Enhancement of Soil Phosphorus Bioavailability by Arbuscular Mycorrhizae and Earthworms Through Regulating Soil Bacterial Community and Plant Nutrient Balance Under Salt Stress

Soil salinization is an important factor limiting plant phosphorus (P) uptake and crop production. This study aimed to investigate the effects of arbuscular mycorrhizal fungi (AMFs) and earthworms in enhancing soil P bioavailability by regulating soil salt ions and altering the soil bacterial community under salt stress. Methods Treatments with or without earthworms and with or without AMFs in a high-salinity soil were applied. The results showed that the maize biomass and plant P, Ca and Mg contents were signicantly increased by earthworms and AMF inoculation, and the highest plant P, Ca and Mg contents were observed with earthworm application alone. Earthworms and AMFs signicantly decreased the soil stable inorganic P (hydroxyapatite) proportion and increased the soil available dicalcium phosphate proportion. AMFs signicantly increased soil phosphatase activity and inorganic P fraction contents. Earthworms and AMFs signicantly increased soil bacterial Chao1 and phylogenetic diversity. Structural equation model analysis showed that the most important driver of soil P mineralization was soil bacterial diversity, followed by soil Ca 2+ and total salt concentration. Network analysis suggested that the response of bacteria to soil Ca 2+ but not salt concentration positively correlated with soil P availability. Earthworms and AMFs could stimulate certain bacteria harbouring the phoX alkaline phosphatase gene to increase soil phosphatase activity and soil P availability. 2016b). The mechanism through which earthworms and AMFs interact to affect maize P utilization in high-salinity soil needs to be further studied. Here, we explored the individual and combined roles of AMFs and earthworms in enhancing soil P availability and maize P uptake in high-salinity soil. Soil bacterial communities in the rhizosphere associated with earthworms and AMFs were assessed by Illumina MiSeq sequencing analysis. Correlations between maize P uptake, soil inorganic P fractions, soil salt ions, soil phosphatase activity and soil microbial diversity and composition were monitored. Our hypotheses were as follows: and

The results showed that the maize biomass and plant P, Ca and Mg contents were signi cantly increased by earthworms and AMF inoculation, and the highest plant P, Ca and Mg contents were observed with earthworm application alone. Earthworms and AMFs signi cantly decreased the soil stable inorganic P (hydroxyapatite) proportion and increased the soil available dicalcium phosphate proportion. AMFs signi cantly increased soil phosphatase activity and inorganic P fraction contents. Earthworms and AMFs signi cantly increased soil bacterial Chao1 and phylogenetic diversity. Structural equation model analysis showed that the most important driver of soil P mineralization was soil bacterial diversity, followed by soil Ca 2+ and total salt concentration. Network analysis suggested that the response of bacteria to soil Ca 2+ but not salt concentration positively correlated with soil P availability. Earthworms and AMFs could stimulate certain bacteria harbouring the phoX alkaline phosphatase gene to increase soil phosphatase activity and soil P availability.

Conclusions
In conclusion, earthworms and AMFs could enhance soil P bioavailability by stimulating soil P-cycling bacteria to activate soil stable inorganic P and by improving the plant cation nutrient balance under salt stress.

Introduction
Phosphorus (P) limitation ofcrop productivity is of particular concern in saline soils due to the high ionic adsorption and alkaline pH (Bano and Fatima, 2009). High ionic strength can enhance the adsorption of soil P by clay minerals and Fe/Al oxides, which leads to a decrease in P activity (Arai and Sparks, 2007). In saline soils, excessive salt ions strongly interfere with plant phosphorus, magnesium and calcium uptake and result in a nutritional imbalance in maize (Farooq et al., 2015). Moreover, high salinity could have adverse effects on the preservation and mineralization of soil organic P by in uencing organic matter inputs, destroying soil structure and decreasing soil microbial biomass and potential enzyme activities (Singh, 2016). A previous study demonstrated that for plants simultaneously subjected to soil salinity and soil P de ciency, the impact of the latter constraint on plant growth was predominant (Talbi Zribi et al., 2011). Therefore, understanding the links between soil P availability, soil ion alteration and plant nutrient uptake is important in saline soils.
Arbuscular mycorrhizal fungi (AMFs) symbiose with many crop plants and help their host plants acquire nutrients and resist salt stress (Porcel et al., 2012). AMFs could enhance soil organic P mineralization and increase soil P availability by releasing photosynthetic carbon to trigger soil phosphate-solubilizing bacterium growth (Zhang et al., 2016a). AMFs could release protons to mobilize insoluble P and extend the nutrient absorptive surface by the formation of extensive mycorrhizal hyphae (Smith and Smith, 2011). The existing information about mycorrhizal plants in saline soils focuses on the enhancement of plant nutrition acquisition and physiological tolerance to salt stress (Tigka and Ipsilantis, 2020;Wang et al., 2020). In contrast, scarce information is available on the possible role of AMFs in the soil microbial community in driving the P cycle or P transformation in saline soil. Hidriet al. (2019) reported that the bacterium Bacillus subtilis and AMFs could promote plant growth and contribute to increased plant nutrients and soil enzyme activities such as those of urease and alkaline phosphatase under salt stress. Therefore, how AMFs in uence the soil bacterial community and their potential roles in enhancing P availability in saline soil need to be further studied.
Earthworms can in uence soil P availability for plants and microorganisms by regulating soil P mobilization and sorption (Le Bayon and Binet 2006). The carboxyl groups in earthworm mucus can compete for P absorption sites, such as the cations Ca, Al and Fe, to increase soil P solubilization (Bianco and Defez 2010).
The behaviour of earthworms, including excavation, excretion and limivorous action, can increase soil microbial activity and produce high phosphatase activity in vermicomposts to improve P cycling (Gusain and Suthar, 2020). All of the above reports about the roles of earthworms in P availability were conducted in nonsaline soil. Just Oo et al. (2015) reported that vermicompost was effective in alleviating salinity, increasing soil total extractable phosphorus and improving crop growth. Our previous studies demonstrated that earthworms could increase soil P availability and improve the P absorption by maize byregulating the soil microbial community in saline soil (Zhang et al., 2016b;Zhang et al., 2018). However, the relations between the absorption of plant P and the cations affected by earthworms under salt stress are still not clear.
In non-P active soil, the availability of soil phytate-P and insoluble inorganic P (calcium phosphates) was effectively improved by earthworms and AMF interactions (Cao et al., 2015a;Cao et al., 2015b). Our previous studies showed that earthworms and AMFs cooperatively increased soil P availability in high-salinity soil (EC 8.56 dSm − 1 , Zhang et al., 2018) and promoted maize P absorption in low-salinity soil (EC 4.41 dSm − 1 , Zhang et al., 2016b). The mechanism through which earthworms and AMFs interact to affect maize P utilization in high-salinity soil needs to be further studied. Here, we explored the individual and combined roles of AMFs and earthworms in enhancing soil P availability and maize P uptake in high-salinity soil. Soil bacterial communities in the rhizosphere associated with earthworms and AMFs were assessed by Illumina MiSeq sequencing analysis. Correlations between maize P uptake, soil inorganic P fractions, soil salt ions, soil phosphatase activity and soil microbial diversity and composition were monitored. Our hypotheses were as follows: i) AMFs and earthworms relieve the negative effects of salt stress on maize P uptake by improving the plant nutrient balance and increasing soil P bioavailability; ii) AMFs increase soil P availability by enhancing soil phosphatase activity and regulating soil P-cycling bacteria to activate soil stable inorganic P; and iii) earthworms regulate maize cation utilization to enhance maize P utilization (Fig. 1). Four treatments were set up as follows: CK, high-salinity soil with maize planting; E, high-salinity soil with maize planting and earthworm addition; AM, high-salinity soil with maize planting and AMF inoculation; and E + AM, high-salinity soil with maize planting and earthworm and AMF addition. There were four replicates for each treatment. One germinated maize seed was sown in each pot. The mycorrhizal inoculum (Funneliformis mosseae, Beijing Academy of Agricultural and Forestry Sciences) was applied at 2.5 g kg − 1 soil. Equal amounts of sterilized AMF inoculum were added to treatments without AMFs (CK and E). Ten similar adult earthworms (Eisenia fetida, 3.09 ± 0.08g fresh weight) were inoculated at the third-leaf stage of the maize. Eisenia fetida was chosen because it could be easily obtained and it was the only species that largely cultivated and applied to modify soil quality in China (Guo et al., 2019). Earthworms were incubated and adapted in the experimental soil for 14 days before addition. The surface and guts of earthworms were pretreated to eliminate the impact of natural AMF propagules. All pots were maintained at 65% of the soil waterholding capacity by weighing and supplementing deionized water.

Soil and plant sample harvesting
The soil and plant samples were harvested 60 days after sowing. The shoots and roots were separately collected. Aliquotsof roots were frozen at -20°C for mycorrhizal colonization. The other plant samples were oven-dried for elemental analysis. To prevent the potential in uence of exotic bacteria, the top 2 cm of surface soil was removed before collecting soil samples. The well-sieved and mixed soil was frozen for measuring phosphatase activity (-20°C) and soil DNA extraction (-70°C) or air-dried for detecting soil organic P fractions and salt ions. Earthworms were recovered, washed, and weighed.

Assays for root mycorrhizal colonization and plant nutrients
To determine the intensity of mycorrhizal colonization (M%) ofmaize roots, the cut root segments (1 cm) were stained with Trypan blue (McGonigle et al., 1990), and the MYCOCALC program was used to calculate M%. The dried and ground plant samples were digested with a mixture of HCl/HNO (3/1 v/v) and examined by inductively coupled plasma-optical emission spectroscopy for plant P, Ca and Mg contents.

Assays for soil salt ions and soil inorganic P fractions
Soil soluble cations and anions were extracted with deionized water (soil/water 1/5 w/v) (Bao, 2000). Soil Na + was determined by ame photometry, whereas soil Ca 2+ and soil Mg 2+ were analysed by EDTA titration following a standard procedure. The concentration of soil Cl − was determined by titration with 0.025 mol L − 1 silver nitrate solutions using potassium chromate indicator (Bao, 2000). Soil HCO 3 − was determined by neutralization titration with double indicators, namely, phenolphthalein and bromophenol blue (Bao 2000). To study the P characteristics and transformation, soil inorganic P fractions, including calcium phosphate (Ca-P), nonoccluded aluminium phosphate (Al-P) and iron phosphate (Fe-P), and occluded phosphate (Oc-P), were analysed by sequential extractions (Jiang and Gu, 1989). Calcium phosphates, including dicalcium phosphate (Ca 2 -P), octocalcium phosphate (Ca 8 -P) and hydroxyapatite (Ca 10 -P), were determined (Jiang and Gu, 1989). Brie y, 1 g of dry soil was rst extracted by 0.25 mol L − 1 NaHCO 3 for Ca 2 -P determination; second, the NaHCO 3 -extracted soil was dissolved in 0.5 mol L − 1 NH 4 OAc to measure Ca 8 -P; third, the NH 4 OAcextracted soil was dissolved in 0.5 mol L − 1 NH 4 F to determine nonoccluded Al-P; fourth, 0.5 mol L − 1 NaOH and 0.1 mol L − 1 Na 2 CO 3 were added to the NH 4 F-extracted soil to determine nonoccluded Fe-P; fth, the extracted soil from the fourth step was added to 0.5 mol L − 1 NaOH and digested with 10 ml of a H 2 SO 4 /HClO 4 /HNO 3 mixture (1/2/7 v/v/v) to determine Oc-P; and sixth, the extracted soil from the fth step was extracted with 0.5 mol L − 1 (1/2H 2 SO 4 ) to determine Ca 10 -P.All extracts were analysed for P fractions by the molybdate blue method (Murphy and Riley, 1962). The mobility and mineralization magnitude of inorganic P fractions was in accordance with their extraction order, i.e., Ca 2 -P > Ca 8 -P > nonoccluded Al-P > nonoccluded Fe-P > occluded P > Ca 10 -P. Soil alkaline (pH 10) phosphatase was determined at 420 nm (Tabatabai, 1982

Individual soil P fractions and phosphatase activity
Analysis of the inorganic P fraction distribution showed that the largest proportions of soil inorganic P among all treatments were Ca-P, which accounted for 48-55%, and Al-P, which accounted for 18-21% ( Fig. 3). Mycorrhizal inoculation (treatments AM and E + AM) signi cantly decreased the proportions of Ca 10 -P (19-21%) and Al-P (12-15%) while increasing the proportion of Oc-P (60-90%) (p < 0.05). The proportion of Ca 2 -P was signi cantly higher in the E + AM treatment than in the CK treatment (p < 0.05), indicating an increase in soil active P. Compared to those in the controls, the Ca 2 -P, nonoccluded Al-P and nonoccluded Fe-P contents were signi cantly increased in all treatments (Table S1; p < 0.05). Treatments AM and E + AM signi cantly increased the soil Oc-P and Ca 8 -P contents and soil alkaline phosphatase activity (Table S1 and

Mycorrhizal colonization and soil bacterial community
The intensity of mycorrhizal colonization of the roots was higher in treatments AM (13%-29%) and E + AM (25%-42%) than in the controls (0.2%-0.5%), indicating successful AMF colonization of maize roots. All initial earthworms were recovered, no signi cant change was found in their biomass at harvest. There was also no signi cant difference in their biomass between treatments E and E + AM. Metagenomic pro ling analysis of Illumina-based 16S rRNA showed that treatments E, AM and E + AM increased the OTU numbers by 27%, 43% and 38%, respectively, compared with those in the controls (1128 OTUs). Good sequencing coverage was obtained. The rarefaction curves were shown in Fig. S1. The Chao1 and phylogenetic diversity indices were signi cantly increased in all treatments, indicating increased soil bacterial diversity (Table 1). Variations in the relative abundance of the dominant bacterial species (phyla) among all treatments are shown in Fig. 4.
The phyla Proteobacteria, Bacteroidetes and Actinobacteria were the dominant species associated with earthworms and AMF inoculation. The most abundant bacterial groups with signi cant differences at the phylum, class, order and family levels are summarized inTable S2. Treatments E, AM and E + AM signi cantly increased the relative abundance of the phylumChloro exi (p < 0.05). The AM treatment signi cantly increased phylum Acidobacteria abundance, and the E + AM treatment signi cantly decreased Bacteroidetes abundance (p < 0.05). At the class level, the E + AM treatment signi cantly increased the relative abundance of Anaerolineae and Acidimicrobiia (p < 0.05). Additionally, the AM treatment signi cantly increased the relative abundance of Acidobacteria and an unidenti ed species (p < 0.05). At the order level, earthworms and AMFs signi cantly decreased the relative abundance of Flavobacteriales, Alteromonadales and Cytophagales (p < 0.05). At the family level, earthworms and AMFs signi cantly increased the relative abundance of an unidenti ed species and Bacillaceae (p < 0.05). Earthworm presence signi cantly increased Oceanospirillaceae abundance (p < 0.05). As shown in Fig. 5 3.5. Network analysis of soil P availability, salt ions and their associated bacteria As shown in Fig. 6, the soil P availability and salt ions quanti ed in this study were signi cantly correlated with various bacterial species (p < 0.05). The network was divided into four groups: the decreases in soil total salt content and soil Ca 2+ induced by the addition of earthworms and AMFs resulted in more increasing bacterial clusters (43) than decreasing bacterial clusters (10). The increases in soil P availability and soil HCO 3 − concentration in all treatments resulted in more increasing bacterial clusters (49) than decreasing bacterial clusters (5). The bacteria associated with decreasing soil salt content (41) and Ca 2+ (39) had the highest diversity, which generally increased with the addition of earthworms and AMFs. For instance, earthworms and AMFs elevated the abundance of the genera Chloro exi, Acidimicrobiales and Anaerolineaceae (Table S3). The signi cant correlation relationships between soil chemical properties and their associated bacteria in the network indicated that the bacterial taxa were developed and shaped by a combination of factors.

Factors driving soil P availability
To investigate the predominant factor increasing soil P availability, a hypothesized SEM model (Fig. 9a) was established, and its tting parameters were excellent (χ 2 , 5.476; p, 0.602; CFI, 1.000; RMSEA, 0.000). Our SEM explained 76% of the variance in soil P availability (Fig. 9a). When considering both direct and indirect effects, soil bacterial diversity was the most important driver controlling soil P availability, followed by soil Ca 2+ concentration (λ = -0.589) and salt concentration (λ = -0.529) (Fig. 9b). Soil P availability was directly affected by soil bacterial diversity and phosphatase activity and indirectly affected by soil Ca 2+ , salt concentration and soil HCO 3 − (Fig. 9a). Soil P availability was positively driven by soil bacterial diversity, phosphatase activity and soil HCO 3 − but negatively affected by soil Ca 2+ and salt concentration (Fig. 9b).

Discussion
Maize P uptake regulated by earthworms and mycorrhizae in saline soil It is well known that high salinity decreases plant P uptake and adversely affects crop growth ( (Akhter et al., 2003). Earthworms and AMFs increased the maize P content, which was signi cantly negatively correlated with the soil salt concentration (r = -0.547, p < 0.05, Fig. 7). These results indicated that the decrease in soil salt content induced by earthworms and AMFs could contribute to the P uptake by maize.
Magnesium plays a major role in plant phosphorus absorption and transport and enhances P use e ciency (Grzebis et al., 2018). Calcium could stabilize cell wall structure, regulate ion (K + , Na + and Mg 2+ ) transport and selectivity and improve plant osmoregulation (Elkelish et al., 2019). Earthworms, mycorrhizae and their combination signi cantly increased the shoot Mg and Ca contents, which were signi cantly positively correlated with maize P content (Fig. 8, p < 0.01). Thus, we speculated that earthworms and mycorrhizae probably enhance maize P uptake by increasing Mg and Ca utilization. Tuffen et al. (2002) demonstrated that earthworm activity could enhance 32 P transfer by AMF mycelium to leek (Allium porrum L.) in non-saline soil. In this study, the E treatment resulted in the greatest magnitude of increase in maize P, Mg and Ca contents (Fig. 2), indicating that the addition of earthworms did not enhance the capacity of AMFs to transfer these soil nutrients to their host maize. However, earthworms addition increased the hyphal length density by 17.3% compared to that under AMF inoculation alone (Zhang et al., 2018). Therefore, we considered and increased the proportion of soil dicalcium phosphate (Ca 2 -P) (Fig. 3).Simultaneously, the soil Ca 10 -P proportion was signi cantly negatively correlated with the soil Ca 2 -P proportion (Fig. 8, p < 0.05). These results indicate a transformation of the soil stable P form (Ca 10 -P) into active or available P (Ca 2 -P) driven by earthworm and AMF combination. Additionally, AMF inoculation signi cantly enhanced soil phosphatase activity and increased the soil inorganic P fractions (Ca 2 -P, nonoccluded Al-P, Ca 8 -P, nonoccluded Fe-P and Oc-P) ( Table 1 and Table S1). Soil phosphatase activity was negatively correlated with the soil Ca 10 -P proportion (Fig. 8, p < 0.01) and positively related to the increased soil inorganic P fractions (r > 0.640, p < 0.01). Therefore, we suggest that mycorrhizae could strengthen soil phosphatase activity to promote stable P (Ca 10 -P) dissolution and mineralization. Thus, AMF appears to have a dominant role in activating hard dissolved P by increasing soil phosphatase activity under salt stress.

Enhancement of soil phosphatase activity by earthworms and mycorrhizae via altering soil bacteria
To overcome the limitation of bioavailable inorganic orthophosphate, plants and bacteria secrete numerous phosphatases to cleave orthophosphate from complex organic P substrates (Lidbury et al., 2017). In this study, the separate and combined addition of earthworms and AMFs signi cantly increased the Chao1 and phylogenetic diversity indices (Table 1), which were signi cantly positively correlated with soil alkaline phosphatase (r > 0.656, p < 0.01). These results suggested that earthworms and AMFs probably increased soil bacterial diversity to enhance soil phosphatase activity. To understand the drivers of soil P availability, an SEM was constructed to jointly investigate the multiple disparate pathways associated with P activation. The SEM provided evidence that soil bacterial diversity was the most important driver of soil P availability, followed by soil Ca 2+ concentration and salt concentration (Fig. 9b). This indicated that low bacterial diversity was the key factor limiting soil P activation, while the increase in soil bacterial diversity induced by earthworms and AMFs resulted in increased soil P availability (Zhang et al., 2018). Network analysis could provide new insights into the relationships between soil nutrients and their associated bacteria under complex conditions (Zheng et al., 2018). Our network analysis showed that the change in bacterial diversity was largely dependent on the soil salt concentration and Ca 2+ concentration (Fig. 6). Simultaneously, the response of bacteria toCa 2+ concentration but not salt concentration had a signi cant positive correlation with soil P availability.
Identifying the physiological attributes of the dominant bacterial taxa is critical for understanding the microbial controls on some key soil processes, e.g., soil carbon and nutrient cycling (Manuel et al., 2018).
Previous studies reported that abundant phyla, including Acidobacteria and Chloro exi, harboured the phoX alkaline phosphatase gene and participated in organic phosphorus cycling in multiple soil environments (Ragot et al., 2017;Shao et al., 2020). Earthworms and AMFs signi cantly increased the abundance of the phylum Chloro exi, and AMFs alone increased the abundance of the phylum Acidobacteria. The abundances of Chloro exi and Acidobacteria were positively correlated with the soil phosphatase and soil Olsen-P contents (Fig. 7, p < 0.05). Thus, we speculated that earthworms and AMFs may stimulate the growth of Chloro exi and Acidobacteria to increase soil phosphatase activity and thus enhance soil P availability. LEfSe analysis reveals some predominant phylotypes in smaller taxonomic categories. Xanthomonadales and Oceanospirillaceaespecies were signi cantly higher in the E treatment than in other treatments (Fig. 5), indicating that earthworms could stimulate these bacteria to participate in P degradation and accumulation (Martin et al., 2006;Sosa et al., 2017). Earthworms and AMF inoculation increased the abundance of Chloro exi and Anaerolineales (Fig. 5), which are dominant in carbon-rich soil and associated with cellulose hydrolysis (Lian et al., 2017;Pinnell et al., 2014). Simultaneously, the earthworm and AMF combination decreased the abundance of Cytophagale, which is an e cient degrader of complex organic compounds in relatively low-carbon soil (Zhang et al., 2013). The shift inbacterial composition from the low-carbon type (Cytophagales) to the rich-carbon type (Chloro exi and Anaerolineales) induced by earthworm and mycorrhizal activities suggested an improvement in saline soil nutrient conditions.

Conclusion
The present study indicated that earthworms and AMFs could interact to relieve the negative effects of salt stress on maize P uptake and growth. Earthworms predominantly enhanced maize P uptake by increasing plant Ca and Mg utilization, and AMFs played a dominant role in activating hard dissolvable P (Ca 10 -P) by increasing soil alkaline phosphatase activity. Soil bacterial diversity was the most important driver controlling saline soil P availability under the combined regulation of earthworms and AMFs.  Figure 1 Schematic representation of the cooperative effect of earthworms and arbuscular mycorrhizael fungi (AMF) on P cycling processes in a high salinity soil. We hypothesized that the cooperation of earthworms and AMF could increase soil P availability and promote maize P absorption and growth (a); AMF increased the soil P availabity by increasing soil phosphatase activity to activate stable inorganic phosphate (hydroxyapatite, Ca10-P) transforming into plant-available P form (Dicalcium phosphate, Ca2-P) (b); Earthworms improved mazie cations (Ca and Mg) utilization to enhance the maize P uptake (c); Earthworms and AMF could increase soil bacterial diversity and P-cycling bacteria to increase soil phosphatase activity and soil P availability (d).

Figure 2
Maize biomass and nutrients uptake in the salinity soil inoculated with earthworms and/or mycorrhizae.
Data are the means of four replicates ± SE and were compared using Tukey's tests. The values with the same lower case letter are not signi cantly different. CK is the soil without mycorrhizae and earthworms; E is the soil with earthworms inoculation; AM is soil with mycorrhizae addition; E+AM is the soil inoculated mycorrhizae and earthworms.

Figure 3
Percent distribution of inorganic P fractions among different treatments in a salinity soil. Dicalcium phosphate (Ca2-P), octocalcium phosphate (Ca8-P), non-occluded aluminum phosphate (Al-P), non-occluded iron phosphate (Fe-P), occluded phosphate (Oc-P) and hydroxyapatite (Ca10-P) were analyzed and listed. CK is the soil without mycorrhizae and earthworms; E is the soil with earthworms inoculation; AM is soil with mycorrhizae addition; E+AM is the soil inoculated mycorrhizae and earthworms. Variation of the dominant bacterial species at phylum level. The right half circle shows the species composition in the sample. The colour of the outer ribbon indicates the phlum group from which it comes.
The width of the inner ribbon represents the relative abundance of the species in the corresponding sample. The colour of the left half circle represents the sample from which it comes. CK is the salinity soil without mycorrhizae and earthworms; E is the salinity soil with earthworms inoculation; AM is salinity soil with mycorrhizae addition; AE is the salinity soil inoculated mycorrhizae and earthworms. Supervised comparison identi es differential abundance of bacteria using LEfSe (LDA, linear discriminant analysis >3.5). CK is the salinity soil without mycorrhizae and earthworms; E is the salinity soil with earthworms inoculation; AM is salinity soil with mycorrhizae addition; E+AM is the salinity soil inoculated mycorrhizae and earthworms.   The proportion of variation explained (R2) appears above response variables. Data of soil salt and available P were referenced to Zhang et al., 2018 Supplementary Files This is a list of supplementary les associated with this preprint. Click to download. SupplementaryInformation.docx