of 20 plots were selected for sampling. The 5 surface soil core (0 to 10 cm) samples (five-point sampling method) from each plot were mixed into a single sample. Soil samples were then passed through a 2-mm mesh screen to remove most roots, animals, and stones. One subsample was preserved at −80°C for molecular analysis, and the other subsample was air dried for physiochemical analysis. The measurement methods for soil and vegetation properties are described in the supplemental material.

DNA extraction, MiSeq sequencing, and bioinformatics analysis.

The procedures for the molecular analysis of microbial DNA from soil samples were the same as those described by Luo et al. (65). Briefly, DNA was extracted with an E.Z.N.A. R Soil DNA kit (Omega Bio-tek, Norcross, GA, USA). The V3 to V4 hypervariable region of the bacterial 16S rRNA gene was amplified using the universal primers 338F and 806R. High-throughput sequencing of PCR amplification products was performed on an Illumina MiSeq platform (Majorbio Bio-pharm Technology Co., Ltd., Shanghai, China). After removing chimeric sequences, the remaining sequences were assigned to operational taxonomic units (OTUs) at similarities of 97% using UPARSE (66). OTU representative sequences were classified taxonomically with BLAST against the Silva 128 16S rRNA database, using 0.8 confidence values as the cutoff (67). We obtained 1,118,879 high-quality reads from 20 soil samples and identified 4,340 bacterial OTUs. Bacterial alpha and beta diversities were calculated at the same sequencing depth (random resampling was performed at a depth of 36,150 sequences per sample).

Quantitative PCR.

N cycling genes (nifH, AOA amoA, AOB amoA, nirS, nirK, and nosZ) were quantified in triplicates by qPCR using a CFX96 system (Bio-Rad, San Diego, CA, USA). The qPCR mixture contained 2 μl of template DNA, 12.5 μl of 2× SYBR green qPCR Master Mix (TaKaRa, Dalian, China), and 1 μl of primer (Table 3) in a total 25-ml reaction volume. A melting curve analysis was performed following each assay to ensure the specificity of amplification. The reaction efficiencies of qPCRs were 81.5% for nifH, 86.6% for AOA amoA, 89.4% for AOB amoA, 83.7% for nirS, 79.8% for nirK, and 79.2% for nosZ. The abundances of all functional genes were finally calculated as copy number per gram of dry soil.

TABLE 3.

PCR primers used for quantitative real-time PCR

Target	Primer	Primer sequence (5′→3′)	Amplicon length (bp)	Annealing		Elongation		Reference(s)
				Temp (°C)	Time (s)	Temp (°C)	Time (s)
nifH	nifH-F	AAAGGYGGWATCGGYAARTCCACCAC	458	55	60	72	59	23, 50
	nifH-R	TTGTTSGCSGCRTACATSGCCATCAT
AOA amoA	amoAF	STAATGGTCTGGCTTAGACG	635	55	60	72	45	23, 69
	amoAR	GCGGCCATCCATCTGTATGT
AOB amoA	amoA-1F	GGGGTTTCTACTGGTGGT	491	57	30	72	45	22, 69
	amoA-2R	CCCCTCKGSAAAGCCTTCTTC
nirS	cd3aF	AACGYSAAGGARACSGG	420	58	30	72	30	70
	3cdR	GASTTCGGRTGSGTCTTSAYGAA
nirK	nirK1F	GGMATGGTKCCSTGGCA	515	58	30	72	30	71
	nirK5R	GCCTCGATCAGRTTRTGG
nosZ	nosZ-F	CGYTGTTCMTCGACAGCCAG	454	60	30	72	30	72
	nosZ-R	CGSACCTTSTTGCCSTYGCG

Statistical analyses.

The alpha diversity index (species richness and Shannon-Wiener index) and beta diversity (based on Bray-Curtis dissimilarity) of the bacterial and plant communities were calculated in the R package Vegan (68). Nonmetric multidimensional scaling (NMDS) and analysis of similarity (ANOSIM) were used to test the differences in the compositions of the bacterial communities or in NFGs among different meadows. Redundancy analysis (RDA) was performed to assess the correlation between the structures of the bacterial or plant communities and the environmental variables. Before RDA, stepwise regression and the Monte Carlo permutation test (CANOCO for Windows Version 5.0) were performed for the forward selection of environment variables that were statistically significant (P < 0.05). To better understand the influences of soil physical properties, soil chemical properties, and plant variables on the bacterial community structure, we used the selected variables for variation partition analysis (VPA). A model based on stepwise regression was created to predict the relationship between environmental variables and NFGs. A confidence interval of 95% (P < 0.05) was considered to be statistically significant, and statistical analyses and Pearson correlation tests were performed using SPSS 20.0 (IBM SPSS statistics, Chicago, IL, USA).

Accession number(s).

The obtained original bacterial sequences were deposited in the NCBI BioProject database (accession no. SRP193701).