The data for the meta-analysis were retrieved from Web of Science (WOS) and the China National Knowledge Infrastructure (CNKI), and the keywords searched included “potassium-solubilizing bacteria”, “yield”, “rhizosphere potassium-solubilizing bacteria”, “microorganisms”, “enzyme activity” and “potassium solubilization”. For the meta-analysis, 102 relevant published studies were collated, including 95 English and 7 Chinese studies ( Figure 1 ), containing 846 pairs of paired observations(Among them, 806 pairs of data are from English databases, and 40 pairs of data are from Chinese databases). The selected studies had to satisfy the following requirements: (1) The retrieved variables included the mean, standard deviation (or standard error), and repeat group. (2) All the control and treatment groups belonged to the same ecosystem and experienced the same environmental and growth conditions. (3) Data from only the control and treatment groups inoculated with potassium-solubilizing microorganisms were analyzed in multiple controlled studies with experimental inoculation treatments. (4) The extracted literature was limited to articles published from 2015 to 2025.

We also recorded climatic characteristics such as the mean annual temperature (MAT) and mean annual precipitation (MAP), collecting climatic data from the NASA Surface Meteorology and Solar Energy Location site (https://power.larc.nasa.gov/) on the basis of the latitude and longitude of the study site, if not reported in the original study.

Data from all published literature tables were extracted manually, whereas data from figures were extracted via the Get Data (https://getdata.sourceforge.net/download.html) tool.

We used Meta-Win 2.0 software (https://www.metawinsoft.com/) to analyze the data for random effects. The effect value was calculated via the natural logarithm of the response ratio (lnRR) in Equation (2) (Luo et al., 2018), which was then converted to upper and lower 95% confidence intervals (CI) in Equation (4) to evaluate the impact of the KSM on crop yield, soil physicochemical parameters, and crop growth indicators. In several studies, the standard error (SE) in Equation (1) had to be translated to the standard deviation (SD) in Equation (1), which was calculated as (Jian et al., 2016):

The variance of the effect value is as follows:

where X ¯ t is the mean of the variable in the treatment group and where X ¯ c is the mean of the variable in the control group in Equation (3). S_t is the standard deviation of the variable in the treatment group; S_c is the standard deviation of the variable in the control group; n is the sample size; and n_t and n_c are the sample sizes of the treatment and control groups, respectively in Equation (3). We converted lnRR and its corresponding confidence intervals to the corresponding percentage change.

The I² statistic was used to assess heterogeneity, and the test value Q was calculated as follows in Equation (5) and (6) (Huo et al., 2017):

where w_i is the weight of group i of data, n is the number of effect sizes, E_i is the effect size of the ith group, and E is the average of all data effect sizes in Equation (5). Calculations were made using the random effect model when the heterogeneity test result I² was greater than 50% and the fixed effect model when it was less than 50%. The Rosenthal loss of safety coefficient method was used for the publication bias test. N>5 m+10 (N is the loss of safety coefficient, and m is the sample size) indicates no publication bias. We used AMOS software (IBM SPSS AMOS 20.0.0) to untangle the indirect and direct effects of climate (MAT and MAP), microbial count, soil available potassium content, crop growth indicators, and soil enzyme activities on crop yields (Eisenhauer et al., 2015) via structural equation modeling (SEM). Before modeling, all the data were normalized (ensuring that the data followed a standard normal distribution with a mean value of 0 and standard deviation of 1), and an a priori model was designed based on the known effects and relationships among the drivers that had a significant impact on crop yields in our previous analyses. SEM was used instead of multiple regressions since directions can be assigned to several relationships, resulting in multiple explanatory and response variables in one model. Furthermore, the structure of such a model can reveal whether a significant bivariate relationship results from a significant relationship between two given variables and a third variable. Despite using fail-safe N to assess publication bias, our results may still be affected by the underreporting of studies with non-significant or negative findings. While subgroup analysis explored substantial heterogeneity (indicated by I²), residual heterogeneity could remain due to unmeasured factors like experimental duration, KSM strain potency, and unrecorded soil microbial communities, which may limit the generalizability of pooled effects (Jin et al., 2015; Mathur and VanderWeele, 2022).

The calculation of the effect size on yield revealed that KSM inoculation improved overall crop yield by 23.43% compared with the control, confirming its potential as a yield-enhancing agricultural practice ( Figure 2 ). Note: The four KSM genera analyzed (Aspergillus, Pseudomonas, Bacillus, Enterobacter) were selected based on two criteria: (1) their high frequency of reporting in existing KSM-related studies (accounting for >70% of eligible literature included in this meta-analysis); (2) their well-documented potassium-solubilizing mechanisms and agricultural application records, ensuring data robustness and practical relevance. Among these genera, Aspergillus (+36.27%) and Pseudomonas (+32.86%) were consistently demonstrated to outperform Bacillus and Enterobacter in boosting crop yields (P<0.05).

Soil texture emerged as a key modulating factor: clay loam soil (+38.82%) showed a significantly stronger yield response to KSM than sandy loam and loam (P<0.05), while soil pH had no substantial impact. Crop type and family also drove marked differences: vegetable crops (+34.25%) exhibited the greatest yield increase relative to food crops and fruit trees (P<0.05), with cruciferous (+42.74%) and leguminous (+40.08%) crops outperforming Poaceae and Anacardiaceae (P<0.05). Additionally, greenhouse experiments (+34.17%) yielded better results than field trials (P<0.05), whereas soil water potential and EC values did not induce significant variations in yield response.

A key chain emerged from the correlation analysis ( Figure 5 ): soil available potassium was significantly and positively correlated with both crop shoot K concentrations (P<0.05) and root K concentrations (P<0.05), while root K concentrations further positively linked to shoot K concentrations (P<0.05). Critically, both soil available potassium and crop shoot K concentrations showed direct positive correlations with crop yield (P<0.05), highlighting that KSM’s yield-enhancing effect is likely mediated through improved soil K availability and subsequent crop K uptake.

The results of the random forest analysis are presented in Figure 6 . We assigned significance to factors influencing yield and other growth indicator changes by considering those with a significance level greater than 0.1. For yield, the key control factor is microbial count and soil available K. For root length, the key control factor is soil available K. For plant height, the key control factors are root length, soil available K, MAT, microbial count, soil pH, MAP, shoot K concentrations, and soil water potential. For leaf area, significant control factors included soil pH, microbial count, and plant height. For root K concentrations, significant control factors included microbial count, MAT, and soil available K. For shoot K concentrations, the significant control factors included soil available K and root K concentrations.

Structural equation modeling (SEM) was used to investigate how the various explanatory variables and their interactions influenced crop yield under KSM inoculation ( Figure 7 ). The final structural equation model explained 19% of the overall variation in crop yield (χ² = 5.09; P = 0.955, AIC = 113.09). The microbial count, MAP, MAT, soil available potassium, root K concentration, root length, leaf area, and oxidase were the most crucial factors affecting crop yield, both directly and indirectly. Overall, the microbial count indirectly affects crop yield by altering root length (path coefficient = 0.76), root K concentration (path coefficient = 0.93), and leaf area (path coefficient = 1.49). MAP indirectly affects crop yield by influencing root length (path coefficient = 0.81), root K concentration (path coefficient = 0.92), and leaf area (path coefficient = 1.69). The MAT had an indirect effect (path coefficient = 0.81), and the root K concentration (path coefficient = 0.92) and leaf area (path coefficient = 1.69) were affected. MAT also indirectly improved crop yield by influencing root K concentrations (path coefficient =–0.21) and leaf area (path coefficient =–0.15). Soil available potassium indirectly affected crop yield by influencing root length (path coefficient = 0.31) and root K concentrations (path coefficient = 0.29). The root K concentrations also indirectly influenced crop yield because of their impact on root length (path coefficient =–0.32). Root length had a direct positive effect on crop yield (path coefficient = 0.21), and leaf area had a direct positive effect on crop yield (path coefficient = 0.24). Oxidase had a direct positive effect on crop yield (path coefficient = 0.14). These results confirm the potential of KSM to increase crop production through inoculation.

Our results showed that KSM significantly increased crop yield, soil available K, and crop potassium uptake and growth, confirming their role in bioweathering potassium-containing minerals and mediating soil potassium transformation (Chen et al., 2022). Elevated root and shoot K concentrations further promote crop growth and yield, which aligns with Olaniyan who emphasized that enhanced potassium nutrition improves root vigor and photosynthetic efficiency (Olaniyan et al., 2022).

Crop yield and development varied with KSM genus, soil texture, and other contextual factors. Among tested microorganisms, Aspergillus showed the strongest overall effect on crop K content and yield, outperforming Bacillus and Pseudomonas. This superiority stems from its synergistic mechanisms: it secretes multiple extracellular organic acids (e.g., oxalic acid) to disrupt potassium-bearing mineral lattices and releases phosphatases/glucosidases (Ashrafi-Saiedlou et al., 2024), while producing phytohormones and iron transporters to optimize root physiological functions (Kumawat et al., 2024). Unlike Bacillus (which primarily relies on acidolysis), Aspergillus also improves rhizosphere microecology and soil aggregation (Imran et al., 2021), making it a key candidate for microbial preparation development and potash fertilizer efficiency improvement.

With respect to soil texture, clay loam showed the strongest yield response to KSM, which is attributed to the unique improvement effect of KSM on this soil type. Clay loam typically suffers from poor aeration that hinders root nutrient uptake (Hsiao et al., 2018), but KSM inoculation maintains stable available potassium levels during the crop reproductive period (Singh et al., 2022) and produces extracellular polysaccharides to promote soil aggregation (Wang et al., 2022). This structural improvement increases aeration and nutrient diffusion, enhancing root K uptake and root length (a trend consistent with the significant increase in root K concentration and length observed in clay loam). These findings indicate that KSM can targeted alleviate the limitations of clay loam soils, providing a solution for yield improvement in such agricultural systems.

With respect to crop classification, our synthesis indicated that vegetable crops responded more positively to KSM than other crop types, which can be explained by their biological traits: most vegetables have short growth cycles, high potassium demand, and quality formation closely linked to potassium nutrition (Zoerb et al., 2014). Compared with cereals, vegetables such as sugar beet exhibit higher K mobilization efficiency per unit root length (El Dessougi et al., 2002; Wang et al., 2011), making them more sensitive to KSM-mediated potassium supplementation. This finding provides a basis for targeted KSM application—prioritizing vegetable crops in agricultural production to maximize the return on microbial inoculation.

With respect to field management, greenhouse conditions significantly enhanced KSM efficacy relative to field trials, as the controlled temperature, humidity, and light optimize both crop growth and microbial survival (Shamshiri et al., 2018; Farvardin et al., 2024). This suggests that KSM application should be paired with appropriate environmental regulation to fully exploit its potential.

Under stress types, KSM consistently improved crop performance across drought and salinity gradients, which is achieved through dual mechanisms: supplementing potassium to maintain physiological functions (e.g., photosynthesis, enzyme activity) (Munsif et al., 2022) and enhancing stress defense systems (e.g., increasing antioxidant enzyme activity) (Feng et al., 2019). For salinity stress, KSM also helps maintain K⁺/Na⁺ balance (Romero-Munar and Aroca, 2023), further confirming its role in improving crop stress resilience.

This study demonstrated a significant linear correlation among soil available potassium, root/shoot potassium concentrations, and crop yield following KSM inoculation, consistent with previous findings (Shen et al., 2016; Ai et al., 2022). The increase in yield can be attributed to a well-defined physiological pathway initiated by KSM-mediated potassium solubilization.

KSM enhances the availability of soil potassium, leading to elevated potassium uptake by roots and subsequent translocation to shoots (Chen et al., 2021). The increased potassium levels stimulate root growth, including surface area, vigor, and biomass, which further expands nutrient absorption capacity and improves water retention (Sustr et al., 2019). Simultaneously, potassium promotes shoot development by facilitating cell expansion, leaf area enlargement, and photosynthetic efficiency (Hu et al., 2022; Ding et al., 2024).

Our SEM analysis revealed a multi-level regulatory network of KSM on yield, with soil available potassium and microbial count as core indirect drivers, and root length/leaf area as direct effectors. Soil available potassium promotes root tip cell division and elongation by regulating growth hormone transport (Sheng, 2005), while microbial count modulates root traits and nutrient availability (Sun et al., 2023), thereby establishing a coherent pathway from potassium activation to root growth and, ultimately, yield increase. Climate factors also play a role: MAP improves soil moisture to enhance root uptake (Li et al., 2023a), while MAT shows dual effects—moderate temperatures promote metabolism, but extreme temperatures inhibit root K uptake (Jacobson et al., 1957), highlighting the need for temperature adaptation in KSM application.

These findings have important implications for sustainable agriculture: KSM can reduce reliance on mineral potash fertilizers by improving soil potassium use efficiency, particularly in clay loam and vegetable cropping systems. However, practical application must consider contextual limitations, for example, KSM efficacy may decline in highly saline soils due to microbial survival constraints. Future research should focus on optimizing inoculation strategies and developing climate-adapted KSM formulations to enhance applicability across diverse agricultural environments.