Our study sites were located in the Yellow River Delta National Nature Reserve (118°33′–119°20′E, 37°35′–38°12′N), near Bohai and Laizhou Bay, China (Figure 1). Mean annual temperature and precipitation are approximately 12.4°C and 550 mm, respectively. Soil types are classified as fluvial soil formed by alluvial sediment of the Yellow River and saline soil formed by long-term seawater intrusion. In the water-salt interaction regions, we selected seven salinity concentrations (represented by electrical conductivity), including S1 (1.61 dS m^–1), S2 (2.47 dS m^–1), S3 (3.67 dS m^–1), S4 (5.21 dS m^–1), S5 (9.24 dS m^–1), S6 (14.53 dS m^–1), and S7 (29.90 dS m^–1) (Table 1 and Figure 1). The sites of S1 and S2 are the interaction regions of the Yellow River and Bohai bay; S3 and S4 are affected by the seasonal tide; and S5, S6, and S7 are the degraded wetlands due to dam construction. Plant communities are dominated by Phragmites australis, following Suaeda glauca, Tamarix chinensis, Cynanchum, etc. We established study plots in August 2019 using a randomized within-wetland nested design. In each of the seven salinity concentrations, four plots (each approximately 10 m × 10 m) were randomly selected to collect soil samples (0–20 cm), with at least a 100-m distance between each plot. To alleviate spatial heterogeneity, we randomly took five soil cores using a soil auger (diameter of 5 cm) within each plot and pooled them into one soil sample (n = 4 samples per salinity concentration). After sieving soils to 2 mm to remove fine root, plant, and animal debris, soils were homogenized, and analyzed as below.

Air-dried soils (sieved to 2 mm) were used to prepare soil slurries (1:5 w/v), oscillate for 30 min, and subsequently measure electrical conductivity (EC) and pH on an electrode meter (Jenco 6173, Jenco, United States). Fresh soil samples (sieved to 2 mm) were oven-dried (105°C for 48 h) to determine soil water content (SWC). Air-dried soil samples (sieved to 0.15 mm) were prepared to analyze SOC using the potassium dichromate oxidation method (Nelson and Sommers, 1982). Total nitrogen (TN) was measured on a Vario MACRO Cube (Elementar vario EL III, Germany). Total phosphorus (TP) was detected using a UV-VIS spectrophotometer (PERSEE, WA, United States) after soil digestion using H₂SO₄ and HClO₄ (Cheng et al., 2016).

Soil ASs, including GluN, GalN, and MurA, were determined according to a modified procedure referring to Zhang and Amelung (1996). In brief, air-dried soil samples (sieved to 0.15 mm) containing 0.3 mg N at least were hydrolyzed using 15 ml 6 M HCl at 105°C for 8 h. Each hydrolysate was filtered and dried at 52°C on a rotary vacuum evaporator after adding 100 μl of myoinositol (internal standard 1 with response factors, Rf of 1). The dried residue was dissolved using deionized water and purified with KOH neutralization. The solution was fully dried and dissolved with absolute methanol. The dissolved residue was dried by N₂ gas at 45°C, subsequently redissolved with 1 ml deionized water and 100 μl N-methylglucamine (internal standard 2), and then lyophilized.

Each lyophilized residue was derivatized with a 300-μl derivatization reagent. The reactive solution was completely mixed and heated at 75–80°C for 35 min. The derivative was heated (75–80°C, 25 min) for acetylation after cooling and adding 1 ml acetic anhydride. After cooling to room temperature, we orderly added 1.5 ml dichloromethane and 1 ml 1 M HCl and fully mixed to isolate the organic phase. The organic phase was washed with deionized water three times to absolutely remove residual anhydride. The remaining organic phase was dried with N₂ gas at 45°C and then dissolved with 100 μl ethyl acetate-hexane (1:1 v/v). Finally, the extracted ASs were detected and analyzed on an Agilent 7890B GC (Agilent Technologies, Santa Clara, CA, United States). In terms of the response factors of standards (i.e., GluN, GalN, MurA, and mannosamine) relative to the internal standard 1 (Rf = 1), we can calculate the concentration of soil ASs; the internal standard 2 is used to check the derivatization process and calculate the recovery of soil ASs (Zhang and Amelung, 1996; Liang et al., 2012).

One-way ANOVA was conducted to assess the effects of salinity on soil physicochemical properties and AS parameters using R software (R version 3.6.0). We performed redundancy analysis to detect the relationships of GluN/MurA, GluN/SOC, GalN/SOC, and MurA/SOC (log-transformed) with soil physicochemical properties on Canoco (version 4.5 for Windows; Ithaca, NY, United States). Path analysis was used to evaluate the causality among soil physicochemical properties and ASs along a salinity gradient, using AMOS software (AMOS 17.0.2 student version; Amos Development, Crawfordville, FL, United States). The maximum likelihood method was used to fit measured data to the model. The adequate model goodness of fit was tested using root-mean-square error of approximation (RMSEA).

Salinity resulted in the significant changes in soil pH (F = 2.84, P = 0.04), SOC (F = 13.89, P < 0.001), and TN (F = 11.24, P < 0.001), whereas there was no change in SWC, SOC/TN, and TP (Table 1). High salinity soils (S5, S6, and S7) had lower pH than low salinity soils (S1), decreasing by 5.82%. SOC and TN gradually decreased along with increased salinity, from 10.13 to 2.93 g kg^–1 and from 1.07 to 0.41 g kg^–1, respectively. Although salinity did not change SWC, high salinity coastal wetlands had low SWC (Table 1).

Salinity greatly influenced the contents of soil AS (F = 29.37, P < 0.001), GluN (F = 25.84, P < 0.001), GalN (F = 35.65, P < 0.001), and MurA (F = 10.77, P < 0.001) (Figure 2). The contents of AS, GluN, and GalN gradually declined across a salinity gradient, which decreased by 90.56% (Figure 2A), 91.11% (Figure 2B), and 90.62% (Figure 2C), respectively. The contents of MurA were higher in low salinity soils (S1 and S2) than that of other soil salinity concentrations. Additionally, MurA showed no change from S3 to S7 (Figure 2D).

The ratios of AS, GluN, and GalN to SOC showed significant responses to salinity (P < 0.001), whereas there was no change in the ratio of MurA to SOC (Figure 3). The ratios of AS, GluN, and GalN to SOC showed the similar trend with AS content across a salinity gradient. The AS/SOC decreased from 36.62 to 12.32 g kg^–1 (Figure 3A; F = 39.20, P < 0.001); GluN/SOC showed a decline from 24.22 to 7.76 g kg^–1 (Figure 3B; F = 43.27, P < 0.001); and GalN/SOC reduced from 11.50 to 3.75 g kg^–1 (Figure 3C; F = 24.00, P < 0.001) from S1 to S7.

Salinity resulted in a significant change in GluN/MurA (Figure 4B; F = 31.32, P < 0.001), but there was no change in GluN/MurA (Figure 4A). Low salinity soils had high GluN/MurA, and high salinity lowered GluN/MurA in soils (Figure 4B).

Redundancy analysis demonstrated that soil physicochemical properties explained 82.19% (axis 1 of 80.1% and axis 2 of 2.1%) variation of the GluN/MurA, GluN/SOC, GalN/SOC, and MurA/SOC (Figure 5). The ratios of GluN/SOC, GalN/SOC, and GluN/MurA were negatively related to soil salinity (represented by EC) and positively associated with SWC, pH, SOC, and TN, whereas not related to SOC/TN and TP. Soil physicochemical properties showed no impact on the MurA/SOC (Figure 5).

Path analysis further evaluated the causal relationships among soil physicochemical properties, GluN/MurA, and AS/SOC across a salinity gradient in coastal wetlands and elucidated the relative contributions of these soil variables to explain changed AS/SOC (Figure 6). Salinity, pH, SOC, TN, and GluN/MurA were the main abiotic and biotic factors influencing AS/SOC, accounting for 84% variation of the AS/SOC. Soil salinity showed a negative effect on soil chemical properties and AS/SOC; soil pH and TN indirectly and positively influenced AS/SOC via changing the GluN/MurA; and SOC showed a positive direct impact on AS/SOC (Figure 6).

In this study, we found that the AS content was 34.82–368.6 mg kg^–1 along a salinity gradient under coastal wetlands in the Yellow River delta, which was several or even tens of times lower than AS content in terrestrial ecosystems (i.e., forest, grassland, and agriculture) (Lauer et al., 2011; Xia et al., 2019; Chen et al., 2020). Coastal wetlands are predominately controlled by water and salinity due to tides and underground seawater (Ma et al., 2017; Steinmuller et al., 2019). The restrained plant productivity leads to low SOM (Zhao et al., 2020), decreasing the accumulation of microbial AS. Also, plant phenology shows a quick response to changed soil salinity (Sun et al., 2021). High salinity negatively influences plant phenology and alters the seasonality of plant photosynthesis or productivity. Especially in dry seasons, the exacerbated soil salinity stress restrains aboveground plant growth, root germination, and reproduction (related to plant reproduction the next year) (Baldwin et al., 2014; Sun et al., 2021), reducing C input to soils, consequently resulting in the decrease in microbial AS. In addition, the SOM was not easily utilized by soil microorganisms with a low decomposition rate (Steinmuller and Chambers, 2019; Ward, 2020), which is likely to result in low AS synthesis.

As the main component of the microbial cell wall, soil AS has certain specificity and stability, which cannot only represent microbial residues but also reflect the dynamic changes of ecosystems (Lauer et al., 2011; Shao et al., 2019). Increased soil salinity decreased AS content, indicating that salinity stress would suppress microbial residue accumulation in coastal wetland soils. Microbial residues are mainly originated from the fast iteration of microbial cell production and death, thus high microbial biomass could promote microbial residue retention in soils (Liang et al., 2017). A recent study has reported that increased salinity results in a decline in microbial biomass, associated with the decreased microbial residues in the saline-alkali soils (Chen et al., 2021). In addition, microorganisms utilize more SOM to synthesize several stress-resistant substances (e.g., extracellular polysaccharides) to maintain microbial survival and integrity and less SOM investment in microbial cell growth and metabolism (Malik et al., 2020), thus resulting in low microbial residues in high salinity soils.

The GluN and MurA can be used to represent fungal and bacterial residues, respectively (Joergensen, 2018). The GluN accounted for 65.67% of total AS; MurA accounted for 3.19% of total AS; and the content of GluN was approximately 20 times of MurA in coastal wetland soils. These findings are similar to the previous studies that the content of soil GluN was approximately 15–20 times of MurA in terrestrial ecosystems (Shao et al., 2019; Wang et al., 2021). Although increased salinity stress greatly decreased the contents of GluN and MurA, the proportion of GluN and MurA in total AS showed an increment and decline, respectively. The inconsistent changes in fungal and bacterial residues are probably ascribed to the different responses of fungal and bacterial communities to salinity. The 80% of fungi in soil systems are symbiotic with plants; and changed plant productivity and traits would influence fungal activity and biomass (Zheng et al., 2021). In our studied regions, increased soil salinity inhibits the growth and photosynthate synthesis of Phragmites australis; and few Phragmites australis survive in high or extreme salinity soils (Guan et al., 2017). The decreased plant litterfall and root exudates lead to less available C input to soil fungi, particularly plant symbiotic fungi, to assimilate less C into fungal biomass, thus reducing fungal residue retention in soils. MurA is uniquely derived from bacteria, which can reflect the change in the bacterial community (Glaser et al., 2004). Low salinity soils had higher bacterial residues, and there was no change in bacterial residues from medium salinity to high salinity. The dynamic of bacterial residues along with soil salinity gradient may be attributed to the salinity tolerance of the bacterial community. Some specific bacterial taxa (e.g., Halomonas and Chloroflexi) can survive in high salinity soil (Xu et al., 2021), which maintains the balance between production and degradation of bacterial residues, resulting in no change in bacterial residues.

With the developments of microbial biomarkers (i.e., AS) and microbial C pump theory (Zhang and Amelung, 1996; Liang et al., 2017), more studies concentrate on microbial residue contribution to SOC pool (Kallenbach et al., 2016; Ma et al., 2018; Deng and Liang, 2022). The AS extracted from soils accounts for a major percentage of dead microbial biomass (Joergensen, 2018). Therefore, AS levels are used to evaluate the extent to which microbial residue contributes to the SOC pool, providing significant information on the “chronic” responses of microbial residues to changing ecosystems (Shao et al., 2019). The AS/SOC can represent microbial residue contributions to SOC. The AS/SOC was gradually declined with increasing salinity, approximately decreasing by 70% from low salinity to high salinity, indicating that high salinity stress greatly inhibits microbial residue contributions to the SOC pool. We found that salinity negatively affected fungal residue contributions to SOC, and there was no change in bacterial residue contributions to SOC. These findings imply that fungal residues predominately drive SOC storage, which is consistent with previous views (Wang et al., 2021). The different effects of salinity on fungal and bacterial residue contributions to SOC are likely ascribed to change soil physicochemical properties. Our study shows the positive relationship of GluN/SOC with SWC, pH, SOC, and TN, suggesting that changed soil water and nutrient conditions regulate fungal residue accumulation in the SOC pool. Additionally, the chemical stability of bacterial residues is lower than fungal residues, which are easily decomposed and assimilated by soil microorganisms (Joergensen, 2018; Ma et al., 2022). The fast turnover of bacterial residues is not conducive to the retention of microbial residue-C belowground in coastal wetlands.

Current studies pay more attention to the plant community effect on SOC cycling, and plant-derived C mainly drives wetland soil C storage (Duarte et al., 2013; Xia et al., 2021). Microbial C pump theory has proposed that microorganisms utilize the decomposed plant detritus and micromolecular SOM to synthesize microbial cells, contributing to the SOC pool via microbial necromass (Kästner and Miltner, 2018). Low salinity held high AS/SOC, which indicates more soil microbial residue-C accumulation in low salinity coastal wetlands. Abundant soil water, SOC, and N promote the synthesis of microbial products, enhancing the contribution of microbial residues to existing SOC stocks. High soil water and nutrient stimulate plant growth and photosynthetic C input into soils (Chu et al., 2019); and microbes assimilate more plant-derived C to synthesize microbial biomass, ultimately increasing microbial residue-C storage in coastal wetland soils. In addition, we also found that AS/SOC was positively related to pH. Some studies have clarified that pH is tightly associated with soil cation, such as calcium and aluminum ions (Gruba and Mulder, 2015). Microbial residues can persist in soils for decades or even centuries in the form of stable SOM through mineral association (Sokol et al., 2019; Buckeridge et al., 2020), substantially enhancing soil C storage. Overall, the abundant soil resource (i.e., water, C substrate, and N nutrient) can promote SOC storage by increasing microbial residue accumulation and stability in low salinity coastal wetlands.

In high salinity coastal wetlands, salinity stress greatly restrained the contribution of microbial residues to SOC. This might be explained by the two mechanisms, namely, microbial stress tolerance strategy and low available C. In extreme environmental stress (e.g., high salinity or drought), microorganisms select a stress tolerance strategy responding to environmental changes (Krause et al., 2014; Malik et al., 2020). Microorganisms invest more C to synthesize osmolytes (e.g., mycose) or extracellular polymeric substances (e.g., polysaccharides) to protect microbial cell integrity and to enhance microbial stress-resistance in high salinity soils. Meanwhile, less C was used for microbial cell growth and proliferation, thus reducing the source of microbial residues. The low microbial cell death and microbial residue production both lower the retention of microbial residues in the SOC pool of coastal wetlands. In addition, high salinity soils held few SOC and low SOC/TN, which decrease microbial C availability. Low soil available C inhibits the production of microbial biomass and residues, resulting in the low contribution of microbial residues to SOC storage in high salinity coastal wetlands.

Path analysis was used to evaluate the causal relationships among microbial residue-C contribution, soil physicochemical properties, and fungal/bacterial residues, clarifying the main factors driving microbial residue contribution to SOC. Changed soil salinity, pH, SOC, TN, and GluN/MurA explained 84% variation of microbial residue-C contribution to SOC. Soil nutrient regulates microbial residue contribution to SOC through varied fungal and bacterial residues associated with their stability along a salinity gradient in coastal wetlands.

Although soil physicochemical properties (e.g., salinity, water, SOC, and N) were substantially different among coastal wetlands with different salinity concentrations, the changes in microbial residues and their contributions to SOC were influenced. However, we cannot neglect the impacts of system experience conditions (e.g., hydrology, microclimate, and topography) on microbial residues in different geographical locations. Together, our study presents supporting evidence on how salinity affects the fate and magnitude of microbial residues in coastal wetlands through changed soil properties. Further research is needed to accurately assess microbial residues in changing wetlands considering system experience conditions and other fluctuant environmental variants.

The ratio of GluN to MurA cannot only indicate the responses of fungal and bacterial communities to environmental changes but also represent the relative contributions of fungal and bacterial residues to the SOC pool (Liang et al., 2015; Joergensen, 2018). We found a gradual reduction in GluN/MurA across increasing salinity, implying that fungi are more competitive in low salinity soils, and bacteria are more tolerant in high salinity soils of coastal wetlands. The positive relationships of GluN/MurA to SWC, SOC, and TN indicate that suitable soil water, salinity, and nutrients favor fungi and stimulate fungal residue accumulation, reserving more microbial residue-C in the SOC pool.