Above-ground biomass (AGB) is a key indicator for studying grassland productivity and evaluating carbon sequestration capacity; it is also a key area of interest in hyperspectral ecological remote sensing. In this study, we use data from a typical alpine meadow in the Qinghai–Tibet Plateau during the main growing season (July–September), compare the results of various feature selection algorithms to extract an optimal subset of spectral variables, and use machine learning methods and data mining techniques to build an AGB prediction model and realize the optimal inversion of above-ground grassland biomass. The results show that the Lasso and RFE_SVM band filtering machine learning models can effectively select the global optimal feature and improve the prediction effect of the model. The analysis also compares the support vector machine (SVM), least squares regression boosting (LSB), and Gaussian process regression (GPR) AGB inversion models; our findings show that the results of the three models are similar, with the GPR machine learning model achieving the best outcomes. In addition, through the analysis of different data combinations, it is found that the accuracy of AGB inversion can be significantly improved by combining the spectral characteristics with the growing season. Finally, by constructing a machine learning interpretable model to analyze the specific role of features, it was found that the same band plays different roles in different records, and the related results can provide a scientific basis for the research of grassland resource monitoring and estimation.

Alpine wetlands are highly sensitive to climate warming and anthropogenic disturbances such as grazing, highlighting the urgent need to identify operational indicators for monitoring soil functional changes. In this study, the Zequ National Wetland Park on the Qinghai–Tibet Plateau was selected as the study area. At the plot scale (n = 66), vegetation structure (aboveground biomass, vegetation height, and coverage), soil nutrient properties (soil organic carbon (SOC), total nitrogen (TN), total phosphorus (TP), ammonium nitrogen (NH₄⁺–N), nitrate nitrogen (NO₃⁻–N), available phosphorus (AP)), soil enzyme activities (β-glucosidase (BG), N-acetylglucosaminidase (NAG), and acid phosphatase (ACP)) were measured simultaneously. Spearman correlation analysis and redundancy analysis (RDA) were applied to examine their statistical relationships. Descriptive statistics revealed pronounced variability among plots, with aboveground biomass ranging from 115.43 to 1505.27 g·m⁻², AP from 0.75 to 70.23 mg·kg⁻¹, and BG activity from 0.25 to 14.71 μmol·g⁻¹·h⁻¹, indicating strong spatial heterogeneity in alpine wetlands. Both correlation and RDA results consistently showed that nutrient availability—particularly inorganic nitrogen and AP—was more closely associated with soil enzyme activities, whereas total nutrient contents exhibited a relatively limited ability to explain short-term variations in soil functional processes. These findings suggest that a combined indicator framework integrating nutrient availability and soil enzyme activities has strong potential for the early detection of soil quality changes and degradation in alpine wetlands, thereby providing quantitative support for sustainable wetland management and restoration assessment.

As a typical ecologically degraded mining area, the Jiangcang Mine in Qinghai is characterized by severely depleted soil nutrients and reduced biodiversity, making scientifically grounded soil-amelioration measures urgently necessary to facilitate vegetation reconstruction and enhance soil ecological functions. To determine the optimal fertilization rate, we conducted a two-factor randomized block experiment over an approximately two-year period after plant sowing, using pelletized organic fertilizer and sheep manure as the primary amendments, with three replicates per treatment, and with application rates selected based on commonly used ranges in alpine grassland restoration and the availability of local organic resources in the mining area. Sheep-manure treatments were set at three levels at 0 (S0), 20 (S1), and 40 (S2) kg·m⁻², while pelletized organic fertilizer was applied at three rates at 0.0 (F0), 1.5 (F1), and 3.0 (F2) kg·m⁻². The combination of the two factors resulted in nine treatments: S0F0, S0F1, S0F2, S1F0, S1F1, S1F2, S2F0, S2F1, and S2F2. The results showed that fertilization significantly improved vegetation height, canopy cover, plant density, and aboveground biomass, with the strongest promotive effects observed under S2F2 and S2F1. Compared with other treatments, S2F2 markedly increased soil moisture content, pH, soil organic matter, available nitrogen, available phosphorus, as well as total nitrogen, 6.96-, 2.91-, 1.70-, 5.04-, 2.51-, and 3.91-fold relative to the control, respectively. The S0F2 treatment significantly enhanced bacterial Observed Richness, Shannon, and Chao1 indices, as well as simultaneously increasing fungal Observed Richness and Chao1 index. Vegetation height and density exhibited the strongest positive correlation under S2F1, whereas vegetation cover and aboveground biomass were most strongly correlated under S2F2. A gray relational analysis performed on 15 indicators ranked S2F0 as having the highest relational degree and comprehensive score, followed by S2F2 and S2F1. In summary, the combined application of approximately 40 kg·m⁻² of sheep manure without pelletized organic fertilizer showed the highest comprehensive restoration performance under the experimental conditions of this alpine mining area.

Improving surface-soil carbon accumulation is important for restoring reclaimed alpine mining soils. We evaluated nine treatments in the Muli mining area, Qinghai, China: a control (CK) and eight sheep manure cake rates (0.025–0.060 m³·m⁻²). Soil total organic carbon (TOC), particulate organic carbon (POC), microbial biomass carbon (MBC), dissolved organic carbon (DOC), 0–10 cm soil organic carbon (SOC) stock, carbon pool management index (CPMI), carbon fixation genes (cbbL and cbbM), and microbial community sequencing profiles were assessed. Sheep manure cake significantly increased the measured carbon fractions and CPMI. The 0–10 cm SOC stock was highest under SM3 (75.82 ± 3.52 Mg C ha⁻¹), followed by SM4 (65.99 ± 3.49 Mg C ha⁻¹), compared with 20.74 ± 1.52 Mg C ha⁻¹ in CK. cbbL and cbbM gene copy numbers peaked under SM2 (0.030 m³·m⁻²), reaching 1.51 × 10⁹ and 1.91 × 10⁷ copies·g⁻¹, respectively. Under the prespecified equal criterion-layer weights, TOPSIS ranked SM2 first; however, sensitivity analysis showed that the leading treatment changed when criterion weights were altered. Thus, SM3–SM4 showed the strongest measured surface-carbon accumulation, whereas SM2 was the highest-ranked treatment only within the adopted TOPSIS weighting scheme.

Alpine wetland ecosystems on the Qinghai–Tibet Plateau are highly vulnerable to hydrological alterations, yet whether alpine wetland plants occupying different hydrological niches exhibit distinct adaptive strategies under waterlogging remains poorly understood. Here, two dominant alpine wetland species, Kobresia tibetica and Blysmus sinocompressus, were exposed to four waterlogging treatments: control (CK), slight waterlogging (SW), moderate waterlogging (MSW), and severe waterlogging (SSW). Root functional traits, rhizosphere soil enzyme activities, physiological responses, and root transcriptomic profiles were analyzed to elucidate species-specific responses to waterlogging stress. Slight waterlogging promoted root growth, whereas moderate and severe waterlogging inhibited root development and impaired physiological functions in both species. Root transcriptomic analysis under moderate waterlogging identified 123 and 2806 differentially expressed genes in K. tibetica and B. sinocompressus, respectively. These genes were mainly associated with redox regulation and stress responses in K. tibetica, whereas carbon metabolism, energy metabolism, and stress-related pathways were enriched in B. sinocompressus. Overall, the two alpine wetland species exhibited contrasting species-specific waterlogging response strategies, with B. sinocompressus showing greater capacity to maintain belowground functional stability under moderate and severe waterlogging. Quantitative reverse transcription polymerase chain reaction (qRT-PCR) validation further confirmed that waterlogging-responsive genes involved in anaerobic metabolism and antioxidant regulation showed species-specific expression patterns under severe waterlogging conditions.

Drought–rewatering events are becoming increasingly frequent on the Qinghai–Tibet Plateau, yet how drought severity shapes post-drought recovery of root functions and associated rhizosphere enzyme activities remains unclear. Here, Poa pratensis from Zeku Zequ National Wetland Park was subjected to a control treatment (75–80% field capacity, FC), mild (60–65% FC), moderate (40–50% FC), or severe drought (30–35% FC) for 14 days, followed by 14 days of rewatering. Root functional traits, leaf physiological responses, chlorophyll-related traits, and rhizosphere soil enzyme activities were evaluated. Drought severity progressively impaired root function, chlorophyll-related traits, and rhizosphere enzyme activities, whereas moderate drought induced the strongest osmotic adjustment and antioxidant enzyme responses. After rewatering, most traits under mild drought approached control levels, moderate drought showed partial restoration, and severe drought retained substantial deficits in root function, water status, chlorophyll-related traits, and rhizosphere enzyme activities. Pearson correlation analysis and PCA further revealed coordinated variation among root functions, physiological status, and rhizosphere enzyme activities. These findings demonstrate that drought severity strongly influences post-drought restoration trajectories of root functions and rhizosphere processes.