Spatio-temporal dynamics of arbuscular mycorrhizal fungi and soil organic carbon in coastal saline soil of China

A comprehensive understanding of the relationship between arbuscular mycorrhizal (AM) fungi and coastal saline soil organic carbon (SOC) is crucial for analysis of the function of coastal wetlands in soil carbon sequestration. In a field experiment, the temporal and spatial dynamics of AM fungi, glomalin-related soil protein (GRSP) – which is described as a N-linked glycoprotein and the putative gene product of AM fungi, SOC, and soil aggregates were investigated in halophyte Kosteletzkya virginica rhizosphere soil of coastal saline areas of North Jiangsu, China. Soil samples were collected from a depth of up to 30 cm in two plantation regions from August 2012 to May 2013. Results showed K. virginica formed a strong symbiotic relationship to AM fungi. AM colonization and spore density were the highest in the 10–20 cm soil layer of Jinhai farm in August 2012, because of the presence of numerous fibrous roots in this soil layer. The total GRSP and SOC were the highest in the 0–10 cm soil layer in May 2013 and November 2012, respectively. Correlation coefficient analysis revealed that AM colonization and spore density were positively correlated with total GRSP. Meanwhile, total GRSP was significantly positively correlated with large macroaggregates (>3 mm), SOC, total P, Olsen P, and soil microbial biomass carbon (SMBC), but negatively correlated with microaggregates (<0.25 mm), soil EC, total N, and pH. SOC was positively correlated with spore density, large macroaggregates, small macroaggregates (2–0.25 mm), alkaline N, and SMBC and negatively correlated with microaggregates, EC, pH, and total K. Although it may be a statistical artifact, we found an interesting phenomenon that there was no significant correlation between soil aggregates and AM colonization or spore density. Hence, total GRSP is a vital source of saline soil C pool and an important biological indicator for evaluating coastal saline SOC pool and soil fertility, while AM colonization or spore density may not be.

Spatial and temporal distribution of total GRSP. In Qingkou saltern and Jinhai farm, the total GRSP content decreased with increasing soil depth and ranged from 0.08 mg·g −1 to 0.69 mg·g −1 and 0.32 mg·g −1 to 2.43 mg·g −1 , respectively ( Fig. 2A,B). The total GRSP content in K. virginica plantation is significantly higher in Qingkou saltern than that in control plot in the same layer (p < 0.05). Irrespective of soil layer, the total GRSP content decreased from August 2012 to the lowest values in February 2013 and then returned to high levels in May 2013. edaphic factors. Ten edaphic factors were assessed to comprehensively determine the local soil condition. The results were presented by a radar chart. The plantation soil pH, available nitrogen, total nitrogen, and soil microbial biomass carbon (SMBC) are all significantly higher in Qingkou saltern than in control plot from August 2012 to May 2013 (p < 0.05, Fig. 3A-D). The SOC content ranged from 6.34 mg·g −1 to 14.2 mg·g −1 in the plantation. The plantation SOC in the 20-30 cm soil layer is significantly higher than that in the 10-20 cm soil layer in August and November 2012 and significantly lower than that in control plot soil in February 2013 (p < 0.05). The electrical conductivity (EC) in all plantations is significantly lower than that in control plot soils in each soil layer (p < 0.05). The EC increased in the plantation with increasing depths of the soil layer. In August 2012 and May 2013, the Olsen phosphorus of the plantation is significantly higher than that of control plot in each soil layer (p < 0.05). The total phosphorus of the plantation is significantly higher than that of control plot (p = 0.024), except for that in February 2013. However, no significant difference was observed between K content in the plantation and control plot in each soil layer (p > 0.05).
In Jinhai farm, soil pH was not significantly different between the plantation and control plot from August 2012 to May 2013 ( Fig. 3E-H). The EC and Olsen potassium in the plantation are significantly lower than those in control plot in each soil layer (p < 0.05). By contrast, the total nitrogen, SMBC, and SOC in the plantation are significantly higher than those in control plot in each soil layer in four measurements (p < 0.05), except for SOC in the 20-30 cm soil layer during November 2012 and February 2013. The SOC content ranged from 2.69 mg·g −1 to 10.71 mg·g −1 in the plantation. The Olsen phosphorus in the plantation is significantly higher than that in control plot in each soil layer during November 2012 and May 2013 (p < 0.05).
Spatial and temporal distribution of soil aggregates. In Qingkou saltern, irrespective of plantation or control plot in the four periods, soil aggregates are mainly in the fractions of > 5 mm, 1-0.25 mm, and  microaggregates (MI) (Fig. 4A-D). In August 2012, all soil aggregate fractions between the plantation and control plot exhibited no significant difference. In November 2012, the soil aggregate contents of 3-5, 2-3, and 1-2 mm fractions in the plantation are significantly higher than those in control plot in the three soil layers (p < 0.05). However, the content of > 5 mm fraction in the plantation is significantly lower than that in control plot in the 0-10 cm soil layer (p = 0.026). The large macroaggregates (LM) and small macroaggregates (SM) contents in the plantation are significantly higher with time than those in control plot in February 2013, but the MI content in the plantation is significantly lower than that in control plot in the three soil layers. In May 2013, the soil aggregate contents in 3-5, 2-3, and 1-2 mm fractions are higher than those in control plot, and the MI is significantly lower than that in control plot in each soil layer.
In Jinhai farm, the results of the parameters tested are similar among the four periods ( Fig. 4E-H). In the plantation, soil aggregates are mainly in > 5 mm and MI. In control plot, soil aggregates are mainly in MI in the three soil layers during the four periods. In all soil layers and periods, the contents of LM and SM in the plantation are significantly higher in Jinhai farm than those in control plot, and the MI content in control plot is significantly higher than that in Jinhai farm in the plantation (p = 0.031).

AM fungus and edaphic factors. The inoculation of AM fungi can form good symbiotic association with
K. virginica roots under various salt stress conditions in greenhouse 35,36 . This study demonstrated that K. virginica in the coastal saline soil of North Jiangsu could also form symbiotic relationships to AM fungi. He et al. reported that the maximal value of the AM colonization and spore density occurred at the 0-10 cm soil layer in farming-pastoral zone 37 . In this study, the AM colonization rate and spore density in the 10-20 cm soil layer are higher than those in the 0-10 cm soil layer because of the presence of numerous fibrous roots in the former. For the same reason, K. virginica possessed numerous fibrous roots during August 2012, and the highest AM colonization rate was detected during this period. AM fungal hyphae in roots can be related to the absorption and translocation of low-mobility nutrients, such as P, in soil and water from distant areas that are inaccessible to plant roots 38 . In the present study, increased hyphal colonization could help in efficient nutrient and water absorption and transportation from soil to host, leading to increased nutrient demands in August. This result is in accordance with the reports that the maximum abundance of AM fungi occurs during summer and the colonization declines during winter and early spring [39][40][41] . However, Füzy et al. reported that AM colonization peaked in late spring to early summer and exhibited a second peak later in autumn 42 . This may be due to different soil and air temperature changes and plant growth characteristics leading to the difference of AM colonization.  www.nature.com/scientificreports www.nature.com/scientificreports/ In this study, the AM colonization and spore density in the two research sites decreased with increasing soil depth (Fig. 1A,B), consistent with the results of He et al. 37,43 , Taniguchi et al. 44 , Wang et al. 45 , and Zhang et al. 41 . This phenomenon might be due to the usual distribution of the main roots of K. virginica in 0-20 cm soil. Moreover, the EC and pH increased but the oxygen concentration decreased with increasing soil depth ( Fig. 3A-G). The correlation coefficient analysis results also showed significantly negative correlation of AM colonization and spore density with pH and EC (Table 1). These results corroborate reports on reduction in root mycorrhizal rate at high salinity levels 35,36,[46][47][48][49] . This may be due to salinity be able to hamper colonization capacity, spore germination, and growth of AM fungal hyphae 46,50,51 . Füzy et al. suggested that drought may play an important role in governing mycorrhizal activity in saline habitats because AM fungi may help plants acquire water from soil 42 ; in addition, the formation of additional arbuscules may facilitate the transfer of water and nutrients to plants. Although mycorrhizal colonization is reduced with increasing salt levels, the symbiosis between AM fungi and halophytes may be strengthened in saline environments once the partnership has been established 52 .
Bencherif et al. found that the number of AM fungal spores increased with soil salinity level 49 . Hildebrandt et al. 53 and Becerra et al. 54 also reported different halophytic plants associated with numerous AM fungus spore populations in saline soils. In addition, Aliasgharzad et al. 13 reported that salt stress can stimulate AM fungi sporulation. Bencherif et al. 49 suggested that sporulation can be considered a resistance behavior to help AM fungi survive adverse environmental conditions. In contrast to these findings, the present results indicated a significant reduction in spore count as salinity level increased. This phenomenon could be explained by the inhibitory nature of high salinity levels on spore-producing hypha 55 . Zhang et al. 41 suggested that the high spore number found in October coincides with the end of growth season and can also reflect the accumulation of spores throughout the growing season. In the present study, the highest spore density was observed in Qingkou saltern and Jinhai farm during November and August (Fig. 1), respectively. This finding could be due to the effects of the differences in soil type, edaphic factors, and climate characteristics in the two research areas on spore density and hyphal growth; thus, the number of spores might be different in various types of soil 56 .
In contrast to previous research 41,57 , the present work showed that seashore saline soil total nitrogen was negatively correlated with AM colonization and spore density. This finding could be due to the higher amounts of N that accumulated in mycorrhizal plants than in non-mycorrhizal plants 47 . With AM fungi colonization, K. virginica mycorrhizal became stronger and intensively absorbed nitrogen from saline soil, leading to a shortage of supply of N in this period. Additionally, the entire soil microbial community became more active with the improvement in soil physico-chemical properties and soil fertility, resulting in significantly intensified ammonification and nitrification and accelerated organic nitrogen mineralization. In the present study, the available P content in plantation soils is higher than that in control plot and was positively correlated with AM colonization. This trend might be explained by the fact that enhanced AM colonization can increase the available P content 36 . Magallon-Servín et al. 58 suggested that AM can produce numerous organic acids, which can solubilize mineral P. For example, the metabolic activity of the K. virginica root system enhances the reproduction of AM fungi and other soil microorganisms, thereby accelerating the release of soluble phosphatase into the soil 36 .
AM fungus and GRSP in coastal saline soil. AM colonization is positively correlated with GRSP content 41,59-63 . Consistent with previous works, the present study reported that AM colonization, similar to spore density, was positively correlated with GRSP, illustrating the presence of the majority of GRSP in AM fungal hyphae and spores 64 . Similar to the findings of He et al. 43 and Taniguchi et al. 44 , the current results showed that the total GRSP content in each plantation decreased with soil depth. In Qingkou saltern, the average total GRSP contents in the 30 cm soil layer are 35.44% and 62.5% lower than those in the 20 and 10 cm soil layers, respectively. In Jinhai farm, the average total GRSP contents in the 30 cm soil layer are 33.98% and 56.75% lower than those in the 20 and 10 cm soil layers, respectively. These findings may be due to the distribution depth of the K. virginica root. The usual distribution depth of the mature K. virginica roots is 20 cm, which supports the result of Taniguchi et al. 44 and confirms that the reduction of root biomass may cause the decrease in AM fungi colonization and GRSP content with soil depth. The microbial activity was strong in highly fertile soil in the 0-20 cm layer, especially at 10 cm, and favored microbial reproduction expansion and GRSP accumulation. Thus, GRSP content was concentrated in the 0-20 cm soil layer, especially in 0-10 cm.
GRSP is a very stable biomolecule with a half-life of 6-42 years in soil 22,63 , but concentrations can fluctuate throughout the growing season 65 . Moreover, inoculation of AM fungi stimulated the synthesis of easily extractable GRSP (EE-GRSP), a fraction of soil GRSP 11 , which is newly produced by the hyphae and spores of AM fungi 62 . In the present study, the highest concentrations of total GRSP in saline soil were observed in May or August, and the lowest was detected in February. These results are consistent with those reported in previous studies 43,65 , which indicated that the highest total GRSP content was observed in May. Zhang et al. 41 reported that both total GRSP and EE-GRSP contents were the highest in August and lowest in October. This finding may be due to the corresponding mycorrhizal infection in different sampling periods. The local area has more suitable temperature and water and light conditions in May and August, and the plants are in a vigorous growth period. In addition, the accumulation of total GRSP decreased possibly due to the decomposition of the labile part of GRSP (EE-GRSP). Hence, the total GRSP content fluctuated seasonally because AM fungi and K. virginica were affected by seasonal changes, demonstrating time heterogeneity.
Different kinds of soil texture exhibit varied influences on total GRSP content. In agricultural, natural grassland, desert, and mangrove forest ecosystems, the total GRSP contents are 1-21 mg·g −1 , 4.5-5.0 mg·g −1 , 2.49-4.11 mg·g −1 , and 0.46-1.38 mg·g −1 , respectively 21,25,41,59 . In the present study, the total GRSP content in the K. virginica plantation in Jinhai farm is 0.85 m·g −1 to 2.43 mg·g −1 , which is lower than that in natural grassland soil but higher than that in the mangrove forest ecosystem. This finding indicates that K. virginica could establish good symbiotic association with AMF 35,36 and contribute to the restoration of saline soil. The total GRSP content is higher in Jinhai farm than in Qingkou saltern (Fig. 2) because of the heavy clay soil texture in the latter and the Scientific RepoRtS | (2020) 10:9781 | https://doi.org/10.1038/s41598-020-66976-w www.nature.com/scientificreports www.nature.com/scientificreports/ sandy loam in the former. AM fungi obtained more suitable soil conditions for growth in the Jinhai farm soil, thus demonstrating higher AM colonization and spore density in August, 2012 (Fig. 1B). Furthermore, GRSP concentration may be influenced by soil mineral and fertility. In the current study, total GRSP showed a significantly positive relationship to SOC, total phosphorus, Olsen phosphorus, and SMBC, consistent with the findings of many researches 22,41,43,57,66 . This result could be due to the requirement of mycorrhizal fungal growth and metabolism 67 . Rillig reported that total GRSP is an important part of soil N 59 ; in addition, many studies indicated that GRSP content is positively correlated with available nitrogen 41,43,57 . In the present study, total GRSP was significantly correlated with available nitrogen and negatively correlated with soil total nitrogen. This can be explained by former results that there is a higher portion of available nitrogen in the plantation plots.
The amount of GRSP was negatively correlated with soil EC and pH. Our results are strongly supported by other authors, who reported the significant negative effects of soil salinity on GRSP 48,68,69 . However, Ji et al. 59 and Lovelock et al. 70 found that GRSP levels in soil and in vitro cultures were negatively correlated with hyphal length, suggesting that the production of this compound may be a stress response. Hammer and Rillig 8 reported that the GRSP content increased up to a threshold level of 150 mM NaCl but decreased with further increase in the salt concentration. Garcia et al. reports that maximum salinity stress (2.0 dS m −1 ) increased 6% and 18% GRSP production than 1.0 dS m −1 and 0.6 dS m −1 , respectively 71 . Hence, GRSP may be involved in the inducible stress responses of AM fungi to salinity. GRSP and soil aggregates in coastal saline soil. The distribution patterns of soil aggregates differed in the two sites. In Qingkou saltern, both in plantation and control plot, soil aggregates in all soil layers are mainly in > 5 mm, 1-0.25 mm, and MI. In Jinhai farm, soil aggregates are mainly in > 5 mm and MI in the plantation, and 95% are MI in control plot. This finding could be directly due to difference in soil texture in the studied areas; that is, Qingkou saltern has heavy clay with high soil bulk density and low nutrition, whereas Jinhai farm has sandy loam and up to 65.50% sand ( Table 2). Soil in these two sites showed very poor granulation structure. Such results suggest that soil aggregation decreased as levels of sand and carbonate increased, likely due to concurrent decreases in levels of clay in the soil 4 . In addition, the contents of macroaggregates (>0.25 mm) effectively increased in the plantation soil compared with that in CK soil. Macroaggregates were negatively correlated with microaggregates (<0.25 mm), suggesting that the former was gradually formed from the latter during planting K. virginica periods. This finding may be related to amount of fibrous roots can wind up dispersion soil particles and form large aggregates 72 . Furthermore, the stability of soil aggregates is probably affected more by the direct and indirect actions of the plant-fungal system, rather than by plant root metabolism 73 . In present study, SM content was positively correlated with SMBC, and LM content was positively correlated with total GRSP; as such, microorganisms are one of the most active and important biological factors.
Wright et al. 74 indicated that glomalin is an insoluble glue-like substance only released by AM fungi into the soil environment during hyphal turnover and after the death of the fungus 64 ; this compound may contribute to binding within microaggregates and macroaggregates. Wright and Anderson 75 found that aggregate stability and GRSP were linearly correlated (r = 0.73, n = 54, p < 0.001) across all treatments from different cropping systems. In relation to this finding, the present results indicated the positive correlation between LM and TG, with coefficient of 0.566 (p < 0.01), and the negative correlation between MI and total GRSP, with coefficient of 0.568 (p < 0.01). Hence, soil GSRP presents strong cementing ability, inducing the formation of aggregates with increased structural stability 63,76 . In addition, external hyphae can promote the formation of soil aggregates by physical tangles and changing soil dry-wet circulation 9,77 . Bearden and Petersen 19 suggested that mycorrhizae primarily influence the stability of macroaggregates. By contrast, no significant correlation was found between soil aggregate and AM colonization or spore density (p > 0.05) in the present study. This may be due to the formation of soil aggregates takes a long time, but the fibrous roots used to check AM colonization have a shorter time to grow from the main and lateral roots of K. virginica. Meanwhile, it may also be affected by the detection methods and environmental factors, and there is a large gap between the test results and the actual situation of AM colonization or spore density. Rillig et al. 59 reported that the direct effect of GRSP on aggregate stability is higher www.nature.com/scientificreports www.nature.com/scientificreports/ than the total (direct and indirect) effect of hyphae on soil aggregate stability in oak-hickory (Quercus-Carya) forests of eastern North America; this phenomenon also can partly explain the weak correlation between soil aggregate and AM fungi. Barto et al. 78 suggested that abiotic factors (mowing, grazing, and fertilization) can be more important for determining soil aggregation than biotic factors (root length and mass, AM colonization, extraradical AMF hyphal length), especially in highly aggregated soils. Hence, contrary to He et al. 's 43 research conclusion that spore density, colonisation of hyphae, the contents of glomalin can be used as parameters to monitor the development of organic carbon dynamic and nutrition cycle in sand soil, AM colonization or spore density may not be used to evaluate coastal saline SOC pool and soil fertility. Furthermore, we cannot ignore another possibility that the lack of correlation between soil aggregate and AM fungi was a statistical artifact. The mechanism underlying this phenomenon has not been fully understood yet. The results also showed that SM content was negatively correlated with soil EC, similar to the report of Lax et al. 79 . This probably due to large amount of Na ions complicate the formation of colloidal aggregates in coastal saline soil 8 .
AM fungus and SOC in coastal saline soil. Coastal soil ecosystems dominated by plants play a critical role in the global sequestration of C 80 , and the SOC pool acts as a crucial regulator of C fluxes between biosphere and atmosphere. Our previous studies revealed that plant biomass improved by AM fungi inoculation might be beneficial to coastal soil ecosystems in sequestering large amounts of C 81 . We also found that AM fungi inoculation could strongly promote plant dry biomass and nutrient uptake by K. virginica, regardless of salinity level 36,81 . In the present work, after planting K. virginica for 3 years, the SOC in the plantation is significantly higher than that in control plot in each soil layer in several periods (p < 0.05). This finding could be due to the ability of K. virginica to sequester SOC within numerous living biomass aboveground and belowground, litter, and dead wood 81 . Moreover, the mycorrhizal roots of K. virginica can provide AM fungi with photosynthetic C, which, in turn, is delivered to soil via fungal hyphae 38 . This finding is confirmed by the positive correlation between spore density and SOC (p < 0.05) ( Table 1). Rillig 22 stated that the C in GRSP largely contributes to the total soil C pool. In tropical soils, the amount of C in GRSP is estimated to be 37%, which represents 3% of soil C pools 67 . The total GRSP-to-SOC ratio of sand and soil is between 34.88% and 66.85% 41 . In the present study, total GRSP was positively correlated with SOC (p < 0.05), with the highest total GRSP-to-SOC ratio of up to 53.29% in Jinhai farm. Hence, total GRSP is a vital source of saline soil C pool, and important biological indicator for evaluating coastal saline SOC pool and soil fertility.
In addition, as insoluble glue, GRSP can stabilize soil aggregates and significantly reduce organic matter degradation by protecting labile compounds within soil aggregates 75 . In agreement with these findings, the present study showed that LM was positively correlated with SOC and total GRSP (p < 0.01). Meanwhile, the negative correlation of MI content with SOC and total GRSP (p < 0.01) illustrates the importance of soil aggregate stabilization in inhibiting the degradation of organic matter and promoting soil C sequestration 4-6 . In our previous study, the introduced microbe (Glomus mosseae, and a phosphate-solubilizing fungus Mortierella sp.) can collaborate with indigenous microorganisms to promote the humification of organic materials 81 . Plant-AM fungi mutualism can improve the reestablishment of vegetation in bare saline-alkaline soil and drives the vegetation restoration to a community dominated by the original species 81,82 . In the present work, the significant positive relationship among AM colonization, total GRSP, SMBC, and SOC suggests that AM fungi play a crucial role in stimulating the growth of indigenous microorganisms, enhancing the stability of saline SOC pool and soil fertility, and promoting the reestablishment of vegetation 41 when K. virginica was introduced into the coastal saline soil of North Jiangsu and other sites in China.
The SOC contents in the 20-30 cm soil layer of the two plantations intensively fluctuated following spatio-temporal dynamics and are even lower than that in control plot. These results suggest that SOC in the 20-30 cm soil layer of K. virginica plantation demonstrated instability and spatio-temporal heterogeneity in coastal saline soil. However, further research is required to understand the relative mechanisms. conclusion A strong symbiotic relationship was found between K. virginica and AM fungi after their introduction into the coastal saline soil of North Jiangsu for 3 years. This study demonstrated highly variable temporal and spatial patterns in the dynamics of AM fungi, total GRSP, and SOC, which were affected negatively by soil salinity. The significantly positive relation among AM fungi, total GRSP, and SOC were also confirmed. The results also revealed that soil aggregate stabilization, especially soil large macroaggregates (>3 mm), is crucial to maintain the stability of total GRSP and saline SOC pool. Although it may be a statistical artifact, a new phenomenon that no significant correlation exists between soil aggregate and AM fungi was observed, which contradicts previous findings. Hence, total GRSP is a vital source of saline soil C pool and an important biological indicator for evaluating coastal saline SOC pool and soil fertility, while AM colonization or spore density may not be. Future research must investigate the relation between AM fungi and saline soil aggregate to improve understanding of their roles in coastal ecosystems.

Materials and Methods
Study site. North Jiangsu experiences a typical temperate and monsoonal climate with four clearly distinct seasons. The study sites selected are Qingkou saltern of Lianyungang City (34°45′N, 119°11′E) and Jinhai farm of Yancheng City (32°59′N, 120°46′E), which are located in North Jiangsu and possess the highest annual average temperatures of 25.8 °C and 28.1 °C, respectively, in July and the lowest annual average temperatures of 0.7 °C and 2.0 °C, respectively, in January. The annual average precipitations of Lianyungang and Yancheng are 896.7 and 1020.5 mm, respectively, with 50% of the precipitation occurring from June to September. The soil types of Qingkou saltern and Jinhai farm are coastal meadow saline soil, and the soil textures are heavy clay and sandy loam, respectively (Table 2). (2020) 10:9781 | https://doi.org/10.1038/s41598-020-66976-w www.nature.com/scientificreports www.nature.com/scientificreports/ Field and experimental sampling. At each study site, the 1200 m 2 area without vegetation for high salt content was divided into six parts. The experimental design was full factorial, three parts were randomly selected as K. virginica plantation, and three K. virginica seeds were planted in one hole according to 0.5 m × 0.5 m design on May 4, 2009. 10 d after germination, germinants were thinned from 3 to 1 in each hole and each part had 800 plants. The residual three parts were selected as control plot without K. virginica. After seedling thinning, K. virginica was allowed to grow unmanaged from 2009 onwards. And five sites in each part were randomly selected for soil sampling, which started on August 15, 2012 and every three months thereafter until May 25, 2013. Soil samples (n = 5) were collected from section at depths of 0-10, 10-20, and 20-30 cm. Prior to sample collection, the upper layer of soil (approximately 5 mm) was scraped off to remove litter. The collected soil samples were stored in sealed polyethylene bags placed in an insulated container and transported to a laboratory. The soil samples (50 g) for soil microbial biomass C analyses were dried at room temperature, passed through a 4 mm sieve, and stored in sealed plastic bags at 4 °C until analysis. The remaining samples were divided into two parts after air drying. One part (100 g) was milled to obtain particles with size < 2 mm for determination of physico-chemical properties, and the other part (350 g) was used for water stability structure analysis.
Analyses of total GRSP. Total GRSP was extracted from 1 g of soil by using the method described by Wright and Upadhyaya 21 . Extraction was conducted with 8 ml of 50 mM Na citrate (pH 8.0) at autoclave cycles of 121 °C for 60 min until the supernatant showed no red brown color typical of GRSP. Based on our previous experiment, two to three cycles are recommended. The fractions were determined through Bradford assay using bovine serum albumin as standard.
Analyses of AM fungus spore and colonization. Soil samples (25 g each) were used to determine spore density of AM fungus. Spore number was determined by wet sieving in a 40 µm mesh and decanting, followed by sucrose density centrifugation. The suspension was carefully decanted and added with 40% sucrose solution. AM fungus spores were counted under a stereoscopic microscope at 40 ×. Sporocarps were dissected with forceps, and the released spores were counted. Spore density was expressed as number of spores per 10 g of dry soil.
Fresh roots were cut into 0.5-1.0 cm segments and washed until free of soil. The roots were then stained with 0.5% (w/v) acid fuchsin solution according to the method described by Phillips & Hayman 83 and Zhao & He 84 . AM colonization was quantified by glass slide method, where 50 randomly selected 1 cm root segment units were microscopically examined 85 . Total colonization was expressed as the percentage of root segments colonized for a root sample.
Analyses of soil aggregates. Soil aggregates were fractionated using a wet-sieving procedure 86,87 . After capillary wetting of 100 g of air-dried soil to field capacity, the samples were immersed in water on a nest of 5, 3, 2, 1, and 0.25 mm sieves and shaken vertically at 3 cm height for 50 times during a 2 min period. This wet-sieving procedure resulted in fractions of >3 mm LM, 2-0.25 mm SM, and <0.25 mm MI. Soil aggregates retained on each sieve were backwashed into pre-weighed containers, oven dried at 50 °C for 2-3 days, and weighed.
Analyses of general soil properties. Soil pH was analyzed in a 1:5 soil-to-water ratio. SOC was determined by dichromate oxidization 88 . EC of soil was measured with a conductivity meter (Model DDS-11A; Leizi, Shanghai, China). SMBC was analyzed by fumigation-extraction method 89 . Available nitrogen was measured using alkaline hydrolysis diffusion method 90 . Olsen potassium in soil was determined by ammonium acetate through flame photometry. Olsen phosphorus was determined by chlorostannus-reduced molybdophosphoric blue color method after extraction with 0.5 M sodium bicarbonate for 30 min 90 . total nitrogen, total phosphorus, and total potassium concentrations of soils were determined using the method of Olsen et al. 90 .
Statistical analysis. Data were subjected to ANOVA using IBM SPSS Statistics (version 19.0; IBM Corp., Armonk, NY, USA). Differences were considered significant at p < 0.05. The means of main effects were compared using least significant difference test after a significant ANOVA test result. Pearson linear correlations among the parameters were evaluated with SPSS 19.0.