Spatial Variations in Organic Carbon Pools and Their Responses to Different Annual Straw Return Rates in Surface Paddy Soils in South China

Abstract

To identify the effects of straw return on different organic carbon pools in surface paddy soils (0–20 cm), a total of 33 soil samples under different annual straw return rates (SRr) was collected, and then the samples were analyzed based on a 100-day incubation. The data from acid hydrolysis-incubation experiments were fitted to a three-pool first-order kinetics model that divided soil organic carbon (SOC) into active (C_a), slow (C_s) and resistant (C_r) pools. The results showed that the mean pool sizes of C_a, C_s, and C_r were 0.27, 10.26, and 13.46 g·kg⁻¹, representing a mean of 1.35%, 41.91%, and 56.74% of the total SOC (TOC), respectively. The SOC pools in the surface paddy soils in Dongxiang had a small C_a pool but had longer mean residence times of the C_a and C_s pools than those in other regions in China. The three carbon pools were less affected by the paddy soil type but showed obvious spatial variations. The SRr contributed a strong positive effect on the variability of C_s and C_r, especially on C_s variability, while it had very little effect on C_a variability. Soil available nitrogen dominated the variability in TOC and C_r compared to the other soil properties. Therefore, the C_s pool is more sensitive than the other carbon pools to long-term straw return.

1. Introduction

In China, almost 6.10 × 10¹⁰ t·a⁻¹ of agricultural straw is produced, and the estimated potential of farmland carbon sequestration is over 2.3 × 10⁴ Tg·a⁻¹ [1]. Straw return (SR) could help increase the soil organic carbon (SOC) via the direct addition of straw carbon to soil [2,3,4]; it is widely considered a sustainable farming technology for increasing soil carbon sinks and promoting soil fertility [5,6,7].

However, SOC is not a homogenous carbon pool [8,9]; rather, it consists of a continuum of thousands of diverse carbon compounds, with mean residence times (MRTs) ranging from hours to millennia [10,11]. In the CENTURY model, according to the differences in MRT, SOC is divided into active (C_a), slow (C_s), and resistant (C_r) carbon pools [12]. Previous studies have reported the SR could influence the turnover and distributions of SOC pools [13,14,15,16], and different SOC pools respond differently to different annual straw return rates (SRr) [17]. Therefore, identifying the regional patterns and controls of multiple pools of SOC under different SRr will improve our understanding of SOC dynamics in croplands as well as the mechanisms of sequestration in carbon cycling.

The latest study demonstrated that the SR had larger impacts on C_a than on C_s and C_r pools in different cropland soils across China [18]. However, the effects of SR on the different SOC pools were also restricted to the SRr and SR years at a regional scale. In south China, long-term SR could significantly increase the total SOC (TOC) and the active SOC fractions in the double-cropped paddies [15]. Short-term SR could increase the active SOC fractions, but it did not change the TOC in the topsoil in a double-cropped rice paddy in northeast China [19]. In addition, a meta-analysis of 446 sets of data from 95 studies in China suggested that the proper duration of SR was 6–9 years, while it would decrease the SOC by 17.06–20.05% after 10 years of SR [20]. Moreover, different SRr affected the SR sequestration of C in soil [21,22]; generally, a 50% SRr was suggested to be the best option for improving SOC sequestration (both the TOC and the active SOC fractions) in the rice-wheat rotation system of China [14]. In addition, different SRr would make a difference on the priming effect of SR, which could regulate the decomposition of the stable SOC fraction through affecting soil properties, microbial community, and enzyme activities [7,23,24,25].

Previous studies have focused more on the effect of SR on the C_a pool; how rice SR affects the C_s and C_r pools in the surface paddy soils in south China remains unclear. Which C pool is more sensitive to rice SR lasting more than 10 years with different SRr values also needs to be identified. The latter two carbon pools account for a larger and more stable proportion of the TOC, the turnover and dynamics of which are more important for soil carbon sequestration in cropland [18,26].

In this study, a combination method using lab incubation and a three-pool first-order kinetics model was used to estimate the pool sizes of C_a, C_s, and C_r in surface paddy soils with different SRr values in Dongxiang County, south China. Then, we examined the spatial variations in C_a, C_s, and C_r pools in the surface paddy soils in the study area. Furthermore, the effects of SRr and soil properties on the spatial variations of the three pools were explored. The objectives of this study were to (1) examine the spatial distribution of the C_a, C_s, and C_r pools in the surface paddy soils under different SRr values; (2) identify the main controls of the spatial variability of the three C pools; and (3) reveal the effects of the SRr on the three C pools.

2. Materials and Methods

2.1. Study Area

The study area is located in Dongxiang County, Jiangxi Province, southeast China (28°02′~28°30′ N, 116°20′~116°51′ E) (Figure 1) with a total area of 1270 km². The region has a subtropical monsoon humid climate, with an annual average temperature of 18.6 °C and an annual average precipitation of 2180.6 mm. It is in the transition zone between the hills in eastern Jiangxi Province and the Poyang Lake Plain, with sufficient sunshine and rainfall. The farmland is mainly planted with double-cropped rice, and hydromorphic paddy soil is the main type of paddy soil in the study area, most of which is hydromorphic yellow-mud soil, followed by hydromorphic eel-mud soil, and finally, a small amount of hydromorphic alluvial soil [27].

Through field interviews and investigations, fields with SR for 12 to 15 years were used as sample plots in this study, and the method of rice SR involved covering the straw stubble on the soil surface. Then, the straw was plowed into the soil before planting the following year, in which rice fields with less than 5% SRr were usually harvested by hand (Figure 1).

2.2. Soil Sampling and Analysis Methods

Firstly, the distribution of paddy soil types in the study area was ascertained based on the soil map of Dongxiang County at a scale of 1:50,000 [27]. Then, the status quos of planting systems and straw return at various sites were identified through field interview and survey. Finally, a total of 33 soil samples in the surface paddy soils (up to 20 cm) with long term consistent cropping systems but with different SRr values was collected in November 2019, including 22 from hydromorphic yellow-mud paddy soils (HYMS), 7 from hydromorphic eel-mud paddy soils (HEMS), 2 from hydromorphic alluvial sandy paddy soils (HASS), 1 from hydromorphic alluvial paddy soil (HAS), and 1 from submerged eel-mud paddy soil (SEMS). The surface soil of the central point and its four corners for each sampling field was collected within a range of 10 m × 10 m. All soil samples were mixed thoroughly, air dried, crushed, and passed through a 2–mm sieve, and visible roots were removed. Part of each soil sample was then transported to the laboratory and stored in sealed containers at 4 °C before the incubation experiment, and the remaining soil samples were reserved for soil physicochemical property analysis.

The SRr mentioned in this paper represents the annual straw return rate of a certain land under the same return method for many years. The SRr data in the Figure 1 were obtained by field measurement, which can be calculated as follows:

$${{\displaystyle SRr}}_i=\frac{{\displaystyle \sum _{j=1}^n\raisebox{1ex}{${{\displaystyle HSS}}_j$}\!\left/ \!\raisebox{-1ex}{${{\displaystyle THS}}_j$}\right.}}{n}$$

(1)

where SRr_i is the average straw return rate (%) of paddy rice in the sampling field i; n is quantity of randomly selected rice plants in the sampling field i (n = 10); HSS_j represents the height of stubble straw (cm) of the rice plant j; THS_j represents the total height of straw (cm) of the rice plant j.

Determination of soil physicochemical properties followed the methods recommended by Lu (2000) [28]. Soil urease activity (Urease) was determined using an indophenol blue spectrophotometry method, soil catalase activity (CAT) was determined by a permanganate titration method, and soil acid phosphatase activity (ACP) was determined using a phenyl disodium phosphate colorimetric method [29]. The C_r concentration was determined using a K₂Cr₂O₇ oxidation method from the residue of acid hydrolysis [30].

2.3. Incubation Method

One hundred grams of each air-dried soil sample was placed in a 250–mL special glass jar (adjusted to 65% of field water holding capacity) and incubated for 100 days in the dark at 25 °C in a biochemical incubator [31]. The CO₂ emitted from SOC decomposition was absorbed by 20 mL of 0.5 M NaOH. The absorbing liquid was measured at 1, 3, 5, 7, 11, 18, 25, 31, 38, 48, 58, 68, 78, 93, and 100 days from the start of the incubation. The CO₂–C produced through carbon mineralization was determined at different times during the incubation by titration with standard HCl solution [18].

2.4. Three-Pool Separation and Statistical Analyses

According to the differences in mean residence times, the mineralizable SOC in soil was separated into the C_a, C_s, and C_r pools by a three-pool first-order kinetics equation in the CENTURY model [32]. This equation can be expressed as follows:

$$Csoct=Ca\times \exp (-Ka\times t)+Cs\times \exp (-Ks\times t)+Cr\times \exp (-Kr\times t)$$

(2)

where C_soct is the total SOC at time t; C_a, C_s, and C_r represent the pool sizes of the active, slow, and resistant SOC, respectively; and K_a, K_s, and K_r are the corresponding decomposition rate constants, which are calculated as the reciprocals (K⁻¹) of the respective MRTs in the three-pool first-order model. The size of the C_r pool is estimated by acid hydrolysis, with the field MRT commonly assumed to be 1000 years [32]. Additionally, the laboratory-derived MRT was scaled to the field MRT by assuming a Q₁₀ (2^{(25–MAT)/10}), where MAT is the mean annual temperature in the field [33]. This conversion process can be calculated as follows:

$${MRT}_{lab}={MRT}_{field}/Q_{10}$$

(3)

where MRT_lab represents the mean residence time in the laboratory; MRT_field represents the mean residence time in the field; and Q₁₀ is the temperature sensitivity coefficient.

Equation (2) was fitted with a nonlinear regression (PROC NLIN METHOD = Marquardt algorithm) (SAS 9.3, SAS Institute Inc., Cary, NC, USA) that uses an iterative process to estimate the parameters (C_a, K_a and K_s) [34].

The Shapiro-Wilk normality test was performed for each group of samples. One-way ANOVA (Scheffe–Tamhane T2) was used to analyze the differences in the CO₂–C produced from the different types of paddy soils during incubation, the composition of the C_a, C_s, and C_r pools, and their mean residence times in the different types of paddy soils. The Kruskal-Wallis test was used to analyze data that did not conform to a normal distribution. Spearman’s tests were used to determine which of the corresponding soil physicochemical properties, soil enzyme activities, paddy soil types, and straw return values were significantly correlated with the pool sizes of C_a, C_s, and C_r. Linear regression was used to determine the relative effect of each correlated predictor variable from Spearman’s tests based on the coefficient of determination (R²). All of the above statistical analyses were performed using the Statistical Package for Social Sciences (SPSS) software (IBM SPSS Statistics 26, Chicago, IL, USA) [35].

3. Results

3.1. Characteristics of SOC Decomposition

The mean SOC decomposition rates (CO₂–C) of the five types of paddy soil (HYMS, HEMS, HASS, HAS, and SEMS) during the 100-day laboratory incubation declined from 19.3 to 2.5 mg·kg⁻¹ soil d⁻¹, from 18.8 to 2.2 mg·kg⁻¹ soil d⁻¹, from 22.2 to 3.2 mg·kg⁻¹ soil d⁻¹, from 19.7 to 2.1 mg·kg⁻¹ soil d⁻¹, and from 20.2 to 5.1 mg·kg⁻¹ soil d⁻¹, respectively (Figure 2a). Although the sample sizes of paddy soils were different, the basic trends of the decomposition processes were consistent, with rapid decomposition initially, which then gradually reached a steady decomposition state. No significant differences were observed in the decomposition rates of HYMS, HEMS, and HASS (sample size ≥ 2) during incubation (p > 0.05), except for HEMS and HASS at day 68 (p < 0.05) (Figure 2a).

Accordingly, the absolute cumulative mineralization of CO₂–C in the five types of paddy soils (HYMS, HEMS, HASS, HAS, and SEMS) increased logarithmically with incubation time, which varied from 29.3 to 515.5 mg·kg⁻¹ soil, from 28.1 to 427.8 mg·kg⁻¹ soil, from 22.2 to 574.9 mg·kg⁻¹ soil, from 19.7 to 387.4 mg·kg⁻¹ soil, and from 20.2 to 654.3 mg·kg⁻¹ soil, respectively (Figure 2b). Additionally, no significant variations were found in the absolute cumulative mineralization of HYMS, HEMS, and HASS during the 100-day incubation (p > 0.05) (Figure 2b).

3.2. SOC Pools and Their Regional Patterns

No significant differences were found in the pool sizes of C_a, C_s, and C_r in the surface paddy soils between HYMS, HEMS, and HASS or in their ratios with TOC (p > 0.05) (

Table 1. Sizes of the active, slow, and resistant SOC pools and their mean residence times in the surface soils (0–20 cm) from different types of paddy soil.

Soil Type	Ca (g·kg−1)	Ca/TOC (%)	Cs (g·kg−1)	Cs/TOC (%)	Cr (g·kg−1)	Cr/TOC (%)	MRTa (days)	MRTs (year)	n
HYMS	0.25 ± 0.10	1.14 ± 0.79	11.27 ± 3.93	42.76 ± 7.06	14.66 ± 4.97	56.10 ± 6.85	22.88 ± 7.80	10.30 ± 5.93	22
HEMS	0.32 ± 0.09	1.82 ± 1.00	8.30 ± 3.26	42.03 ± 5.00	11.06 ± 4.43	56.16 ± 4.37	29.54 ± 15.41	30.00 ± 33.03	7
HASS	0.28 ± 0.06	1.23 ± 0.11	10.76 ± 5.85	47.25 ± 19.69	11.08 ± 2.74	51.53 ± 19.80	30.54 ± 13.16	24.52 ± 8.85	2
HAS	0.38	3.55	1.08	9.98	9.33	86.47	39.84	22.09	1
SEMS	0.20	0.87	10.03	43.66	12.74	55.47	7.29	7.13	1
Total/ Mean	0.27 ± 0.09	1.35 ± 0.92	10.26 ± 4.17	41.91 ± 9.15	13.46 ± 4.83	56.74 ± 8.74	24.8 ± 10.77	15.6 ± 17.42	33

TOC, total soil organic carbon; C_a, active SOC pool; C_s, slow SOC pool; C_r, resistant SOC pool; MRT_a, mean residence time of C_a; MRT_s, mean residence time of C_s. The same applies below; n, sample size.

). The average pool sizes of C_a, C_s, and C_r in the five types of paddy soil ranged from 0.20 to 0.38 g·kg⁻¹, from 1.08 to 11.27 g·kg⁻¹, and from 9.33 to 14.66 g·kg⁻¹, respectively, composing a mean of 1.35%, 41.91%, and 56.74% of the total amount of TOC, respectively.

In total (Table 1), the average MRT_a and MRT_s were 24.8 days and 15.6 years in the surface paddy soils, which varied from 7.29 to 39.84 days and from 7.13 to 30.00 years, respectively. Additionally, no significant variations were observed between HYMS, HEMS, and HASS.

Large spatial differences were observed among the pool sizes of TOC, C_a, C_s, and C_r in the surface paddy soils in Dongxiang County (Table 1 and Figure 3). The distribution characteristics of the small C_a pool were significantly different from those of the TOC pool (Figure 3a,b), while the larger C_s and C_r pools showed a more consistent spatial distribution with the TOC pool (Figure 3a,c,d), especially in the C_r pool with the largest proportion of the TOC (Figure 3a,d). On the whole, the regional patterns of the TOC, C_s, and C_r were consistent with the distribution of corresponding SRr values in Dongxiang County (Figure 1).

3.3. Effects of Straw Return and Soil Properties on SOC Pools

Spearman’s correlation analyses showed that SRr, TN, TP, AN, and ACP were significantly positively correlated with TOC, C_s, and C_r but not with C_a, and pH was significantly negatively correlated with TOC, C_s, and C_r but significantly positively correlated with C_a (

Table 2. Correlation analyses (Spearman’s tests) between the sizes and MRTs of the SOC pools and different factors, including annual straw return rate, soil physiochemical properties, enzyme activities, and paddy soil type.

Item	SRr	pH	TN	TP	TK	Sand	Silt	Clay	CEC	AP	AN	AK	CAT	ACP	Urease	ST
TOC	0.88 **	−0.57 **	0.86 **	0.50 **	−0.13	0.05	0.35 *	−0.23	0.14	0.51 **	0.91 **	0.22	0.05	0.75 **	0.22	−0.35 *
Ca	−0.22	0.35 *	−0.13	−0.19	0.02	0.12	−0.19	0.04	−0.04	−0.05	−0.12	−0.15	0.00	−0.32	−0.07	0.22
Cs	0.83 **	−0.51 **	0.70 **	0.42 *	−0.16	0.08	0.28	−0.26	0.10	0.33	0.75 **	0.10	0.16	0.73 **	0.25	−0.30
Cr	0.79 **	−0.52 **	0.86 **	0.52 **	−0.12	0.01	0.37 *	−0.20	0.22	0.56 **	0.91 **	0.32	0.03	0.68 **	0.25	−0.32
MRTa	−0.07	0.16	0.07	−0.02	0.16	0.00	−0.09	0.21	−0.03	0.11	0.01	0.05	0.13	−0.20	−0.12	0.09
MRTs	−0.14	0.04	−0.14	0.15	0.01	0.23	−0.34	−0.09	0.30	0.26	−0.09	0.22	0.28	−0.32	−0.03	0.34

TOC, total soil organic carbon; C_a, active SOC pool; C_s, slow SOC pool; C_r, resistant SOC pool; MRT_a, mean residence time of C_a; MRT_s, mean residence time of C_s; SRr, annual straw return rate; TN, total nitrogen; TP, total phosphorus; TK, total potassium; Sand, soil sand content; Silt; soil silt content; Clay, soil clay content; CEC, cation exchange capacity; AP, available soil phosphorus; AN, available soil nitrogen; AK, available potassium; CAT, soil catalase activity; ACP, soil acid phosphatase activity; Urease, soil urease activity; ST, paddy soil type. The same applies below. * significant at p < 0.05; ** significant at p < 0.01.

). In addition, both silt and AP were significantly positively correlated with TOC and C_r. The ST was only significantly negatively correlated with the TOC pool in the surface paddy soils (Table 2). However, none of these factors, including soil physicochemical properties, soil enzyme activity, paddy soil type, and SRr, were significantly correlated with the MRT_a and MRT_s (Table 2).

Further linear regression analysis showed that SRr could explain 64.1% of the C_s variability and 52.3% of the C_r variability, suggesting a strong influence on the spatial variation in the C_s and C_r pools (Figure 4).

Of the controlling factors, the relative impact of different factors on the TOC, C_s, and C_r pools varied greatly (Figure 5). AN had the greatest relative impact on the variability of both the TOC and the C_r pools in these factors (p < 0.05) and SRr made a larger contribution to C_s variability than did the other factors (p < 0.05), while SRr played a small role in the TOC and C_r variability, especially C_r variability (Figure 5).

4. Discussion

In this study, the SOC decomposition rates in the surface paddy soils during laboratory incubation were obviously lower than the values reported in Wang et al. (2016) [36] and Wang et al. (2018) [37]. C_a was the main decomposition fraction of SOC in the initial stage. This difference could be attributed to a smaller C_a/TOC (1.35%) in the surface paddy soils in this study area compared with the values in Wang et al. (2016) (2.90%) and Wang et al. (2018) (2.72%), under the similar TOC in them. Additionally, both the SOC decomposition rates and the absolute cumulative mineralization varied little with the paddy soil type during incubation (Figure 2). That could be ascribed to the little difference of C_a pools in the different types of paddy soil, including the pool sizes, the C_a/TOC, and their MRT_a values (Table 1).

Large differences were found between the pool sizes of C_a, C_s, and C_r in the surface paddy soils in the Dongxiang area (Table 1), in which the SOC pools in the surface paddy soils contained a small C_a pool which accounted for less than 2% of the total SOC, and a much larger C_r pool which contributed more than 50% of the total SOC (Table 1). Although no significant differences were found in the pool sizes of TOC, C_a, C_s, and C_r in the different types of paddy soil (Table 1), there was still an obvious spatial variation in their pool sizes (Figure 3). The spatial distribution pattern was probably closely related to the SRr of sampling fields because of the similar distribution trends. The ratio of C_a to TOC in the paddy soils in this study was lower than that in China’s paddy soils at the national level reported in Wang et al. (2017) [18], while the ratio of C_s to TOC in the paddy soils in this study was higher than the national average reported in Wang et al. (2017) [18]. Additionally, both the average MRT_a (24.8 days) and the average MRT_s (15.6 years) in this study were clearly longer those values (11 days and 5.4 years) in Wang et al. (2021) [34], suggesting more stable C_a and C_s pools in the surface paddy soils in the Dongxiang area.

Here, the SRr had a large effect on the variability of TOC, C_s, and C_r but not on C_a (Table 2, Figure 4); by contrast, most previous studies reported that the SR could significantly enhance C_a [3,4,5,6]. The contradiction occurred because existing studies focused mainly on the active SOC fractions but did not include the stable C_s and C_r fractions [38], and their SR term was relatively short (<10 years), both of which would conclude in conflicting research results [39,40,41]. In this study, the SRr had a greater effect on C_s variability, while AN controlled the variability in TOC and C_r more (Figure 5). This result can be explained by the fact that more of the straw C returned to the soil was converted into the C_s pool under long-term SR. In addition, the latest studies have suggested that stable SOC is more vulnerable to priming by exogenous C (i.e., straw return) input than is labile SOC [38]. With a large amount of straw returned into the soil, the C use of soil microorganisms would switch from SOC to straw with higher nutrient availability, and the resistant SOC pool would be protected from the decrease by the priming effect and its low accessibility [23]. In this process, SR made a larger difference in the turnover of stable C_s and C_r than in C_a, and the C_s pool was more easily affected by the priming effect of SR than was the C_r pool in long-term paddy management. Thus, more attention should be given to the turnover and stability of the C_s pool in surface paddy soils under long-term SR in the Dongxiang region.

5. Conclusions

The decomposition of SOC was little influenced by the paddy soil type during incubation. The SOC pools averaged 1.35% C_a, 41.91% C_s, and 56.74% C_r in the surface paddy soils in Dongxiang, with a MRT_a of 24.8 days and a MRT_s of 15.6 years. Compared to other regions, the C_a pool in the study area was smaller but was more stable due to a longer MRT_a. The pool sizes and composition of the three C pools were less affected by the paddy soil type but still showed large spatial variation. The SRr had a strong effect on the spatial variability of C_s and C_r, especially on C_s variability, but had very little effect on C_a variability. AN controlled the spatial variability in TOC and C_r more than the other soil properties did.