The dominant microorganisms vary with aggregates sizes in promoting soil carbon accumulation under straw application

ABSTRACT Unraveling the influence of microbes on C content at aggregate scale is pivotal for promoting soil C accumulation. Previous studies were based mainly on the mutual transformation process between aggregates, the links between the microorganisms in initial aggregates and inner C content and aggregate sizes were still unclear. In this study, the classified aggregates (> 5 mm, 2–5 mm, 1–2 mm, 0.25–1 mm, and < 0.25 mm) were individually incubated for 10 months under 13C-labeled maize straw application to analyze the relationship between microbial community structure in independent aggregates and inner C accumulation under straw addition. The results show that the SOC content increased in independent aggregates under straw addition, with higher stable C accumulation in < 0.25 mm than in > 2 mm aggregates. Aggregates of > 5 mm were more capable of improving unstable C accumulation and C derived from straw (Cstraw) than smaller aggregates. Fungi and Gram-negative bacteria (G-) were more important to increasing C accumulation in > 2 mm aggregates, whereas Gram-positive (G+) bacteria dominated in < 2 mm aggregates. The results indicate that the contribution of microorganisms within aggregates to inner C accumulation was associated with aggregate sizes.


Introduction
Soil organic carbon (SOC) is the largest terrestrial C reservoir and plays a key role in the global C cycle (Wan et al. 2019). Small changes in the soil C pool might affect the global C balance significantly (Murugan et al. 2019). Due to the long term and large-scale use of soil by humans, the balance between soil and atmospheric C has gradually been affected. Hence, understanding the mechanisms of cycling SOC is essential to increase the accumulation of organic carbon (Sokol and Bradford 2019).
Aggregates, as the basic units of soil, are considered to be stable pools of SOC (Blanco-Canqui and Lal 2004) and contain nearly 90% of sequestered SOC (Somasundaram et al. 2017). Some researchers divide the aggregate sizes into micro-(< 0.25 mm) and macro-(> 0.25 mm) aggregates (Jastrow 1996;Six et al. 2004), and there is less consistency in the study of C in different sizes aggregates. In a previous publication, greater C content was found in macro-aggregates than in micro-aggregates in soils dominated by 2:1 clay mineralogy (Six et al. 2004), and micro-aggregates were suggested to contain the largest stable soil organic C pools (Kong et al. 2005). Similarly, some research found that occluded micro-aggregates inside macro-aggregates contributing greatly to C protection than macro-aggregates (Totsche et al. 2018). The research on the C content in different aggregates can provide more details for C dynamics, therefore, more studies on the relationship between soil C content and aggregate sizes should be illustrated for improving C accumulation at aggregate scale.
As is known, soil aggregates provide different micro-environments (Rabbi et al. 2016) and attract diverse microbial communities (Blaud et al. 2012) that can colonize them well, depending on the readily available carbon abundance, organic matter complexity, and other indexes (Mikha and Rice 2004;Upton et al. 2019). In general, microbial processes, such as microbe-triggered courses and microbial-originated polysaccharides, could promote the mutual transformation of macro-and micro-aggregates ). In the process, physically protecting SOC by macroaggregate formation has been regarded to be an important process in SOC sequestration (Pulleman and Marinissen 2004). In addition, as drivers of the soil C cycling and sequestration (Lehmann and Rillig 2015), microorganisms contribute to the mineralization and biodegradation of SOC and the formation of new organic metabolites . The effect of microorganisms on C at aggregate scale has been studied widely (Six et al. 2004;Liu et al. 2020). A qualitative difference in microbial communities was found within macro-and micro-aggregates (Miller and Dick 1995), and the microbial biomass in macro-aggregates was suggested to make a larger contribution to SOC storage than in micro-aggregates (Six et al. 2004;De Gryze et al. 2005;Liu et al. 2020). The presence of a large number of arbuscular mycorrhizal fungi (AMF) was considered to be related to C accumulation within >1 mm aggregates, whereas Gram-positive (G+) bacteria and some nematodes increase C retention in smaller aggregates (Zhang et al. 2013). However, previous studies looking at soil samples still contained all aggregate size classes together, which included the process of mutual transformation between aggregates. Currently, there are no studies focusing on initial or independent aggregates. The relationship between the microorganisms residing within independent aggregates and the inner C content, as well as the link between this and aggregate size, were still unclear. This study is instrumental for better investigating how the nature of the microbial community within macro-and micro-aggregates affects the SOC content within each size class.
Crop residue addition can affect SOC retention at aggregate scale and alter its distribution in aggregates (Ding and Han 2014;Wang et al. 2019). As a source of energy, organic addition can stimulate the decomposition of native SOM (Zhang et al. 2018). Therefore, adding organic substrates can alter SOC content during the process of substrate switching through microbial metabolism (Woolf and Lehmann 2019). Furthermore, AMF in the soil not only indirectly stimulates fresh residue decomposition, but also suppresses the decomposition of old C (Wei et al. 2019). As a result, soil C can be accumulated or lost, depending on the balance between the microbial decomposition of new inputs and the protection of new C (Kravchenko et al. 2019). To better understand the relationship between SOC dynamics and the microbes associated with residue addition, it is important to quantify the utilization and fixation of new C during microbial processes (Jiang et al. 2019).
Previous studies can not distinguish in which group of aggregates the microbial processes affecting C accumulation occurred. We have added the 13 C-labeled maize straw to classified aggregate groups, and observed the variation of the inner C content. The difference in C accumulation between the straw addition and no straw addition can then be attributed to straw induced microbial community variations inside independent aggregates. The objective of this study was to determine the influence of straw addition on the relationship between microorganisms in independent aggregate groups and inner C content. We hypothesized that (i) straw application could promote the C accumulation in all independent aggregate groups, and (ii) more C content, including C straw , would be sequestrated in macro-aggregates than micro-aggregates, and (iii) the dominant microbial groups associated with C accumulation vary with aggregate sizes.

Sampling
Soil samples were collected from the Long-Term Residue Retention Experiment, initiated in 1993 at the [37][38] in northern China. The area has a mean altitude of 1100 m above sea level and a continental monsoon climate with a mean annual rainfall of 520 mm and a mean annual temperature of 7-8 °C. The dominant crop grown there is continuous spring maize, which covers more than 50% of the total crop-growing area in Shouyang. During the experimental period, maize was seeded at the end of April without any tillage and harvested in October. Plowing occurred once a year to a depth of about 20 cm. The area was weeded manually twice during the growth period. The study site had sandy clay loam soil, classified as a Calcaric-Fluvic Cambisol (ISS-CAS, 2003;IUSS, 2006). Soils for this study were sampled at a depth of 0-10 cm from the field with no added fertilizer in August 2017, and soil pH was 7.8, and SOC and total N contents were 19.3 and 1.1 g kg −1 , respectively. Up to five soil cores (10 × 10 cm) were sampled and pooled. The fresh soil was immediately taken to the laboratory and then separated manually along the natural cracks to obtain aggregate sizes of < 6 mm. Large pieces of plant material, stones, and visible soil fauna were picked out with tweezers. Then, samples were sieved into five sizes (> 5 mm, 2-5 mm, 1-2 mm, 0.25-1 mm, and < 0.25 mm). Some of the 'classified aggregates' were stored at -20 °C for biochemical analysis and some were air-dried for the analysis of the C contents, and the remaining samples were immediately subjected to the following experiment.

Incubation experiment
In order to determine whether the microorganisms derived from the initial aggregate promote C accumulation, an incubation of classified aggregates (>5 mm, 2-5 mm, 1-2 mm, 0.25-1 mm, and <0.25 mm) with 13 C-labeled maize straw application was conducted. The 13 C-labelled maize straw had a δ 13 C value of 690.5‰, C content of 40%. A total of 30 incubation microcosms were prepared in plastic containers without covers, each with a 200 g dry weight equivalent of soil aggregates. Six microcosms were prepared per aggregate size treatment. All microcosms were adjusted to 50-60% of water holding capacity in an incubation chamber to reach maximum microbial activity and equilibrated for 15 d at 25 °C in the dark. Soil moisture was achieved by adding deionized water and determined gravimetrically every week and adjusted as necessary. Once soil samples had adapted to this optimal environment, 13 C-labeled straw (1 cm height, 1 g per microcosm) was applied to each of 15 microcosms (five sizes, three replicates) and named the S treatment during the incubation. The other 15 microcosms were cultivated without straw, limiting the availability of organic material, in the so-called NS treatment. In addition, the original non-incubated soil aggregates named the CK treatment were used as the reference for the incubation experiment. The S-and NS-treated microcosms were incubated at 25 °C for 300 d. After incubation, some of each sample was removed from each incubator and air-dried to determine the organic C contents and analyze the amount of C derived from straw (C straw ) after picking out all visible debris. The remainder of the samples was stored at -20 °C for biochemical analysis.

Determination of organic C content and δ 13 C value of soil samples
The SOC content was analyzed using an elemental analyzer (C/N Flash EA 112 Series-Leco Truspec). Dissolved organic carbon (DOC) was extracted with distilled water (1:5 w:v). Samples were shaken for 2 h at 250 rpm, then centrifuged at 15,000 rpm for 15 min and filtered. The extracted solution was analyzed with a C analyzer (Multi N/C 3100, Analytic Jena, Germany). Readily oxidizable organic carbon (ROC) was tested with 333 mM KMnO 4 oxidation. The air-dried soil (passed through a 149 μm sieve) containing 15-30 mg C was mixed with 25 mL of 333 mM KMnO 4 and shaken for 1 h at 250 r min −1 , followed by centrifugation at 2000 r min −1 for 5 min. After dilution with deionized water (1:250), the supernatants were detected using a UV spectrophotometer at 565 nm and the ROC content was estimated by the variation in KMnO 4 concentration, assuming that 1 mmol KMnO 4 is equivalent to 9 mg C in the oxidation.
After ball-milling and acidifying with HCl (1 M), the δ 13 C values of the dried samples were determined on an elemental analyzer (vario PYRO cube, Elemantar) coupled with an isotope ratio mass spectrometer (Isoprime 100, Germany). The C isotope value (δ 13 C) was referenced to the international Pee Dee Belemnite standard (R PDB = 0.0112372) and expressed as δ 13 C.
where R sample is the ratio of 13 C/ 12 C in soil samples.

Analysis for organic C and 13 C contents
The proportion of C derived from straw (f; %) to SOC in soil aggregates with straw addition was calculated using the following Equation (2). The contents of aggregate fraction C derived from straw and from native SOC can be calculated according to the following Equations (3) and (4): where δ 13 C mix is the δ 13 C value of the soil aggregate with added straw, δ 13 C straw is the δ 13 C value in the straw residue, δ 13 C soil is the δ 13 C value in the soil aggregate without added straw, C mix is the SOC content in the aggregate with added straw, C straw and C native are the contents of SOC in the aggregate derived from straw and native SOC at the end of the incubation (300 d).

Analysis of enzyme activity
The activity of four soil enzymes, β-glucosidase (BG), β-xylosidase (BXYL), cellobiohydrolase (CBH) and leucine aminopeptidase (LAP), was measured in all soil samples using a 96-well microplate method (DeForest 2009;German et al. 2011). In brief, 1 g of fresh soil was weighed into a beaker with 125 mL of NaHCO 3 buffer (pH 8.0 for alkaline phosphatase) and stirred at 800 rpm for 2.5 min with a stir bar. The slurry was then set to stir at 100 rpm for 3 min. The continuously stirred soil slurry was transferred into black 96-well plates using an eight channel pipet. Every plate was set up for three soil analyses and included a blank well, reference standard well, quench well, negative control well, sample well, and sample control well. 4-Methyllumbelliferyl (MUB) was used as a substrate for BG, BXYL, and CBH activities, whereas 7-amino-4-methylcoumarin (AMC) was used for LAP activity. In detail, 250 μL of sodium acetate buffer were added to the blank wells; 50 μL of 10 mM MUB/AMC-linked substrate solution and 200 μL sodium acetate buffer were added to the reference standard wells; and 50 μL of a 200 mM MUB/AMC-linked substrate solution and 200 μL sodium acetate buffer were added to the quench wells. The sample control wells contained 200 μL soil slurry and a 50 μL sodium acetate buffer; while a 200 μL sodium acetate buffer and 50 μL of a 200 mM MUB/AMC-linked substrate were added to the negative control wells. Finally, the assay wells contained 200 μL soil slurry and 50 μL of a 200 mM MUB/AMC-linked substrate solution. All of the plates were then incubated for 3 h at 25 °C in a dark room/chamber. After incubation, 10 μL of 1 M NaOH were added to each well automatically to stop the reaction and enhance the fluorescence. After 1 min, the fluorescence was detected on a multilabel fluorescence reader (Tecan Infinite F200/M200) using l ¼ 355 nm excitation and l ¼ 450 nm emission filters.

Data calculation and statistical analysis
All data were statistically analyzed using SAS 9.4 for Windows 10. The contents of SOC, ROC, DOC and biomass of all microbial groups were expressed as the mean ± standard error. One-way ANOVA with Duncan's test was used to examine different aggregate size effects on SOC, DOC, ROC, enzyme activity, and microbial biomass. Two-way analysis of variance (ANOVA) with Duncan's test was used to examine straw treatment, soil aggregate size, and their interaction with all the organic C and microbial indicators. Significant differences are presented at p < 0.05, p < 0.01, or p < 0.001 levels. Pearson's correlations among all of the soil properties were calculated with the SAS 9.4 software. In addition, PCA and redundancy analyses (RDA) were performed to analyze the relationship between the soil microbial community composition and C-related indices in aggregates, and they were carried out using CANOCO 5.0 for Windows.

Distribution of C content in aggregates following the incubation
Relative to the non-incubated treatment (CK), the SOC in NS treatment decreased greatly in > 5 mm than in other aggregates ( Figure 1). Compared to CK, the SOC content in S treatment increased by 5.6-6.3% and 7.7%, respectively, in > 2 mm and < 0.25 mm aggregates. Conversely, the ROC and DOC contents significantly (p < 0.01) increased both in S and NS in all aggregate size classes after incubation (Figure 2(a,b)).
For the S and NS treatments, the results showed that the effects of straw treatment, aggregate size classes, and their interaction on the SOC, ROC, and DOC contents were significant (p < 0.05; Table 1). Straw addition increased SOC, ROC and DOC the more in the > 5 mm aggregates, 17.9% (Figure 1), than in any other size classes. Increases in ROC and DOC contents varied drastically in different aggregate classes under straw addition: the ROC and DOC contents (Figure 2(a,b)) were 3.4 g C kg −1 and 0.1 g C kg −1 in > 5 mm aggregates, respectively, which were significantly (p < 0.001) higher by 32.6% and 36.4% in S than in NS, while much lower in the other small aggregates. There was a significant (p < 0.01) increase in ratios of ROC to SOC content and DOC  to SOC content (Figure 2(c,d)) in > 2 mm aggregates under straw addition. As shown in Figure 3(a), the higher C straw content and proportion of C straw to SOC were observed in all aggregate sizes under straw addition and it increased gradually as the aggregate size increased. On the contrary, the proportion of C native to SOC in aggregates decreased gradually as the aggregate size increased (Figure 3(b)).

Soil enzyme activity in aggregates following incubation
Significant (p < 0.01) effects according to aggregate size were observed in the four enzyme activities (Figure 4). These activities increased almost in all sizes in S and NS treatments, relative to the CK. All of the four enzyme activities tended to be higher in S than in NS, being the highest in > 5 mm aggregates. After incubation, significant influences of aggregate size classes (p < 0.001) and straw treatment (p < 0.001) and their interaction (p < 0.05) on β-xylosidase (BG), cellobiohydrolase (CBH), and β-xylosidase (BXYL) activities were observed (Table 1). Table 1. Two-way ANOVA of the effects of the predictors C = straw treatment, A = aggregate size and C * A = their interaction on all considered variables describing soil carbon, enzyme activities and microbial variables in NS and S treatments. The importance of the effects is reported through Fisher's F values and the significance via the coded p values: ***: p < 0.001; **: p < 0.01; *: p < 0.05; ns: not significant.

The microbial biomass in aggregates after the incubation
Soil microbial biomass (measured as total PLFA concentration) in all aggregates was significantly (p < 0.001) lower in NS than in CK (Table 2). After incubation, microbial biomass varied greatly among aggregate sizes and between NS and S treatments. In S, the total PLFA significantly (p < 0.001) increased by 36.9-53.9% in >2 mm aggregates, which was greater than that in < 2 mm aggregates (Table 2). Fungi were significantly (p < 0.01) increased (by 29.8-83.7%) in > 2 mm aggregates but reduced (by 30.8-37.6%) in < 2 mm aggregates, showing the same trend as G-bacteria (Table 2). In particular, AMF could no longer be detected in < 2 mm aggregates after straw addition. In S, the greatest increases in fungi and bacteria biomass were observed in > 5 mm aggregates, where the fungi biomass increased by 83.7%, more rapidly than that of bacteria (which increased by 43.8%). The G+ bacteria and actinomycetes biomasses increased to varying degrees in all aggregates. The ratio of G+ to G-bacteria (G+/G-) was significantly (p < 0.001) increased in all aggregates after straw addition and was higher in < 2 mm aggregates than in those > 2 mm in size ( Table 2). The trend in the ratio of bacteria to fungi (B/F) was exactly the opposite to that of G+/G-bacteria and was significantly (p < 0.001) higher in > 5 mm aggregates after straw addition but significantly (p < 0.001) reduced in < 2 mm aggregates. Overall, bacteria, actinomycetes, fungi, and the two ratios were all affected by aggregate size classes (Table 1).

Relationships between microbial community and organic C indicators in aggregates
The PCA revealed that the soil samples were divided into two groups (> 2 mm and < 2 mm aggregates), based on changes in microbial indices in S and NS ( Figure 5). Considering the  CK: the control treatment without incubation; NS: the treatment without straw addition; S: the treatment with straw addition. B/F: the ratio of bacterial and fungal PLFA; G+/G-: the ratio of G+ and Gbacterial PLFA; AMF: arbuscular mycorrhizal fungi; -: no AM fungi. Values within the same column followed by different lowercase letters indicate significant differences among the different aggregate size classes in the same incubation time at p < 0.05.
differences between > 2 mm and < 2 mm aggregates ( Figure 5), soil microbial contributions to C accumulation were analyzed separately by RDA for the aggregates of > 2 mm and < 2 mm after the incubation ( Figure 6). The amount of variability explained by all the canonical axes in > 2 mm and < 2 mm aggregates were 76.1% and 75.5%, respectively. As shown in Figure 6, SOC content was significantly (p < 0.05) associated with microorganism indicators (microbial PLFA, bacteria, actinomycetes, B/F and G+/G-bacteria) in both > 2 mm and < 2 mm aggregates. More specifically, the SOC was more strongly associated with both fungi and G-bacteria than other microbial indicators in > 2 mm aggregates (Figure 6(a)), while was associated more with G+ bacteria in < 2 mm aggregates ( Figure 6(b)). In addition, DOC and ROC were positively (p < 0.05) correlated with the microorganism indicators (except for AMF and B/F) in >2 mm aggregates, but not in < 2 mm aggregates (Figure 7). Meanwhile, the ROC was positively (p < 0.05) correlated with the C-related enzymes (BG, CBH and BXYL) in < 2 mm aggregates, but not in > 2 mm aggregate (Figure 7).

Discussion
In the previous study, the SOC contents in most aggregates were found lower in NS than in CK, due to the mineralization of some C during the incubation process ). The C sequestered in soil could be turned over by exoenzyme activity or respiration processes of microorganisms, which lead to continuous C loss (Kindler et al. 2006). During the mineralization process, DOC and ROC increased to provide C sources for starving microorganisms after the incubation. Comparing NS to S, significant impacts by aggregate size and straw addition were shown in the SOC content and C straw , indicating that straw addition can promote C accumulation in aggregates. Straw addition stimulated microbial mineralization of native SOC in soil due to the priming effect (Figure 3(b)), and it was affected by the size and structure of aggregates (Juarez et al. 2013;Sarker et al. 2019). Although most of the straw-derived C will be mineralized and released in the form of CO 2 , part of the remaining C can be converted and retained in soil in the form of SOC by microorganisms (Xia et al. 2018). Therefore, an appropriate amount of straw addition can balance part of soil C losses. This is in line with the results presented by other studies . Degradation of added straw and also of the native SOM maybe stimulated by microbial biomass (Blagodatskaya and Kuzyakov 2008), thus leading to the immobilization of C (Zhang et al. 2018). The increases in DOC, ROC, and C straw contents, in particular, were significantly greater in > 5 mm aggregates than in smaller ones, implying that >5 mm aggregates were more capable of improving unstable C and new C accumulation. Variations in the activity and distribution of microorganisms in different sized aggregates could lead to differences in CO 2 emissions and C sequestration during the biological process, which is related to the microenvironment and pore structure inside these aggregates (Jha et al. 2012). A previous study found a similar result: a SOC increase was observed in soil macro-aggregates, not in micro-aggregates and silty clay (Srinivasan et al. 2012), in the 0-10 cm layer with straw retention and macro-aggregates contained more easily decomposable organic C (Puget et al. 2000). In that study, maize residue C was enriched in the particulate organic matter located in macro-aggregates. With good air permeability, the pore structure inside macro-aggregates can provide a convenient place for microbial activity under treatment with added organic substrates (Negassa et al. 2015). A great abundance of labile C can support faster growing microorganisms in macro-aggregates than in micro-aggregates (Bach et al. 2018;Lupwayi et al. 2001) which in turn accelerates the decomposition of straw. It was supported by a positive relationship between SOC, DOC, ROC and microbial PLFA in > 2 mm aggregates (Figure 7(a)). However, a higher mineralization rate of SOC, especially the native C, offsets a big increase in SOC in macro-aggregates (except for 2-5 mm aggregates). As a labile pool (Six et al. 2000), macroaggregates were reported to be more highly susceptible to mineralization than micro-aggregates (Rabbi et al. 2014). In micro-aggregates, the highly recalcitrant SOC and limited oxygen supply (Kuka were not beneficial to C mineralization. Therefore, higher stable C accumulation was observed in < 0.25 mm than in > 2 mm aggregates ( Figure 1) after straw addition, which did not support our second hypothesis. There was no significant difference in SOC content of 0.25-2 mm aggregates between the CK and S, indicating that the rates of C mineralization and sequestration were in balance. Nevertheless, it is unclear whether the trend for C accumulation in soil aggregates is the same as our results in different soil textures (Gao et al. 2017). Therefore, it will be worth exploring whether more C can be sequestrated in soil rich in differently sized aggregates in the future.
In order to obtain C for microbial survival, high activities of BG, CBH, BXYL, and LAP were found in all aggregates after incubation (Figure 4), which could explain why DOC and ROC increased during incubation. Straw treatment and aggregate size had significant (p < 0.05) impacts on all enzyme activities (Table 1). Straw amendment could cause rapid stimulation of soil microbes (Golchin et al. 1994;Liang et al. 2018). In order to accelerate the degradation of straw, microbes could produce the C-related enzymes (BG, CBH, and BXYL) and nitrogen-related enzyme (LAP) to regulate the carbon-to Figure 7. Correlation coefficients among the properties of soil organic C, microbial indices and enzyme activities after the incubation in > 2 mm (a) and < 2 mm (b) aggregates, respectively. *: p < 0.05, **: p < 0.01, ***: p < 0.001.
-nitrogen ratio of the microenvironment (Tiemann and Billings 2011). The macro-aggregates could provide a suitable condition for rapid microorganism activities. This might be explained in part by the high soil enzyme activities in > 5 mm aggregates during straw residue degradation, especially of BG, which is the most easily detectable and abundant enzyme (Turner et al. 2002). Notably, the C-related enzyme activities were significantly (p < 0.05) related to ROC in < 2 mm aggregates ( Figure  7(b)), due to obtaining the available C source for the microbial growth through secreting the corresponding enzymes.
The microbial biomass was lower in NS than in CK (Table 2), because the original microbial habitat was disrupted, which in turn affected the growth of soil microorganisms after the classification of aggregates. After the incubation, PLFA profile analysis showed that the microbial biomass was greatly increased by the addition of straw (Rousk and Bååth 2007). The increase in total PLFA biomass was positively (p < 0.05) correlated with SOC content (Figure 7), which indicated that the rapid degradation of straw in aggregates could also stimulate the growth of microorganisms (Negassa et al. 2015). Fungi and G-bacteria increased significantly in >2 mm aggregates and decreased in < 2 mm aggregates (Table 2). However, only G+ bacteria and actinomycetes increased in <2 mm aggregates, indicating that G-bacteria and fungi respond more strongly to straw addition in > 2 mm aggregates, while G+ bacteria and actinomycetes are more viable in < 2 mm aggregates. Since the nature and mass of the substrate can affect microbial metabolism (Madan et al. 2002), fungi and G-are more inclined to live in a resource-rich environment (Margesin et al. 2009;Bahn et al. 2013) inside macro-aggregates with more unstable C (Lehmann and Rillig 2015), while G+ bacteria compete better for limited environmental resources (Margesin et al. 2009;Bahn et al. 2013). Due to the larger volume of mycelium, fungi communities stimulated by organic substrates tend to colonize macro-aggregate structures (Six et al. 2006;Murugan et al. 2019). This might explain why more AMF was observed in >2 mm aggregates after straw addition, which was considered to be related to decomposition of the recalcitrant ligno-cellulose matrix by stimulating the activity of hyphosphere bacteria or producing a wide range of extracellular enzymes (De Boer et al. 2005) (Figure 4). This was also supported by the positive (p < 0.05) correlation between AMF and SOC in this study (Figure 7(a)). In addition, it can also explain why G+/G-bacteria increase more in micro-aggregates (< 2 mm) than in macro-aggregates (> 2 mm) under straw application, while the B/F, as one of the indicators of ecosystem buffer capacity, displayed the opposite (Bardgett and McAlister 1999). In further research, the dominant microbial indices and strains that can affect C accumulation in different aggregates need to be deeply explored.
SOC content within all aggregates was closely (p < 0.05; Figure 7) related to the activities of inner microorganisms (microbial PLFA, bacteria, and actinomycetes). As the important components of C, DOC and ROC were significantly affected by microbial PLFA, which contributed to SOC in > 2 mm aggregates, not in <2 mm aggregates (Figure 7). This is in line with other research, that the DOC and macro-aggregate fractions (> 0.25 mm) were correlated with soil microbial diversity under straw addition (Bu et al. 2020). A plausible explanation could be that the macro-aggregates are more suitable for an efficient and rapid decomposition of the straw by microorganisms, thus stimulating the increase in labile organic C (Negassa et al. 2015;Dhaliwal et al. 2020). As mentioned above (Figure 6), the significantly raised SOC level in > 2 mm aggregates was accompanied by a significant increase in the biomass of fungi and G-bacteria (p < 0.01), whereas the rise of SOC in <2 mm aggregates was accompanied by G+ bacteria (p < 0.01). There are two possible explanations for this trend. The first is that fungi and G-bacteria strongly depend on C derived from straw residues (Bai et al. 2016), while G+ bacteria is more efficient with recalcitrant SOM decomposition (Waldrop and Firestone 2004), because resistant compounds are more readily degraded by their exoenzymes (Brant et al. 2006). Fungi were observed to be able to absorb and utilize plant material more rapidly and efficiently than bacteria, especially in incorporating straw-derived C (Koechli et al. 2019), which was preferentially accumulated in the fungal biomass (Štursová et al. 2012;Müller et al. 2016). The second is that fungi-dominated soils have more microbial residues in the SOM than those that are bacteria-dominated (Kallenbach et al. 2016). The contribution of total microbial residues to SOC is stronger in macro-aggregates than in micro-aggregates (Murugan et al. 2019), while the microbial residue C in soil is much higher than the living microbial biomass (Simpson et al. 2007). Hence, we conclude that fungi and G-bacteria were more important to increase C accumulation in > 2 mm aggregates, while G+ bacteria dominated in that role for < 2 mm aggregates. Among all of the aggregates, the assistance of actinomycetes is indispensable during the decomposition of macromolecular plant materials (Dou and Wang 2011). This indicated that C in aggregates was closely related to its own microorganisms, and this relationship was directly affected by aggregate sizes. Future studies should consider that whether it is possible to assist the improvement of C accumulation at aggregate scale by selecting different microbial fertilizers.

Conclusions
This study aimed to evaluate the C accumulation in independent aggregates and the contribution of internal microorganisms to this under straw application. Our results showed that straw addition significantly increased SOC content in > 2 mm and < 0.25 mm aggregates, and greater stable C contents were found in < 0.25 mm aggregates than in > 2 mm aggregates. The activities of microorganisms in independent aggregates contributed to inner C accumulation under straw addition. Fungi and G-bacteria were more important to increase C accumulation in > 2 mm aggregates and G+ bacteria dominated in < 2 mm aggregates. Our results imply that the specific microbes naturally occurring in soils could support an approach to increase C accumulation in soils that were regulated by aggregate sizes. Moreover, further studies about long-term experiments are warranted to better evaluate the potential of soils rich in differently sized aggregates to store SOC.