The experiment was conducted at the Gaolan Ecology and Agriculture Research Station, of the Northwest Institute of Eco-Environment and Resources, Chinese Academy of Sciences. This station lies in the northwestern part of the Loess Plateau (Gaolan County, Lanzhou City, Gansu Province: 36°13′′N, 103°47′′E), at an elevation of 1800 m a.s.l. Mean annual rainfall is 327 mm, of which nearly 70% comes between June and September, and the mean annual air temperature is 8.4°C, with monthly minimum and maximum means of −9.1°C (January) and 20.7°C (July). Here, the zonal soil types are derived from ash–calcium and loess parent material, and the predominant zonal vegetation is desert grassland.

To study the key factors affecting soil microbial changes under gravel mulching in the Loess Plateau, three different treatments were established in 2015, in triplicate, consisting of soil mulched with small-sized gravel (2–5 mm dimension, AG), large-sized gravel (40–60 mm dimension, BG), and the control treatment with no mulching (CK). All nine replicate plots (each 3 m × 5 m) were randomly distributed. Gravel samples of different sizes, consisting mainly of silica, were obtained from fluvial material from the Yellow River. All plots had an approximately 100% surface cover of gravel (AG or BG) that was 10-cm thick; this thickness and the corresponding particle size range are commonly applied by farmers in the northwestern Loess Plateau. No plants were planted during the experiment.

Before taking soil samples, the gravel cover layer on the soil surface should be removed first. In each plot, 10 soil samples were randomly collected from a 0–20 cm depth using a soil auger on 25 September 2021. These soil samples were then pooled per plot into one homogenized composite sample. Plant tissues and stones were removed using a 2-mm sieve, after which each sieved sample was divided into two parts. One part was stored in a container with 15 kg of dry ice and sent to the Biomarker Technologies Corporation, Beijing, China (www.biomarker.com.cn) for assessing its microbial community structure via amplicon sequencing, while the other part was air-dried for its soil physicochemical analysis.

Soil organic carbon was determined by potassium dichromate method (Bao, 2008), while total carbon (TC) and total nitrogen (TN) was measured by an elemental analyzer (Vario Macro, Elementar, Germany). Available nitrogen (AN) was determined by the NaOH alkali diffusion method, with the nitrate-N and ammonia-N contents of soil samples measured by ion chromatography (ICS-600, Thermo Fisher Scientific, the United States of America). Soil pH was quantified according to ISO 10390 (ISO, 1994): 5 grams of soil were shaken for 5 min at 25 mL 1 M KCl and allowed to stand for 2 h; pH was then recorded using glass electrodes. Total salt content of soil was determined by the conduction method (Lu, 1999). Dust accumulation (DA) in the gravel mulch layer was assessed using an iron sampling grid (1 m × 1 m) positioned at five points within each plot; dust on the stone surface was washed away by pure water and then oven-dried the solution for 12 h. Particle distribution of dust accumulation was measured by a laser particle size analyzer (Ambivalue BV, Dussen, The Netherlands). Total amount of genomic DNA was extracted from each soil sample by using the PowerSoil DNA Isolation Kit (MoBio Laboratories, Inc., USA), as directed by the manufacturer, and stored at −80°C for subsequent analyses.

The MiSeq sequencing strategies used in this study for bacteria and fungi followed the respective protocols described by Kozich et al. (2013). The 16S rRNA tag-encoded high-throughput sequencing was carried out on the Illumina MiSeq platform by Biomarker Technologies Co., Ltd., Beijing, China. Pairs of reads from the original DNA fragments were aligned according to previously described methodology (Magoč and Salzberg, 2011). Sequencing reads were assigned to each sample according to their unique barcode, and all sequences were analyzed with QIIME (Quantitative Insights Into Microbial Ecology) software and the UPARSE pipeline. The reads were first underwent quality-filtering by applying the method available in QIIME, whose default settings were then used for their Illumina processing. The UPARSE pipeline was then used to identify operational taxonomic units (OTUs) at 97% similarity threshold. For each OTU, a representative sequence was selected for its taxonomic assignment using the RDP Classifier. Next, the estimated species richness was identified by a rarefaction analysis; the Chao1, ACE，Simpson, and Shannon indices of diversity were obtained as described previously (Schloss et al., 2009).

Statistical analyses and testing were implemented in SPSS v19.0 software (IBM, Armonk, NY, United States) and using the Rstudio v4.2.0 software (R Foundation for statistical Computing, Vienna, Austria). Data were subjected to one-way analysis of variance (ANOVA); when an ANOVA’s F-ratio test was significant, the means of each treatment were compared using Fisher’s least significant difference test, with statistical significance set at p < 0.05.

Figure 1 shows the TC and TOC in the variously treated soils. Compared with the control, TC in the AG or BG treatment averaged 21.24 and 22.33 g/kg, respectively. TC in the BG treatment increased 7.1% after 6 years of gravel mulching compared with the control, being the highest point among all treatments (Figure 1A). A similar trend was found for the TOC content (Figure 1B). Although there was no significant difference in the total nitrogen (TN) content among treatments, the available nitrogen (AN) content was higher under the BG than AG treatment. At the same time, ammonia-N content of the BG treatment was significantly higher than that of the control. By contrast, for the nitrate-N content, it was significantly lower in both mulching treatments than the control, especially in the AG treatment.

There was no significant difference in soil pH after 6 years of gravel mulching, but the total salt (TS) content was 47.9 and 48.3% lower for both AG and BG treatments, along with significantly lower EC values, in comparison with the control. We also found a significant change in the accumulated dust in the gravel mulch layer after 6 years of mulching, with the DA of the AG treatment being significantly higher than that of the other two treatments.

Changes to the TC, TOC, and AN in soil and significant increases in crop yields on gravel-covered farmlands during the same cover time from other researchers’ papers are mutually verifiable. For example, watermelon yields were significantly higher after 5 years of gravel mulching than at the onset mulching (Liang et al., 2011). Since the gravel mulch layer covering the soil leads to the isolation of material inputs such as litter, significant changes in soil physico-chemical properties may be due to several factors: (i) Gravel mulching can significantly increase soil moisture and reduce soil erosion, which is well recognized as a serious problem in the Loess Plateau (Li et al., 2000; Wang et al., 2009; Lv et al., 2019). Our previous research showed that the amendments with gravel significantly increased soil moisture, the soil moisture of AG, BG, and CK treatment were 19.1, 19.8, and 13.6%, respectively, after 6-year gravel mulching. Applying gravel mulching also significantly increased the rainfall threshold to generate runoff, and caused fewer runoff events (Qiu et al., 2021). (ii) Gravel mulch collects atmospheric dust; previous research has found that gravel mulch is effective at capturing wind dust, and there is 2–3 times more organic matter in the air than local soils of the Loess Plateau (Li et al., 2001). The changed nitrate-N and ammonia-N contents suggest that gravel mulching may inhibit the nitrification process in the soil, leading to less nitrate-N, or that more it is simply washed away by soil runoff under gravel mulching (Qiu et al., 2021).

Gravel mulching reduced the TS and EC values of soil, a result in line with other findings that suggested gravel mulching could alter the water–salt transport process, prompting some to apply it to saline land management (Sun et al., 2013). This is due to the fact that salt transport in soil is dominated by water, yet gravel mulching increases the rate of infiltration (Lv et al., 2021); hence, as soil salt moves downwards with water infiltration and gravel mulching reduces soil water evaporation (Qiu et al., 2014), less salt accumulates in the upper layer as water evaporates. Concerning the impact of gravel mulch on DA, this is best explained by the fact that the mulch is essentially a porous medium with abundant porosity; accordingly, the AG treatment has a larger specific surface area and can therefore accommodate a greater DA. Albeit at static measure—only indicating the DA in the gravel layer at some time point (i.e., after 6 years of mulching)—it suggests that most of the atmospheric dust captured by the gravel layer seeps into the soil through precipitation solubility, which may have contributed to the changed carbon and nitrogen contents of the experimental soil. Furthermore, previous research has shown that the size of the gravel particles is of importance in changing the channel characteristics of water vapor transport in the gravel mulch layer, which consequently should also change soil water erosion and the loss of important nutrients such as carbon and nitrogen (Li et al., 2000; Qiu et al., 2014, 2021).

From the results of this study, we believe that the significant improvement in the ability of gravel mulching to capture atmospheric dust is one of the main reasons for the increase in soil carbon and nitrogen content after 6-years mulching. In addition, our previous research found gravel mulch significantly reduces runoff caused by rainfall, which may also be one of the reasons for the increase in soil carbon and nitrogen after 6-years mulching (Qiu et al., 2021). As for the decrease in soil carbon and nitrogen content after longer years, we believe that there may be two main reasons. One is that gravel mulching significantly increases the infiltration rate of rainfall, which may take away nutrients from the surface soil (Lv et al., 2021). Secondly, soil microorganisms may play an important role. But the specific mechanism still needs more long-term studies.

Evidently, the 6 years of gravel mulching had an impact on soil mechanical composition (Figure 2). The percentage of 0.1–0.05 mm soil particles was significantly greater under the BG treatment than the two other treatments, with that under the AG treatment significantly surpassing the control as well. Another study had suggested that the combined gravel and soil interface, which causes some minute gravel particles to enter the soil surface and affect both the warming of soil and its water retention function, is one of the main reasons why soil degradation ensues under gravel mulch (Wang et al., 2010). This view is consistent with our results, especially for the small-particle gravel mulching treatment (AG), where the gravel mulching layer gradually compresses subsoil gravel into the soil, while precipitation processes increasingly obscure the gravel mulching layer and soil boundary. In this study, we also measured the particle distribution of residual atmospheric dust in the gravel mulch layer, finding significant differences among the treatments, of which the BG treatment containing a significantly greater proportion of 0–3.9 μm particles. For the 3.9–31.2 μm size range, its percentage of DA under both mulching treatments was significantly lower than that of the control. However, for the 31.2–62.5 μm range, its percentage of DA was significantly higher in the AG treatment than the two other treatments, and also significantly higher in the BG treatment than the control. DA on the gravel surface was significantly correlated with larger-sized portions of soil mechanical composition. This suggest the latter’s variation may be due to a combination of subgrade gravel mixing and the selective dust capture ability of layers of different sized gravel mulch.

The 6 years of gravel mulching significantly changed the soil microbial diversity indexes (Figure 3). The ACE and Chao1 indices showed a similar trend, being higher under the gravel mulching treatments than the control, and also significantly higher under the BG than AG treatment. Both Shannon and Simpson indexes of the AG treatment were significantly lower than the other treatments; the Shannon index of BG treatment was significantly higher than the control, but vice versa for the Simpson index. The ACE and Chao1 indexes are often used to estimate the total number of species (community richness); hence, we may infer that gravel mulching promotes the total number of microbial species in soil. The Shannon and Simpson indices are often used to represent microbial alpha diversity. These results indicate that the richness of soil microbial communities in the mulched treatment was improved compared to the control, although the increase of the AG treatment was not significant. Besides, the diversity of AG treatment was significantly lower than the other treatments. More soil microbial species after 6 years of gravel mulching could be explained by its ability to considerably increase soil temperature (Zhao et al., 2020). The discrepancy between the Shannon and Simpson diversity indexes of the large versus small particle-size treatments may be explained by their air permeability. The treatment using large gravel particles (BG) has a large pore size and a weak resistance to air evaporation resistance, whereas gravel mulching using small particles (AG) has a poor air permeability but a strong air resistance (Qiu et al., 2014), which further inhibits the decline of microbial species.

A Venn plot (Figure 4A) showed that AG, BG and CK shared 1,023 bacterial community. There were 81, 163, and 98 unique bacterial community for AG, BG and CK, respectively. That also meant that AG and BG were more similar with each other, while CK showed less similarity with AG and BG. AG also had fewer unique compounds as compared with CK and BG. A sankey diagram (Figure 4B) showed the trend of relative abundance changes at the bacterial phylum level in different treatments, and Figure 4C shows the 6 years of gravel mulching resulted in a significant impact upon soil bacteria at the phylum level. According to the relative abundance distribution of detected phyla, Acidobacteria was the most abundant phylum in all samples (relative abundance of 18.5–30.8%); the other major dominant bacterial groups (i.e., relative abundance >5%) were Proteobacteria, Actinobacteria, Gemmatimonadetes, and Verrucomicrobia, having relative abundances of 14.9–25.1%, 5.2–31.6%, 18.5–30.8%, 9.0–15.9%, and 5.3–7.5%, respectively. As evinced by Figure 4, gravel mulching significantly changed the relative abundance of bacteria at the phylum level, by increasing that of Acidobacteria vis-à-vis the control, especially under the AG treatment, for which it was significantly higher than the other treatments. However, the relative abundance of Actinobacteria in either gravel mulching treatment was significantly lower than the control, while that in the BG treatment was significantly lower than the other two treatments. At the same time, the relative abundance of Bacteroidetes in the AG treatment was significantly lower than the other treatments, while that in BG treatment was significantly higher than the other treatments. For Gemmatimonadetes, its relative abundance was significantly higher under both gravel mulching treatments than the control, being significantly greater in the AG treatment than the other treatments. The relative abundance of Proteobacteria was also significantly higher under both gravel mulching treatments than the control, with that under the BG treatment being significantly greater than the other two treatments.

From Figure 5, we can see the change in the relative abundances of various bacterial families under the different treatments. For example, while it was significantly increased by gravel mulching generally, the relative abundance of Gemmatimonadaceae in AG treatment was significantly higher than that in other two treatments; a similar trend was found for Nitrosomonadaceae. However, gravel mulching significantly decreased the relative abundance of certain bacterial families such as Chitinophagaceae, whose relative abundance under the AG treatment was significantly lower than that of other treatments.

Few species of fungi were detected in the plot soil, and the impact of gravel mulching on them at the phylum level was not significant, whereas significant gravel mulching-induced changes in the relative abundance of fungi at the genus level were found (Figure 6). For example, gravel mulching significantly increased the relative abundance of Alternaria, it being significantly higher in AG treatment than the other treatments. The relative abundances of Fibulobasidium, Iodophanus and Torula also showed similar trends. However, the relative abundance of Metarhizium featured a different trend, it being significantly reduced by the AG treatment compared with the control.

The bacterial and fungal community structure under the AG treatment differed from that under the BG treatment, which suggests gravel particle size is also related to the changed microbial community structure. After 6 years of gravel mulching, the soil nutrients of the AG and BG treatments had improved, and Bacteroidetes usually exist in environments having higher nutrient levels (Noah et al., 2007). An earlier study pointed out that Proteobacteria are usually abundant in soils with a high organic carbon content (Axelrood et al., 2002). In addition, the ratio of Proteobacteria and Acidobacteria can serve as an indicator to measure the soil fertility (Smit et al., 2001). In our study, compared with no mulching applied, Proteobacteria was more abundant in both gravel mulching treatments, whereas Acidobacteria was less abundant. This also shows that the soil nutrient level is greater after 6 years of gravel mulching than without it. In addition, the increased abundance of Proteobacteria and Bacteroidetes may be due to the enhanced elemental cycling, which in turn should augment soil fertility and thus promote plant growth (Lesaulnier et al., 2008). This could explain, in part, why gravel mulched farmland had significantly increased crop yields during the same cover time. Alternaria is a pathogenic fungus that can infect the leaves of melon crops. Previous researchers have reported on the leaf spot disease caused by Alternaria and the mechanism by which it renders melon seedlings white (Jia and Li, 2002; Lv et al., 2020). The main crops of gravel mulched farmland are melon crops, these dominated by watermelon. Therefore, that gravel mulching promotes the relative abundance of Alternaria is likely a chief reason for the characteristic decline in farmland production and land degradation that arises in the later period of gravel mulching.

PCA results for bacteria and fungal communities are shown in Figure 7. PCA was dissimilarity able to distinguish between the AG and CK bacteria communities from that of the BG treatment using the first principal component (PC1). The fungal community of the CK treatment differed in the first principal component (PC2) from those of the AG and BG treatments. PCA results showed that soil bacterial community structure under mulch treatments differed from those found in bare soil. Figures 8, 9 shows the effects of soil physical and chemical properties, DA in the mulching layer, and the sand content of surface soil on the top-10 genera of soil bacteria and fungi and the correlations among them. The sand content is significantly related to the content of TC and TOC in soil, while nitrate-N, TS, and EC in the soil are significantly related to DA in the gravel mulch layer. This indicates the importance of accumulating atmospheric dust as an external material source of soil under gravel mulching. The TC, TOC, AN, nitrate-N, and TS are soil physical and chemical properties greatly influenced the soil microbial community. For example, Halophilic microorganisms such as Nitrilitruptoraceae is very sensitive to changes in the concentrations of soil salt ions (Sorokin et al., 2009), and the abundance of Endocarpon is significantly correlated with salt ions in our study. This result is consistent with many previous studies finding that microbial community structure is affected by the nutrients in soil, such as carbon and nitrogen (Chodak et al., 2013; Delgado-Baquerizo et al., 2018). Given that soil nutrients are the material basis for the survival of soil microorganisms, changes to soil nutrient levels will affect the population growth of microorganisms, which may lead to changes in bacterial ecological strategies and ultimately changes in bacterial community composition over time and space (Noah et al., 2007; Huang et al., 2019). Organic carbon mainly modulates the structure of bacterial communities by regulating the energy and nutrients used by microorganisms in the process of nutrient transformation (Zhang et al., 2016). Among the various factors to affect soil microorganisms under gravel mulching, the capture of dust by gravel mulch and the modified carbon and nitrogen components in the soil evidently play a major driving role. Soil sand content may have altered specific heat, changing soil temperature and evaporation, thereby affecting soil microorganisms. Yet, considering that improving soil sand content in the soil is a process of gradual mixing of the overburden and the soil surface, and that changing both soil salt and pH is also a long-term process, more long-term observations and experimental research are needed to distinguish and gage their roles on longer temporal scales. In this context then, the results of this study can enhance our understanding of the relations between microbial community composition and soil property changes. Nonetheless, we still need more long-term studies from other sites to reveal the underlying mechanisms that drive pronounced and subtle changes in microbial community composition under gravel mulching.