The application of phosphate fertilizers significantly influences soil microbial communities and nutrient cycling. Soil enzymes, which are sensitive to nutrient levels, play a critical role in microbial metabolism. However, the impact of phosphate fertilizers on nutrient limitations within the microbial metabolism of agricultural soils remains poorly understood. In this study, soil samples were collected from a depth of 0–20 cm from a wheat crop subjected to a three-year field experiment with six different phosphorus (P) application rates. Soil β-glucosidase (BG) and leucine aminopeptidase (LAP) activities were highest under the P3 (60 kg P₂O₅ ha⁻¹) treatment over the three-year study period. The responses of soil N-acetyl-β-glucosaminidase (NAG) and alkaline phosphatase (AKP) to increasing P additions varied across different years. The EES C:N, C:P, and vector length were significantly greater than 1. Soil nutrient characteristics accounted for 70.71% of the variation in soil enzyme stoichiometry. The vector length and angle of soil enzymes explained by soil nutrient characteristics were 0.65 and 0.73, respectively. Among these factors, ROC exhibited the largest direct and total effect on the soil enzyme vector length and angle. These research findings offer valuable insights for the management of agricultural fertilizers. Consequently, it is recommended to enhance soil carbon levels to alleviate carbon limitations and improve P utilization efficiency.

Plant patches are a common feature of dryland ecosystems, but how within-patch spatial position is associated with plant functional traits–and whether these patterns differ among organs—remains poorly understood. We sampled twigs, leaves, and flowers of the C₄ desert plant Salsola collina from the centers and edges of natural patches in arid Xinjiang, China, using five plots with three patches per plot. For each organ, we measured the concentrations of carbon (C), nitrogen (N), phosphorus (P), sucrose (SUC), fructose (FRU), total soluble sugars (SS), and starch (ST), and calculated non-structural carbohydrate (NSC) concentrations as the sum of SS and ST. Treating the plot as the experimental unit and accounting for paired center–edge observations within plots, we evaluated the effects of patch position, organ type, and their interaction using two-way ANOVA and distance-based multivariate analyses. Organ type accounted for the largest proportion of multivariate trait variation (R² = 0.90), whereas patch position and the Patch × Organ interaction each explained approximately 4% (all p < 0.001). Significant Patch × Organ interactions were detected for 14 of 17 traits, indicating that center–edge differences varied among organs for most traits. N concentrations were higher at patch edges in all three organs, whereas P concentrations were higher in twigs and flowers only; C showed no significant center–edge differences. NSC concentrations were higher at patch edges in all three organs, while SS was higher in edge twigs and leaves and ST was higher in edge flowers. These patterns were accompanied by contrasting organ-specific differences in carbohydrate composition and elemental stoichiometry. Because environmental variables were not measured concurrently, the observed patterns should be interpreted as spatial associations rather than demonstrated environmental mechanisms. These findings indicate that organ identity and within-patch position are jointly associated with variation in the carbon-, nutrient-, and carbohydrate-related traits of S. collina, highlighting the importance of considering organ-specific variation when evaluating fine-scale spatial heterogeneity in dryland plants.

Biological soil crusts (BSCs) are important components of dryland ecosystems, yet nutrient patterns in adjacent uncovered soils remain unclear. We surveyed 70 sites across the Gurbantunggut Desert and collected 209 composite samples from the adjacent 0–5 cm surface-soil layer, categorized into three groups: uncrusted bare sand (n = 32), soil adjacent to algal–lichen crusts (n = 48), and soil adjacent to moss crusts (n = 129), with bare sand serving as the uncrusted reference category. The results showed that: (1) total nitrogen differed among the three BSC-associated soil categories (p = 0.041), whereas soil organic carbon, total phosphorus, and total potassium did not (all p > 0.05); (2) NO₃⁻-N, NH₄⁺-N, extractable inorganic N, NaHCO₃-extractable phosphorus, and NH₄OAc-extractable potassium differed significantly among categories (all p < 0.001) and were generally highest in soil adjacent to moss crusts; and (3) random forest models explained approximately 38–65% of nutrient variation and identified EC, total nitrogen, and the site-level BSC metric as the leading predictors. The final piecewise structural equation model explained 35.3–76% of nutrient variation. These findings indicate marked regional heterogeneity in extractable nutrient fractions associated with different BSC types.