The rainfall in loess hilly and gully area is concentrated, and mostly comes in the form of rainstorms. The runoff on the slope caused by rainstorms is the main cause of serious soil and water loss in the loess hilly area, and the grassland vegetation has a good inhibitory effect on the runoff on the slope. Therefore, it is of great significance to reveal the role of grassland vegetation in the process of runoff generation, and the mechanisms for controlling soil erosion in this area. In this study, typical grassland slopes in hilly and gully regions of the loess plateau were taken as research objects. Through artificial rainfall in the field, the response rules of the slope rainfall-runoff process to different grass coverage were explored. The results show that: (1) With the increase in rainfall intensity, the inhibitory effect of grassland vegetation on slope runoff decreased, which was mainly reflected in the gradual decrease in runoff rate and runoff coefficient, and the time required to reach stability gradually shortened. (2) Under 60 mm/h rainfall intensity, the sensitivity of runoff coefficient to 31.5% of grass cover change is the lowest, and the cost performance of grass cover with 55% coverage is the highest. (3) Grass coverage inhibited slope runoff by changing the hydraulic characteristics of the slope, but this effect was only obvious in low rainfall intensity and early rainfall. Rainfall in the loess hilly area is characterized by intense rain. The regulating effect of grass cover on slope runoff is not particularly significant under high intensity rainfall. If only considering the regulation of grassland vegetation on slopes, more than 60% grassland coverage is more efficient in inhibiting slope runoff under medium and low intensity rainfall.

Understory vegetation is one of the most important links for improving forest biodiversity, and its restoration is conducive to sustainable forest development, energy flow, and nutrient cycling. However, little is known about the developmental dynamics and main driving factors of the long-time series coverage, biomass, diversity, and species composition of plantation understory vegetation. In a case study of three typical plantations, with a natural secondary forest as reference in the Loess Plateau of China, we collected understory vegetation from a Robinia pseudoacacia Linn. deciduous broad-leaved plantation, Pinus tabulaeformis Carr. evergreen coniferous plantation, and mixed plantation with an age span of 10 to 50 years. (1) The understory plantation coverage and biomass results of stands with different ages showed the R. pseudoacacia plantation to be significantly higher than the P. tabulaeformis plantation, and the species diversity of the P. tabulaeformis plantation changed the most with the stand age. However, the growth resource imbalance, and drastic changes in the stands’ environment caused by excessive intraspecific competition in the early stage of the P. tabulaeformis plantation vegetation restoration, are the main reasons that make the species diversity of undergrowth vegetation of P. tabulaeformis plantation lower than that of other stand types. (2) The understory species composition of the plantations revealed their degree of community stability. Compared to the R. pseudoacacia plantation and P. tabulaeformis plantation, the mixed plantation had higher stability, and its species composition closely resembled a natural secondary forest. The community stability of the P. tabulaeformis plantation was the lowest because it had the lowest coverage, biomass, and species diversity of understory vegetation. However, the understory species composition of the three plantation types converged, which was due to atypical species contribution. (3) The dynamic changes of canopy and soil nutrients were the main driving factors affecting the R. pseudoacacia plantation understory vegetation species composition. Stand density and elevation limited the understory vegetation communities of P. tabulaeformis plantation restoration. Soil bulk density is the key factor affecting understory vegetation in mixed plantations, and this effect weakens with the stand age. In future studies, the focus should be on the converged action and further development trend of atypical species, choosing an appropriate recovery strategy (active or passive), and providing more possibilities for the intensive management of vegetation under different plantations.

Micro-videos have recently gained immense popularity, sparking critical research in micro-video recommendation with significant implications for the entertainment, advertising, and e-commerce industries. However, the lack of large-scale public micro-video datasets poses a major challenge for developing effective recommender systems. To address this challenge, we introduce a very large micro-video recommendation dataset, named "MicroLens", consisting of one billion user-item interaction behaviors, 34 million users, and one million micro-videos. This dataset also contains various raw modality information about videos, including titles, cover images, audio, and full-length videos. MicroLens serves as a benchmark for content-driven micro-video recommendation, enabling researchers to utilize various modalities of video information for recommendation, rather than relying solely on item IDs or off-the-shelf video features extracted from a pre-trained network. Our benchmarking of multiple recommender models and video encoders on MicroLens has yielded valuable insights into the performance of micro-video recommendation. We believe that this dataset will not only benefit the recommender system community but also promote the development of the video understanding field. Our datasets and code are available at https://github.com/westlake-repl/MicroLens.

Conflicts between humans and land use in the process of using water and conflicts between humans and water resources in the process of using land have led to an imbalance between natural ecosystems and socio-economic systems. It is difficult to understand the impact of the processes of water production and consumption on land patches and their ecological effects. A grid-type, basin-distributed hydrological model was established in this study, which was based on land-use units and coupled with groundwater modules to simulate the water production and consumption processes in different units. By combining land use and net primary productivity, the runoff coefficient and the water use efficiency (NPP/ET) of different land units were used as indicators to characterize the interaction between water and land resources. The results showed that the average runoff coefficients of cultivated land, forest land and grassland were 0.7, 0.5 and 0.9, respectively. Moreover, the average runoff coefficients of hills, plains and basins were 0.7, 0.7 and 0.8, respectively. The NPP produced by the average unit, evapotranspiration, in cultivated land, forest land and grassland was 7 (gC/(m²•a))/mm, 0.7 (gC/(m²•a))/mm and 0.2 (gC/(m²•a))/mm, respectively. These results provide quantitative scientific and technological support in favor of the comprehensive ecological management of river basins.

The dynamic changes in vegetation significantly impact the sustainability, safety, and stability of ecosystems in the source region of the Yellow River. However, the spatiotemporal patterns and driving factors of these changes remain unclear. The MODIS NDVI dataset (1998–2018), together with climatic records from meteorological stations and socio-economic statistics, was collected to investigate the spatiotemporal characteristics of vegetation coverage in the study area. For the analysis, we employed linear trend analysis to assess long-term changes, Pearson correlation analysis to examine the relationships between vegetation dynamics and climatic as well as anthropogenic factors, and t-tests to evaluate the statistical significance of the results. The results indicated the following: (1) From 1998 to 2018, vegetation in the source region of the Yellow River generally exhibited an increasing trend, with 92.7% of the area showed improvement, while only 7.3% experienced degradation. The greatest vegetation increase occurred in areas with elevations of 3250–3750 m, whereas vegetation decline was mainly concentrated in regions with elevations of 5250–6250 m. (2) Seasonal differences in vegetation trends were observed, with significant increases in spring, summer, and winter, and a non-significant decrease in autumn. Vegetation degradation in summer and autumn remains a concern, primarily in southeastern and lower-elevation areas, affecting 25% and 27% of the total area, respectively. The maximum annual average NDVI was 0.70, occurring in 2018, while the minimum value was 0.59, observed in 2003. (3) Strong correlations were observed between vegetation dynamics and climatic variables, with temperature and precipitation showing significant positive correlations with vegetation (r = 0.66 and 0.60, respectively; p < 0.01, t-test), suggesting that increases in temperature and precipitation serve as primary drivers for vegetation improvement. (4) Anthropogenic factors, particularly overgrazing and rapid population growth (both human and livestock), were identified as major contributors to the degradation of low-altitude alpine grasslands during summer and autumn periods, with notable impacts observed in counties with higher livestock density and population growth, indicating that for each unit increase in population trend, the NDVI trend decreases by an average of 0.0001. The findings of this research are expected to inform the design and implementation of targeted ecological conservation and restoration strategies in the source region of the Yellow River, such as optimizing land-use planning, guiding reforestation and grassland management efforts, and establishing region-specific policies to mitigate the impacts of climate change and human activities on vegetation ecosystems.