Response of organic carbon mineralization and bacterial communities to soft rock additions in sandy soils

Study site

The experimental plot was located at the Fuping County pilot test base of Shaanxi provincial land engineering technology research institute (109°11′E, 34°42′N). The elevation is 375.8–1420.7 m. The area belongs to the continental monsoon warm zone and has a semi-arid climate. Annual average sunshine duration is approximately 2,389 h, annual average precipitation is 527 mm, annual average temperature is 13.1 °C, and annual average total radiation is 5,187 MJ m⁻². Interannual variation for precipitation is large, with a coefficient of variation (CV) of 21.1%. The water table is approximately at a depth of 75 m.

Experimental design

Soft rock and sand samples were taken from Daji Han Village, Xiaoji Han township, Yuyang district, Yulin city (109°32′E, 38°27′N). Minerals in the soft rock were mainly quartz, montmorillonite, feldspar, calcite, illite, kaolinite and dolomite. Main chemical constituents of soft rock by mass were SiO₂ (65%), Al₂O₃ (14%), Fe₂O₃ and CaO (21%). The mineral in the sand is mainly quartz (SiO₂), and its mass fraction is about 82%. The remaining minerals are mainly feldspar (10% by mass), kaolinite (4% by mass), calcite (2% by mass) and amphibole (2% by mass).

The field test plot was selected to simulate typical conditions of the Mu Us sandy area. The test plot was prepared with a mixing depth of soft rock of 0–30 cm, and 30–70 cm was completely filled with sand (total of 1 m depth). The test plot was set up in 2009. Four treatments were prepared and consisted of different volume ratios of soft rock to sand: 0:1 (CK, blank experimental group), 1:5 (C1), 1:2 (C2), and 1:1 (C3). Each treatment was replicated 3 times (n = 12 trial plots) with an area of 2 m × 2 m per replicate. Considering the uniformity of factors such as illumination and micro-topography, the test plot was arranged from south to north in a “one” shape. Experimental plots were sown twice per year, once with corn (Jincheng 508) and once with wheat (Xiaoyan 22). The same crop varieties were planted over the previous 11 years. All treatments differed only in the volume ratio of soft rock and sand, and received the same fertilizer inputs. Each received 255 kg ha⁻¹ (N), 180 kg ha⁻¹ (P₂O₅) and 90 kg ha⁻¹ (K₂O) of fertilizer consisting of urea (including N 46.4%), diammonium phosphate (including N 16%, containing P₂O₅ 44%), and potassium sulfate (including K₂O 52%). 

Collection of soil samples

After the wheat was harvested in May 2019, five soil samples at a depth of 0–30 cm were collected from each plot and thoroughly mixed. Animal and plant residues were removed using a 2 mm sieve, and then divided into three equal portions. One portion of each soil sample was placed into a 4 °C refrigerator for mineralization incubation tests and determination of NO${}_3^{-}$-N and ammonium (NH${}_4^{+}$-N); one was placed in a −80 °C refrigerator for high-throughput sequencing and real-time fluorescence quantification, and one was naturally air-dried for determination of SOC, TN, AP and available potassium (AK).

Determination method

SOC was determined by potassium dichromate-concentrated sulfuric acid external heating method (Nelson & Sommers, 1996). TN was determined by Kjeldahl digestion, AP was determined by molybdate blue colorimetry, and AK concentrations were measured using atomic absorption spectrometry (Dai, Wang & Fu, 2017). NO${}_3^{-}$-N and NH${}_4^{+}$-N were extracted at a ratio of 10 g fresh soil to 100 mL 2 M KCl. After shaking for 1 h, the extracts were filtered and analyzed by continuous flow analytical system (San++ System, Skalar, Holland) for NO${}_3^{-}$-N and NH${}_4^{+}$-N. The soil temperature (ST) from January to December 2019 was measured by the soil temperature measuring instrument (JC-TW, JCEP, Qingdao, China), starting at 8:00 on the 15th day of each month and ending at 20:00, once every 2 h. Soil bulk density (BD) was measured using the ring knife method, WC of soil was measured gravimetrically (Pandey et al., 2010).

The organic carbon mineralization test uses the lye absorption method (Guo et al., 2019): weigh 30 g of the soil placed in a refrigerator at 4 °C in a beaker, keep the field water holding capacity at about 60%, and pre-incubation it in the incubator at 25 °C for 5 days. Then put the lye and the soil-filled beaker into the incubation bottle to be sealed and incubated in dark. Take out the lye and titrate with dilute acid at the 1st, 2nd, 3rd, 4th, 5th, 8th, 11th, 14th, 18th, 22nd, 26th, 30th, 40th, 50th and 60th days of incubation respectively, and add water to the beaker to a constant weight by weighing method each time.

DNA extraction: 0.5 g of soil samples stored at −80 °C were weighed, and bacterial DNA was extracted according to the instructions of the soil DNA kit (Omega bio-tek, Norcross, GA, USA), and then DNA purity and concentration were determined by NanoDrop 2000 spectrophotometer (Thermo Scientific, Wilmington, USA), and DNA integrity was detected by 1% agarose gel electrophoresis.

PCR amplification and high-throughput sequencing: general primers 338 F (5′-ACTCCTACGGGAGGCAGCAG-3′) and 806R (5′-GGACTACHVGGGTWTCTAAT-3′) in V3-V4 region of 16S rRNA gene were selected for amplification (Caporaso et al., 2012). The formal PCR test used TransGen AP221-02: TransStart Fastpfu DNA Polymerase, 20 µL reaction system: 5 × FastPfu Buffer 4 µL, 2.5 mM dNTPs 2 µL, Forward Primer (5 µM) 0.8 µL, Reverse Primer (5 µM) 0.8 µL, FastPfu Polymerase 0.4 µL, BSA 0.2 µL, Template DNA 10 ng, Supplement ddH2O to 20 µL. The PCR reaction procedure was as follows: 95 °C for 3 min, 30 cycles, 95 °C for 30 s, 55 °C for 30 s, 72 °C for 45 s, 72 °C for 10 min. The amplification product was detected by 2% agarose gel electrophoresis. After measuring the concentration of the purified product, the equimolar number was mixed. Sequencing was performed on the Illumina MiSeq platform (Illumina, San Diego, USA) according to the standard protocol of Majorbio Bio-Pharm Technology Co. Ltd. (Shanghai, China).

Real-time PCR: DNA extraction was as described above. The abundance of the bacterial 16S rRNA gene copy number was determined using an ABI 7300 type real-time PCR instrument (Applied Biosystems, USA). The reaction system is as follows: 2X Taq Plus Master Mix 10 µL, forward primer 338 F (5′-ACTCC TACGGGAGGCAGCAG-3′) 0.8 µL, reverse primer 806R (5′-GGACTACHVGGGTWTCTAAT-3′) 0.8 µL, template DNA 1 µL, ddH2O 7.4 µL. The PCR amplification procedure was as follows: 95 °C for 5 min, 40 cycles, 95 °C for 5 s, 50 °C for 30 s, and 72 °C for 40 s.

Data analysis

Reads of PE obtained by Miseq sequencing were first spliced according to their overlap relationship, and quality control and filtration were conducted for sequence quality. Samples were distinguished according to their barcode and primer sequences at both ends of the sequence to obtain the effective sequence, and sequence direction was corrected. The RDP classifier Bayesian algorithm was used to classify the 97% similar level OTU representative sequence, and community composition of each sample was counted at the Phylum and Genus taxonomic level. Species diversity analysis was performed using OTU.

Rates of SOC mineralization were calculated according to the method of Wang et al. (2014). Data of potential mineralizable organic carbon (C₀), semi-turnover period (T_1∕2) and turnover rate constant (k) were fitted using a first-order kinetic equation (C_t = C₀ (1-e^−kt)) by Origin 2017 software (OriginLab, Hampton, USA). Soil temperature was averaged for calculation. Both Figs. 1–4 and Tables 1–7 used Microsoft Excel 2010 for data calculation and graph drawing, and SPSS 20.0 software (IBM, Stanford, USA) was used for analysis of variance test. Figures 5 and 6 were drawn using the vegan package and the pheatmap package of the R language, respectively. Figure 7 used rda in the vegan package of R language for mapping.

Basic soil properties

Results demonstrate that SOC content of sand can be significantly increased with addition of soft rock to sand (P < 0.05) (Table 1), whereby treatment C3 had the highest SOC content. Contents of TN and NO${}_3^{-}$-N increased with increasing fractions of soft rock, with C3>C2>C1>CK. Contents of NH${}_4^{+}$-N, AK and BD were not significantly different among treatments (P > 0.05). The treatment C2 had a significantly greater AP content when compared to the other treatments, while there was no significant difference observed among the other treatments. ST increased with increasing volume fraction of soft rock, and there were significant differences among treatments. Changes in WC and ST were consistent, and no significant difference was observed among the C2 and C3 treatments. Overall, it is evident that the volume ratio of soft rock and sand impacts basic physical and chemical properties of the resulting compounded soil.

SOC mineralization rate and cumulative mineralization

Results demonstrate that rates of SOC mineralization could be divided into two stages: rapid decline (days 1–11) and slow decline (days 11–60) (Fig. 1). Average values of rates of mineralization of SOC among treatments was C3>C1>C2>CK from greatest to least. On the 11th day of incubation, the SOC mineralization rate was reduced by 79.91% (CK), 79.65% (C1), 58.46% (C2) and 49.25% (C3) when compared with the 1st day, respectively. On the 60 th day of incubation, rates of SOC mineralization was reduced by 93.86% (CK), 89.79% (C1), 85.91% (C2) and 85.52% (C3) when compared with the 1st day, respectively. As presented in Fig. 2 accumulation of SOC by mineralization (C_t) was best described by an exponential function over time, as cumulative release gradually decreased. After 60 days incubation, C_t of all treatments was between 368.58–1046.28 mg kg⁻¹. When compared with the CK treatment, C_t in treatments C1, C2 and C3 increased 94.91%, 71.79% and 183.86%, respectively.

SOC mineralization parameters

Results showed that cumulative mineralization rate of organic carbon (C_r) in each treatment was C3>CK>C2>C1 from greatest to least (Fig. 3). The C3 treatment significantly increased C_r. There was no significant difference between CK, C1 and C2 treatments. When Compared with the CK treatment, C_r in the C1 treatment and C2 treatment decreased by 1.33 and 0.51 percent, respectively, and C3 treatment increased by 5.44 percent. Results of model fitting indicate that C₀ had a similar trend to C_t, as C3>C1>C2>CK. In addition, there was no significant difference between the C1 and C2 treatments. When compared with the CK treatment, k values of C1, C2 and C3 treatments were significantly lesser as CK>C3>C2>C1. Trends in T_1∕2were opposite to that of k (Table 2).

16S rRNA gene copy number

Gene copy numbers for all treatments ranged between 1.76 × 109-2.83 × 109 copies g⁻¹ soil (Fig. 4). Gene copy number of the C1 treatment was highest (2.83 × 109 copies g⁻¹ soil) and was significantly greater by 61.29%, 23.36% and 22.48% (P < 0.05) when compared to CK (1.76 ×109 copies g⁻¹ soil), C2 (2.30 × 109 copies g⁻¹ soil), and C3 (2.31 × 109 copies g⁻¹ soil), respectively. There was no significant difference in 16S rRNA gene copy number when C2 and C3 were compared, however they were significantly greater when compared with CK, 23.51% and 24.06% respectively.

Bacterial community composition

At the Phylum level, the distribution of bacterial abundance in all treatments was relatively consistent and distributed among 12 Phyla (Table 3): Actinobacteria>Proteobacteria >Chloroflexi>Acidobacteria >Bacteroidetes>Firmicutes >Gemmatimonadetes>Cyanobacteria >Saccharibacteria>Verrucomicrobia >Nitrospirae>Planctomycetes, unclassified phyla accounted for less than 1% of abundance (Ji et al., 2017). Actinobacteria, Proteobacteria and Chloroflexi were the three major phyla, accounting for 66.90%–74.29% of total abundance. When compared with the CK treatment, abundance of Actinobacteria treated with C1 increased 13.18%, and decreased with increasing proportions of soft rock. Proteobacteria were the most abundant in the C2 treatment. Abundance of Chloroflexi increased with increasing proportions of soft rock, and was greatest in the C3 treatment. When compared with the CK treatment, abundance of Acidobacteria and Gemmatimonadetes in the C1, C2, and C3 treatments increased (P < 0.05), while Bacteroidetes abundance decreased significantly. There was no significant difference in abundance of Firmicutes among treatments (P > 0.05). When compared with the CK treatment, abundance of Cyanobacteria and Verrucomicrobia was significantly lesser, and was most evident in the C2 treatment. Nitrospirae were the most abundant in the C2 treatment which was significantly greater than the CK treatment. Abundance of Planctomycetes was greatest in the C3 treatment (Table 3).

Analysis of bacterial community composition at the genus level in treatments demonstrated a relatively consistent trend, similar to that observed at the phylum level (Fig. 5). In all treatments, Pseudarthrobacter had the highest abundance, ranging from 5.05% to 7.93%, followed by norank_o_ JG30-KF-CM45 (3.02% to 4.62%). The color of the color block in the heatmap represented the relative abundance of a genus, and clustering was performed based on the similarity of the abundance between the samples, indicating that the C1 and C2 treatment were classified as similar, while CK and C3 treatment were classified as separate clusters. The 50 genera presented in Fig. 5 could be classified into 8 phylums, respectively including: Actinobacteria, Proteobacteria, Chloroflexi, Acidobacteria, Bacteroidetes, Firmicutes, Cyanobacteria and Nitrospirae. Results demonstrate that the dominant genus of bacteria were the same in all treatments, whereby most belonged to the genus Actinobacteria and Proteobacteria.

Bacterial diversity index

After optimization and integration of the original sequence of Miseq data, a total of 185491 V3–V4 regions of 16S rRNA gene conforming to Q20 (accuracy up to 99%) were obtained, with an average length of 414.95 bp. After screening, the four treatments contained a total of 43,356–50,207 high quality sequences and 2,073–2,249 operational taxon units (OTUs). Results showed that the addition of soft rock increased species diversity and richness, and the C2 treatment was most diverse and rich. Table 4 demonstrates that the number of sequencing strips in each treatment was enough to cover most of the microorganisms in the sample, and results were reasonable. The Shannoneven index indicated that the addition of soft rock promoted species uniformity, with the highest uniformity observed in the C2 treatment.

Correlations between diversity indices and organic carbon mineralization parameters showed that (Table 5) Shannon index and Shannoneven index had positive correlations (P < 0.01). The Chao index was positively correlated with SOC and T_1∕2 (P < 0.05), and negatively correlated with k. The parameters C_t and T_1∕2 were positively correlated with SOC, as was C₀. The parameter k was negatively correlated with SOC. Results showed that increases in organic carbon content promoted increased cumulative mineralization and potential mineralizable organic carbon, and promoted increased conversion of stored carbon when species richness increased.

Correlation between bacterial community structure and SOC mineralization parameters and physicochemical properties

The correlation Heatmap demonstrate that values of C_t and C₀were negatively correlated with Proteobacteria (P < 0.01), and positively correlated with Gemmatimonadetes (P < 0.05). Values of C_r were positively correlated with Chloroflexi. Values of C₀/SOC had a positive correlation with Chloroflexi and Acidobacteria, respectively. Values of k were positively correlated with Cyanobacteria and negatively correlated with Nitrospirae. Results demonstrated that degrees of organic carbon mineralization were closely related to the dominant bacteria, while non-dominant bacteria determined the intensity of organic carbon mineralization (Fig. 6). The effect of different soil physical and chemical properties on bacterial community structure were assessed (Fig. 7). Results showed that the first two axes of the RDA analysis explained 67.75% of total variance in community structure, the first axis explained 43.71%, and the second axis explained 24.04%. Nitrate (NO${}_3^{-}$-N), TN, and WC had a greater effect on the bacterial community structure, and the order of influence was NO${}_3^{-}$-N>TN>WC. Spearman correlation analysis was performed on soil bacteria and soil physicochemical properties (Table 6) and results demonstrate that Acidobacteria, Bacteroidetes, and Gemmatimonadetes were correlated with soil properties. Acidobacteria had a significant positive correlation with AK, NO${}_3^{-}$-N, NH${}_4^{+}$-N, and a significant negative correlation with BD. Bacteroidetes had a significant negative correlation with TN, AK, NO${}_3^{-}$-N, ST, and WC, and a significant positive correlation with BD. Gemmatimonadetes was positively correlated with TN, ST, and WC, and negatively correlated with BD.

Bacterial metabolic function

The metabolic functions of bacteria were investigated. Only metabolism of cofactors and vitamins differed among treatments (Table 7). When compared with the CK treatment, there was no significant difference in metabolism of cofactors and vitamins in the C3 treatment, while metabolism of cofactors and vitamins in C1 and C2 treatment were significantly reduced. When differences among all metabolic functions in same treatment were compared, no difference between lipid metabolism and metabolism of terpenoids and polyketides was observed. Furthermore, there were significant differences between other metabolic functions, such as carbohydrate metabolism.

Effects of soft rock on the accumulation of organic carbon in sandy soil

SOC plays an important role in nutrient storage and maintenance in sandy soil ecosystems. As an energy resource for microorganisms, differences in organic carbon content and the degree of mineralization impact availability of soil nutrients (Gallardo & Schlesinger, 1992). Soft rock addition to soils increased the sandy SOC mineralization. Results of this study demonstrated significant differences in SOC accumulation mineralization (C_t) prepared with different ratios of rocky soil as rates were greatest to least as follows: C3>C1>C2>CK (Fig. 2). The C3 treatment significantly increased soil SOC, TN, AP, NO${}_3^{-}$-N and WC content (Table 1). The low nutrient content in the CK treatment likely affected the rate of mineralization of SOC contributed to the slow release of CO₂ (Han, Liu & Luo, 2012). Previously, Dai, Wang & Fu (2017) observed similar results as the treatment with the largest amount of organic carbon mineralization significantly increased SOC, TN, and AP contents. As the content of soft rock gradually increases, the content of clay and silt particles in soil increases, and the soil texture gradually changes from sandy to silty loam (Wang et al., 2016). When the amount of soft rock added was 16.7% (C1) and 33.3% (C2), cumulative mineralization rate of SOC was low and there was no significant difference (Fig. 3). When the content of soft rock reached 50% (C3), the soil cumulative mineralization rate was the largest among all treatments, indicating that SOC accumulation is promoted at a ratio of soft rock to sand of 1:5 to 1:2. Because sandy soil has greater permeability coefficients and low water and fertilizer retention and soft rock has more fine particles with hydrophilic and absorbent properties, mixing soft rock and sand can promote increased organic carbon content and mineralization (Sun & Han, 2018). Therefore, improvements of sandy soil using soft rock can improve soil carbon fixation, and is a new means to realize resource utilization.

Effects of soft rock on organic carbon mineralization rate and fitting parameters of sandy soil

The addition of soft rock changed the mineralization rate of organic carbon in sandy soil. The CK and C3 treatment had similar trends in mineralization reactions during the incubation period, and reached a peak on the third day, which was followed by a rapid decline (Fig. 1). However, research by Ribeiro et al. (2010) observed a peak in organic carbon mineralization rate on the first day. This difference is likely due to the pre-incubation of 5 days, while Ribeiro et al. (2010) used a pre-incubation period of 7 days. Therefore the soil microenvironment was likely still in a stage of microbial activity recovery at the beginning of the reaction. It is also possible that the organic carbon in the compounded soil during the initial stage of mineralization was mostly present in a complex form, containing fewer small molecular compounds that are more easily decomposed. During initial stages, microorganisms need to breakdown complex compounds before they can be absorbed and utilized, therefore the respiration rate showed a rapid rise in the initial stage (Li, 2000). The reaction characteristics of the C1 and C2 treatment were similar, and had a downward trend throughout the incubation period. Treatments could be divided into two stages: rapid decline between days 1–11 and slow decline between days 11–60 day, which was consistent with the of Zhang et al. (2011). In the study of Khalil, Hossain & Schmidhalter (2005), it was believed that the rapid decline stage of organic carbon mineralization rate occurred between days 2–9 because exogenous organics stimulated microbial activity. In a study by Shi et al. (2019) it was suggested that the rapid decline of organic carbon mineralization rates occurred in the first 7 days due to addition of fungal residues which promoted rapid decomposition of active organic carbon. However, in the early stage of mineralization, microorganisms have the strongest metabolic capacity for carbohydrates (Table 7). Microorganisms need to simplify complex organic carbon compounds before absorbing and utilizing them, therefore rapid mineralization takes longer and can fluctuate. In the later stages of incubation, the easily decomposed active components in the soil would have been gradually consumed by the microorganisms and remaining major components, such as lignin and cellulose, etc., which are resistant to decomposition, would inhibit the activity of microorganisms (Li et al., 2018). Therefore, mineralization rates would have the observed trend transitioning from fast to slow. The C₀ value observed in this study was consistent with changes in values of C_t (C3>C1>C2>CK) (Table 2). A likely explanation for the observed differences is that greater proportions of soft rock would result in filled non-capillary pores between sand, increasing capillary pressure and promoting formation of soil aggregate structure and accumulation of potentially mineralizable organic carbon (Chevallier et al., 2004). The results of this study showed that there was no significant difference in the value of k between C1, C2 and C3 treatments, and that it was significantly lower when they were compared to the CK treatment. Changes in T_1∕2 and k values were opposite. Because long-term application of chemical fertilizers would increase the inorganic nitrogen content of soils. Introduction of nitrate and ammonium to soils (Table 1), would result in reactions with compounds such as lignin or phenols resulting in slower decomposition of carbon sources (Liu, Du & Li, 2017).

The relationship between bacterial community structure and organic carbon mineralization

Results of real-time PCR analysis showed that gene copy numbers of bacteria ranged between 1.76 × 109-2.83 × 109 copies g⁻¹ soil (Fig. 4). Addition soft rock resulted in greater gene copy number when compared with the CK treatment (P < 0.05) likely due to the additional nutrient content. The different ratios of soft rock and sand led to different texture characteristics of compounded soil which might impact bacteria counts (Wang et al., 2016). Community composition and abundance of bacteria was similar among treatments at the Phylum and Genus, especially for C1 and C2 (Table 3, Fig. 5). In this study, NO${}_3^{-}$-N, TN and WC had the greatest impact on the bacterial community structure (Fig. 7). In contrast, a study by Wu et al. (2011), observed that SOC, AP and AK had a greater impact on bacterial community composition. Liu et al. (2019) suggested that TN has an important effect on soil bacterial community structure, and soil nutrient content. In addition salt content played an important role in the composition of soil bacteria (Li et al., 2019). For SOC mineralization, the mineralization process was directly affected by microbial abundance, community structure and diversity (Mi et al., 2015). Results demonstrate that Proteobacteria was the most abundance bacteria in the C2 treatment and the abundance of Chloroflexi decreased as the proportion of soft rock increased (Table 3). Organic carbon mineralization (C_t and C₀) was negatively correlated with Proteobacteria (P < 0.01), and cumulative mineralization rate (C_r) was positively correlated with Chloroflexi (P < 0.05) (Fig. 6), suggesting a ratio of soft rock to sand of 1:2 can effectively increase SOC sequestration. Because the dominant bacteria were mostly saprophytic heterotrophic bacteria, which are important decomposers in ecosystems, have a metabolic capacity greater than that of non-dominant bacteria, therefore the carbon cycle can proceed smoothly. Preferred carbon sources bacteria are carbohydrates such as glucose, maltose, and starch (Bianchi, Van Wambeke & Garcin, 1998). We observed that bacteria had a strong ability to metabolize carbohydrates (Table 7). When the ratio of soft rock to sand was 1:2, we observed that diversity index, evenness index and richness index of bacteria were highest among treatments (Table 4), and richness index was negatively correlated with k value (P < 0.01, Table 5), T_1∕2 values had the opposite trend. The k value was positively (P < 0.01) and negatively (P < 0.05) correlated with Cyanobacteria and Nitrospirae, respectively (Fig. 6). Cyanobacteria and Nitrospirae decreased in abundance and increased with increasing soft rock, respectively. In addition, the largest variation in abundance was observed in the C2 treatment (Table 3), indicating that the ratio of soft rock to sand of 1:2 significantly reduced turnover rate of organic carbon and increased its retention time. NO${}_3^{-}$-N, TN and WC had a strong influence on bacterial community structure (Fig. 7), and were correlated with Acidobacteria, Bacteroidetes and Gemmatimonadetes, respectively (Table 6). Urea, as the main source of nitrogen, can react with phenolic compounds in soil or impact the movement of vitamins, thereby reducing carbon loss. In future studies, we will continue to investigate enzyme activity and the role of functional bacteria, and provide theoretical support for further understanding of nutrient fixation in compounded soils.