The issue of multi-scale driving forces within land systems has emerged as one of the pivotal research directions for innovative exploration in the field of land science. However, the understanding of the differences in driving factors across different scales remains relatively inadequate. Based on multi-source data spanning from 2000 to 2020, this study applied GeoDetector models to reveal the evolution of the spatiotemporal pattern of land-use change and the scale differences in driving factors in rural areas of the Loess Plateau region from both prefecture-level scale and township scale. The results indicated that the land-use changes in rural areas of the Loess Plateau had exhibited distinct spatial differentiation characteristics over the past 20 years. Specifically, the land-use change dynamic degree on the Loess Plateau exhibits an upward trend from west to east (slope = 0.031) and a downward trend from north to south (slope = −0.039). Secondly, the spatial scale differences in driving forces of rural land-use change in the Loess Plateau were manifested through variations in dominant factors and differences in the sensitivity of land-use change to various factors. The reasons for these differences lay in scale effects and cumulative effects. These findings would provide decision-making support for policymakers in formulating future sustainable land-use policies for rural areas. Additionally, it would contribute to further advancing the exploration of multi-scale driving forces within rural land systems.

Climate warming and intensified extreme climate weathers have diverse influence on global food production over past decades. However, there is insufficient understanding of the adaptation in future climate change to fully release food production potential on the existing cultivated land. Here we separately simulated spatio-temporal evolution of per unit yield, potential yields and yield gaps for rice, maize, wheat and soybean under four future climate scenarios in the near-term, mid-term and long-term future. The potential yields of rice, maize, wheat and soybean will be significantly rising, and the yield gaps will separately reach 4.01×10⁸ t, 3.41×10⁸ t, 5.71×10⁸ t, 1.11×10⁸ t under the optimal scenario in the long-term future. We divided three types to differentiate future food production potential in different countries, including steady increased production type, fluctuation increased production type and overload production type, and global rice (77.45%), maize (67.46%), wheat (76.64%) and soybean (69.70%) growing countries will be mainly steady increased production type. We intergrate simulation results under differnt future climate scenarios and propose differentiated sustainable strategies for each food production potential type to ensure future global food security and enhance agricultural system resilience.

Against the backdrop of rapid urbanization, the human–land relationship in the mountainous regions of Southwest China (Sichuan, Yunnan, Guangxi, Chongqing, and Guizhou) confronts dual pressures from terrain constraints and development demands, shaping a uniquely complex evolutionary pattern. To clarify the evolutionary laws of the regional human–land system, this study focuses on the period of 2000–2020, integrating land use, socioeconomic, and topographic data to construct a comprehensive analytical framework of “Human Activity Intensity (HAI)–Land Use Dynamic Degree (LUDD)–decoupling model–geographic detector.” This framework is employed to explore the spatio-temporal evolution characteristics of the human–land pattern, the differentiation of decoupling modes, and the underlying driving mechanisms. The key findings are as follows: Human Activity Intensity (HAI) presents a stable spatial pattern of “agglomeration in low-altitude areas and dispersion in high-altitude areas,” undergoing a three-stage temporal evolution of “terrain anchoring–policy constraint–all-round expansion.” Land use dynamics are predominantly governed by terrain: low-altitude river valley plains exhibit significant changes, while high-altitude karst regions remain relatively stable, with an overall policy-responsive fluctuation of “rise–fall–rebound.” Human–land decoupling forms a continuous spectrum encompassing four modes: “collaborative optimization–extensive transition–rigid stagnation–advantageous aggregation,” with strong negative decoupling dominating low-altitude favorable areas and recessive decoupling prevailing in high-altitude mountainous areas. In terms of driving mechanisms, terrain factors serve as the rigid foundation of the human–land relationship, while the urban–rural population structure, urbanization level, and land use intensity act as core human drivers. Additionally, the interaction of factors such as “terrain–economy–transportation” plays a crucial role in the differentiation of decoupling modes. This study clarifies the evolutionary logic of “terrain laying the foundation and human factors shaping the pattern” for the human–land relationship in Southwest China’s mountainous regions, providing scientific support for the coordinated advancement of regional economic development and ecological protection, as well as a Chinese case study for global research on human–land coordination in ecologically fragile mountainous areas.