In arid regions, decreased soil fertility has adversely affected agricultural sustainability. The effects of different amendments in alleviating these issues and increasing soil fertility remain unclear. Herein, a two-year field experiment was conducted to evaluate the properties of grey desert soil and soil respiration (SR) dynamics under six different treatment groups: biochar (BC), leonardite (LD), anionic polyacrylamide (PAM⁻), cationic polyacrylamide (PAM⁺) powder, anionic polyacrylamide solution in water (PAM⁻W), and control (CK). We observed that the BC and LD amendments significantly altered soil pH, organic matter, available nitrogen, available phosphorus, cation exchange capacity, and SR. PAM amendment increased the SR as compared to the control, except in autumn, but PAM did not affect the soil properties. SR under different amendments showed strong seasonal patterns, the highest and lowest SR rates were observed in June and January, respectively. Amendments and seasonal dynamics significantly affected SR, but no interaction was observed between these factors. Temporal variation of SR was substantially influenced by soil temperature at 15 cm of soil depth. Temperature sensitivity of SR (Q₁₀) increased with soil depth and decreased with amendment addition. SR was significantly affected by soil temperature, moisture, air temperature, and their interactions. The outcomes of this study suggested that the BC and LD amendments improved soil fertility and negated the net carbon accumulation by increasing the SR and Q₁₀ in arid agriculture soil.

Introduction: Soil microorganisms maintain ecosystem multifunctionality, yet the group-specific pathways linking microbial diversity, community assembly, and network complexity to soil multifunctionality (SMF) during natural restoration of abandoned alpine grasslands remain unclear. Methods: We examined artificial grassland (AG) and 3-, 12-, and 31-year abandoned grasslands (AB03, AB12, and AB31) in Bayinbuluk, Xinjiang, and analyzed bacterial, fungal, and protistan diversity, community assembly, and co-occurrence networks in relation to SMF. Results: SMF was significantly higher at AB12 and AB31 than at AG. Stochastic processes dominated community assembly in all three groups: homogenizing dispersal in bacteria, drift in fungi, and in protists a shift from dispersal limitation at AB03 (90%) to drift at AB12 (100%). Bacterial network complexity was lowest at AB03, fungal complexity increased at AB31, and protistan complexity was unchanged. Partial least squares path modeling (PLS-PM) revealed divergent microbial pathways: fungal diversity and bacterial and protistan network complexity were positively associated with SMF, whereas fungal network complexity and protistan diversity showed negative associations. Discussion: Microbial–SMF linkages thus reflect a strategy of diversity dependence in fungi versus network dependence in bacteria and protists, indicating that assessing alpine grassland restoration requires group-specific rather than universal microbial indicators.