Groundwater storage (GWS) is essential for supporting agricultural irrigation and revegetation in the water-scarce Yellow River Basin (YRB). Early studies have mainly focused on the impacts of revegetation on GWS, and rarely consider the influences of agricultural irrigation and other human activities, rendering the driving mechanisms of GWS unclear. Here we used NASA’s Gravity Recovery and Climate Experiment (GRACE) satellite data, the PCR-GLOBWB2 hydrologic model, and an Long Short-Term Memory (LSTM) machine learning approach to reveal changes, driving mechanisms, and future trends in GWS in the YRB. Results show that GWS in the YRB decreased by ∼101 Gt in 2003−2020, roughly 24 times the Yellow River’s flow into the sea in 2000. Notably, GWS depletion (−7.7 mm/yr) dominates the observed terrestrial water storage (TWS) losses (−6.0 mm/yr) and accounts for >100% of the net TWS decline. Storage losses are largely explained by increases in evapotranspiration (+6.0 mm/yr) driven by revegetation and agricultural irrigation. This is evident in higher evapotranspiration rates (+3 mm/yr) observed in heavily revegetated areas, with irrigation showing an estimated contribution of −6.6 mm/yr on GWS by the PCR-GLOBWB2 model. GWS losses are projected to persist until 2060 by the LSTM model, with a total storage loss of ∼237 Gt. With GWS declining and natural recharge growth lagging behind the rise in groundwater demand, the YRB confronts a future of groundwater deficits. The study suggests that although groundwater extraction for agricultural and ecological benefits might appear helpful to the region in the short term, this trajectory is physically unsustainable and detrimental to the water-scarce Yellow River.

Accurately locating poor populations is increasingly urgent as global poverty reduction has stalled under the combined pressures of conflicts, climate shocks, rising food prices, pandemics, and growing inequality. Recent studies harnessing geospatial big data and machine learning (ML) have significantly advanced poverty mapping, enabling granular and timely welfare estimates in traditionally data-scarce regions. While much of the existing research has focused on overall out-of-sample predictive performance, there is a lack of understanding regarding where such models underperform and whether key spatial relationships might vary across places. This study investigates spatial heterogeneity in ML-based poverty mapping in East Africa, testing whether spatial regression and ML techniques produce more unbiased predictions. We find that extrapolation into unsurveyed areas suffers from biases that spatial methods do not resolve; welfare is overestimated in impoverished regions, rural areas, and single sector-focused economies, whereas it tends to be underestimated in wealthier, urbanized, and diversified economies. Even as spatial models improve overall predictive accuracy, enhancements in traditionally underperforming areas remain marginal. This underscores the need for more representative training datasets and better remotely sensed proxies, especially for poor and rural regions, in future research related to ML-based poverty mapping. For development agencies, the findings caution against treating ML-based outputs as neutral or universally reliable, highlighting instead the need to pair technical advances with investments in inclusive data collection, integration of spatial theory, and institutional strategies that address structural data inequalities.

Land System Science (LSS) has evolved as a core interdisciplinary field within human-environment system research, with a particular focus on land use and land cover change (LUCC). This article reviews the emergence of LSS, explores its roles in social-ecological research on global environmental change and sustainability, and discusses its challenges and future directions. We develop a conceptual framework that highlights the role of LSS in informing sustainable land management and assessing its impacts on interrelated social-ecological goals (sustainability, resilience, and quality of life, including wellbeing) for transformative planning and governance. To ensure the continued progress of the field and its ability to address evolving global challenges, LSS needs to better implement a systems-based approach through novel methodological developments, deepen the understanding of LUCC complexities, emphasize strong sustainability, bridge global-local gaps, and enhance the science-policy interface. In addition, while LSS is inherently interdisciplinary, its progress requires further broadening and deepening of collaboration and integration among contributing disciplines.

Amidst rapid digitalization and pressing environmental challenges, understanding the environmental implications of digital transformation is crucial for sustainable urban development. Yet, the complex, potentially nonlinear digitalization-environment relationships remain underexplored. This study has two objectives: first, to quantify the nonlinear causal impacts of digital transformation on pollution mitigation and carbon reduction; and second, to unravel the mediating pathways that drive these outcomes. We employ Double Machine Learning (DML) on panel data from 2013 to 2022 across China’s four mega-urban agglomerations to identify the nonlinear environmental impacts of digital transformation. Mediation analysis is then used to examine the technology, structure, governance, and scale pathways. Despite overall progress in both digital transformation and environmental performance, significant regional variations persist. Our DML analysis reveals distinct nonlinearities: an S-shaped relationship between digital transformation and pollution mitigation, and a more complex N-shaped curve for the digital transformation-carbon reduction nexus. Mediation analysis further reveals complex mechanism: while the structure path consistently promotes environmental benefits, technology and scale factors show negative effects, and governance impacts diverge, promoting pollution mitigation but hindering carbon reduction. Translating digital transformation into environmental benefits necessitates a multi-pronged strategy. Key imperatives include prioritizing green technological innovation over sheer digital expansion to mitigate adverse scale effects, and restructuring energy systems towards renewable sources. Furthermore, digital governance must be wielded judiciously, with accountability to enhance specific environmental goals. This research reveals the intricate and context-dependent nature of digital transformation’s environmental effects, providing data-driven insights for regional policies aiming to leveraging digitalization for environmental sustainability, particularly in urban contexts.

Ecosystem services (ES) use and environmental attitudes (EA) represent a central yet theoretically contested and empirically inconclusive relationship in the human-environment feedbacks, highlighting the urgent need for longitudinal and context-specific investigation. Using China as a case study, we quantified direct ES use for basic needs (cooking, drinking, working, and housing) based on provincial-level census data (2000, 2010, 2020), and measured EA using the New Ecological Paradigm (NEP) Scale from the Chinese General Social Survey. Spearman correlation and hierarchical clustering were used to examine the direct ES use and EA relationship. We found that (1) direct ES use for basic needs declined by over 60 % on average from 2000 to 2020, with fewer than 25 % of households still directly relying on ES in 2020, particularly among rural residents in northeastern and southwestern China. (2) NEP scores were lower in rural areas and declined in most provinces from 2010 to 2020. (3) Direct ES use and NEP scores were negatively correlated, with a stronger relationship observed in 2010 than in 2020. (4) Socioeconomically developed provinces with lower direct ES use generally showed higher EA, while higher ES dependence—especially on biomass—was linked to lower EA. Our findings highlight the challenges in fostering sustainable behaviors in regions where direct ES use remains high. Targeted interventions, such as enhanced environmental education and sustainable resource management, are needed to strengthen the linkage between ecosystem dependence and environmental awareness.

Pursuing sustainable growth of Farmland Green Production Efficiency (FGPE) is crucial for achieving multiple Sustainable Development Goals (SDGs), particularly SDG2 (Zero Hunger), SDG 12 (Responsible Consumption and Production), SDG13 (Climate Action), and SDG 15 (Life on Land) in ecologically fragile basins. However, the dynamics, drivers, and challenges of FGPE remain underexplored within the SDGs framework, especially at finer spatial scale and across coupled “society-economy-policy-climate” (SEPC) drivers. This study develops an FGPE assessment framework through SDGs lens, utilizing “elements-processes-functions-drivers” paradigm and data from 447 counties in the Yellow River Basin (YRB) from 2000 to 2022. We apply super-efficiency Slacks-Based Measure and Malmquist-Luenberger (SBM-ML), spatial correlation analysis, and geographically and temporally weighted regression models to assess FGPE growth challenges. Results reveal a “U-shaped” trend in the number of high-FGPE counties, with a rapid increase after 2016. Specifically, FGPE across the YRB increased by 97.6 % from 2000 to 2022, though spatial correlation declined by 44.7 %, indicating weakening spatial spillover effect. Despite the overall progress, the YRB still faces multiple challenges, including uneven regional development, weakening spatial correlation, climate sensitivity, economic structural shift, and weak policy effect. The findings highlight that FGPE improvement align with key SDGs targets, including enhancing food security (SDG 2), promoting sustainable production (SDG 12), increasing climate resilience (SDG 13), and conserving land ecosystems (SDG 15). Region-specific strategies are recommended: enhancing climate resilience and ecological conservation in the upper reaches, promoting technological diffusion via urban-industrial transformation in the middle reaches, and advancing green agricultural technologies with more local financial support in the lower reaches.

Urban green space may impact human health through complex pathways and the effect can vary across different travel contexts. Revealing these disparities in health pathways between different travel contexts may provide essential and practical suggestions for sustainable developments in urban environments. In this study, we investigated the impacts of travel contexts on people’s perceptions and evaluations of green space using a cross-sectional dataset collected in Hong Kong, China. Eight hundred participants in 4 representative communities were recruited through stratified sampling, and we identified 2,913 travel events from their two-day activity-travel diaries after rigorous cross-validation with GPS-derived trajectories. We also derived two green space exposure representations using fine-grained remote sensing imagery and 8 representative green space exposure indicators. Eighty logistical regression models and mixed-effects models were developed to investigate the associations with control of a range of potential uncertainties. Our results indicate solid and consistently positive associations between participants’ measured green space exposure and perceived green space, and significant but variable effects of travel purposes, travel modes, and travel time on participants’ perceptions and evaluations of green space. Walking significantly promotes participants’ perceptions and positive evaluations of urban green space, buses are not significantly associated, and metro trains may depress the perception and evaluation. Our results provide solid evidence on how travel contexts may influence people’s perceptions and evaluations of urban green space and, thus, provide essential insights into environmental health studies and sustainable urban planning that consider green space as an important urban environmental setting.

The digital transformation of territorial spatial planning has underscored the urgent need to integrate ecological network into spatial planning practices. In response, we developed two innovative new tools, the Ecological Linkage Tool (ELT) and the Relative Spatial Conflict Index (RSCI), to enhance ecological networks applications by addressing spatial conflicts and structural resilience. The ELT identified ecological corridors within and outside irregular ecological sources, activation points, and stepping stones in parallel, and then constructed an intact ecological network. By integrating the RSCI and complex network metrics, the spatial conflicts and structural resilience were evaluated. The framework was implemented in the Hohhot-Baotou-Ordos-Yulin (HBOY) urban agglomeration, identifying a total of 5,814 corridors, of which 67 % were classified as intra-patch and 33 % as inter-patch. The number and distribution of these corridors were determined by the size and shape of the ecological sources, and the connectivity of intra-patch corridors was 34 % higher than inter-patch corridors. According to the RSCI, 60 % of the corridors experienced spatial conflicts, with 21 % involving production spaces or composite production-related conflicts. Moreover, Yulin served as a key hub in the ecological network, and Baotou had the highest network efficiency. Compound conflict corridors (involving production, living, and open spaces) had a greater impact on overall ecological network efficiency compared to those with single or dual conflicts. Meanwhile, the failure of 40 % of corridors without spatial conflicts would directly result in a 96.9 % decline in network efficiency, highlighting their critical role in maintaining network functionality. This study provides an enhanced ecological network application solution for the China Spatial Planning Observation Network (CSPON), supporting spatial planning practices.

Improvements in societal well-being are associated with varying levels of environmental impact across countries, posing a major challenge for global sustainability. As climate and development agendas converge, there is growing urgency to understand how countries can advance social progress while minimizing environmental harm. This study is motivated by the hypothesis that a just transition—defined as the shift toward low-carbon societies in which no one is left behind—requires convergence in countries’ ability to decouple social progress from carbon emissions. We introduce the Carbon Efficiency of Social Progress (CESP), a composite indicator constructed as the ratio of per capita carbon footprint to the Social Progress Index (SPI), and analyze its dynamics across 160 countries from 1990 to 2020. Using time-varying factor models based on the Phillips and Sul methodology, we assess whether countries converge in their CESP performance globally and within emission-based subgroups. Our results reveal distinct patterns. Low-emitting countries (below 1.9 tCO₂e per capita) exhibit signs of rapid convergence (β = 0.50; SE = 0.047). High-emitting countries (above 10 tCO₂e per capita) also converge, though at a much slower rate (β = 0.13; SE = 0.179). By contrast, middle-emitting countries (1.9-10 tCO₂e per capita) exhibit significant divergence (β = −0.58; SE = 0.120). These findings indicate that while low-emitting nations follow relatively promising paths, middle- and high-emitting countries face structural barriers that hinder progress toward just transition. Our study contributes to the literature on sustainability convergence and offers insights into how countries can align social development with climate goals under the global just transition framework.

China’s poverty alleviation and elimination campaign (PAEC, 2015-2021) aimed to eliminate absolute poverty and reduce social inequality, making improvements in the accessibility and equity of basic educational facilities a key component. However, existing research on the accessibility of educational facilities in China has predominantly focused on developed urban areas or specific regions, lacking nationwide spatiotemporal assessments. To address this gap, this study systematically evaluates the accessibility and equity of basic educational facilities (kindergartens, primary, and secondary schools) across mainland of China during the PAEC. Travel time cost, derived using multi-source geospatial datasets and the nearest neighbor method, was adopted as the primary indicator for assessing educational accessibility and equity. The results revealed pronounced disparities in accessibility between Eastern and Western China. Overall, both the accessibility and equity of basic educational facilities improved substantially during the campaign, with the average travel time per person reducing by approximately 50 %. Notably, the rate of improvement in impoverished regions was nearly double that observed in non-impoverished areas. Although widespread improvements in educational equity have occurred, the urban-rural disparity persists as a primary barrier to achieving comprehensive educational fairness. This study offers empirical evidence and methodological innovations for optimizing educational resource allocation and provides high-resolution temporal data to support the monitoring and evaluation of progress toward the Sustainable Development Goals (SDGs) in education.

While extensive research has demonstrated the positive association between urban greenspace and elderly health, the mediating role of ecosystem service flows (i.e., the actual supply of ecosystem services accessible to residents) remains poorly understood. Drawing upon data from 215,199 elderly participants in Shanghai’s Seventh National Population Census, this study employed structural equation modeling and bivariate Moran’s I analysis to investigate how three critical ecosystem service flows—air purification, temperature regulation, and outdoor recreation—mediate the relationship between greenspace exposure and elderly health. Our findings indicated that: (1) ecosystem service flows exhibited a significant mediating effect (β = 0.041, p < 0.05), surpassing the direct effect of greenspace alone (β = 0.038, p < 0.05); (2) this mediation varies by baseline health status, with the strongest effect observed among healthier individuals and the weakest among those with severe disabilities; and (3) outdoor recreation exhibited the strongest spatial association with elderly health (Moran’s I = -0.04 to 0.18), outperforming temperature regulation (Moran’s I = -0.05 to 0.14) and air purification (Moran’s I = -0.03 to 0.11). These results highlight that enhancing elderly health through urban greening requires not only expanding green infrastructure but also strategically strengthening the delivery of ecosystem service flows most critical to health outcomes.

Mangrove forests possess the capacity to respond to sea level rise, which can be decomposed into adaptability and resilience. Adaptability is primarily measured by the suitable habitat area of mangroves following landward migration, while resilience refers to the ability of mature mangrove forests to maintain their original distribution and functions under sea level rise. However, existing research rarely distinguishes between adaptability and resilience, nor explicitly differentiates the impacts of anthropogenic pressures on these two aspects. This study developed an integrated framework using Sea Level Affecting Marshes Model (SLAMM) to predict mangrove adaptability and resilience under sea level rise scenarios and Geographical Detectors for Assessing Spatial Factors (GeoDetector) to assess their exposure to anthropogenic disturbances. The research focused on Guangdong Province, the largest mangrove area in China, and provided projections for 2070 under the RCP4.5 and RCP8.5 scenarios. The results suggest that under the combined effects of sea level rise and coastal land use, mangrove resilience would decline more markedly than adaptability. By 2070, suitable mangrove habitat is projected to decline to 49.73 %-72.01 % of the current extent, with only 31.26 %-68.67 % of the present mangrove area persisting as resilient mangroves. A significant portion of the lost mangroves would consist of highly diverse, mature mangrove communities. Furthermore, the spatial distribution of terrestrial anthropogenic pressures, primarily from aquaculture ponds and industrial centers, would exert differential impacts on mangrove responses. Aquaculture would mainly affect mangrove adaptability, while industrial development would primarily influence mangrove resilience. By 2070, 28.89 %-40.23 % of the suitable mangrove habitats would be subjected to high levels of anthropogenic pressure, compared to only 0.92 %-2.08 % of the resilient mangroves. The study’s findings suggest that enhancing mangrove adaptability and resilience in response to sea level rise will require differentiated approaches and measures. The proposed framework, which can be adapted to mangrove habitat studies in other regions with appropriate local datasets, provides practical tools for the adaptive management of mangrove ecosystems under global change.

Competition for comprehensive national power nowadays is essentially a talent competition. International student mobility has traditionally favoured Western countries, but emerging economies are gaining prominence, making accurate quantification of national educational attractiveness increasingly important. Existing methods face limitations: physical attribute approaches fail to capture social dynamics, and network methods focus on flow volume while neglecting distance. We propose the Attractiveness-index (A-index), which comprehensively quantifies countries’ talent attraction capacity by integrating flows, distances, and cross-regional patterns. A-index reveals previously obscured transformations in global education, showing a shift from US dominance to a multipolar landscape where China’s attractiveness grew 397 % (1999-2018), uniquely identifies Australia’s exceptional global reach despite modest raw numbers, and recognises emerging regional educational hubs—insights invisible to traditional metrics. We identify the influences on the A-index’s spatio-temporal evolution and reveal strategic competition patterns among structurally similar economies through spatial spillovers. Sensitivity analysis confirms the A-index’s robustness across different parameter configurations. With a higher correlation coefficient to net student inflow than weighted indegree centrality, the A-index provides a superior methodological foundation for understanding educational resource distribution and developing human-centred strategies in the global talent landscape.

Enhancing net ecosystem productivity (NEP) and water yield (WY) services is critical for sustainable ecosystem management and water security. In 2010, China established National Key Ecological Function Zones (NKEFZs) to restore ecosystems. However, their impacts on carbon-water services dynamics remains poorly quantified. Using a calibrated process-based model (CEVSA-ES), we assessed the effects of vegetation restoration (greening and vegetation type changes) and global climate changes (climate change, elevated CO₂, and nitrogen deposition) on the shifts in NEP and WY trends relative to NKEFZ implementation. Over 2001-2021, both NEP and WY exhibited increasing trends (5.1 Tg C yr^-2 and 0.3 mm yr^-1, respectively), and were the most evident in the water and soil conservation zones, biodiversity maintenance zones, and water conservation zones, respectively. Notably, following the NKEFZs establishment, NEP growth accelerated remarkably from 1.9 Tg C yr^-2 (2001-2010) to 5.6 Tg C yr^-2 (2011-2021), particularly within water conservation zones, whereas WY trends reversed from a decline (-0.5 mm yr^-1) to an increase (0.9 mm yr^-1). While greening drove NEP growth and precipitation governed WY changes during 2001-2021, the post-2010 NEP acceleration was jointly controlled by vegetation restoration and global climate change. Conversely, the WY trend reversal was primarily attributed to shifts in precipitation trends. These findings provide critical insights into how ecological policies can synergistically enhance carbon and water services under a changing climate, offering important implications for sustainable ecological restoration and natural climate solutions.

Global Gridded Crop Models (GGCMs) have been widely used to simulate the impacts of global warming on crop production, but their accuracy in capturing the real-world temperature sensitivity of crop yields remains unclear. Here, we evaluated the performance of eight GGCM emulators (incorporating versus not incorporating cultivar adaptation of crop growing periods at 0.5° × 0.5° resolution) in modelling yield sensitivities to 1 K temperature increase (S_T) and optimized their ensembles against statistically-inferred S_T for maize, rice, and wheat using a Bayesian Model Averaging approach. Our results suggest that multi-GGCM ensembles assuming a fixed crop growing period (i.e., a gradually temperature-adapted crop cultivar) show higher goodness-of-fit to statistically-inferred S_T than those assuming a temperature-sensitive growing period for the crops in major food-producing countries. When setting a temperature-adapted growing period instead of a temperature-sensitive growing period in the GGCM ensembles, the R² between GGCM-simulated and statistically-inferred S_T increased from 0.63 to 0.81 for maize, 0.28 to 0.52 for rice, and 0.40 to 0.85 for wheat, meanwhile the RMSE was reduced for all three crops across their respective top 20 producing countries. The crop models may exaggerate historical responses of crop growing periods to climate warming, resulting in an overestimation of yield S_T for maize and an underestimation of yield S_T for rice and wheat in major food-producing countries. The study highlights the importance of adopting dynamic phenological parameters in GGCM simulations to reflect crop cycle adaptation under global warming.

Amid accelerating climate change and intensifying human activities, flood risks have been accumulating, reinforcing the bidirectional interplay between humans and floods. Within this context, human-flood interaction has emerged as a rapidly expanding research frontier that is attracting growing attention. Here, we review 78 representative studies to examine publication trends, research hotspot shifts, and persistent challenges in this field, combining bibliometric analyses with flood disaster records. We found that, over the past three decades, publications on human-flood interaction have increased exponentially, with their focus shifting from one-way impacts towards two-way human-flood coupling. The global collaboration network displays a “dual-core” structure, with transcontinental collaboration between China and the United States and regional collaborations within Europe, whereas participation from Global South countries remains marginal. In addition, a spatial disconnect persists between the geographic distribution of research hotspot regions and that of flood records. Based on bibliometric mapping, we further highlight three enduring challenges: (i) scale mismatches, (ii) spatiotemporal coupling gaps, and (iii) deficits in behavioral modeling. The insights gained from this study will help guide future research and investment in human-flood interaction, with practical implications for enhancing resilience to flood disasters and promoting sustainable human-water coexistence.

Urban parks provide many essential ecosystem services that significantly enhance the well-being of the massive urban population around the world. Nevertheless, there remains a limited understanding of the extent to which the services that park visitors desire align with those they genuinely appreciate. Here, using survey data from over 20,000 respondents across China, we studied the types of ecosystem services that urban residents want and actually enjoy, factors influencing service satisfaction, and the impact of service demand satisfaction on perceptions of urban parks. Findings reveal significant discrepancies between desired and experienced ecosystem services, particularly for food and water supply, and the need for education. Service demand satisfaction correlates positively with people’s environmental interests, income level, park visiting frequency, and urban vegetation coverage. Additionally, a wider range of ecosystem services experienced, even if undesired, enhances users’ perceptions of park environments. Our study highlights the need for urban park designs that address diverse demands, contributing to sustainable urban planning and improving the quality of life in rapidly urbanizing regions.

Mangroves are vital coastal ecosystems that provide crucial ecological functions, but they exhibit pronounced dynamics of both gain and loss over time. Although previous studies have analyzed global drivers of mangrove change, integrated models that distinguish between gains and losses while accounting for regional variability remain limited. Using a global time-series dataset of mangrove distribution from 2000 to 2022, this study characterizes the spatiotemporal patterns of mangrove gain and loss, and employs machine learning models at global and regional scales to identify key drivers. Our results indicated a modest overall increase in global mangrove area over 2000-2022, accompanied by pronounced regional variability. Southeast Asia experienced substantial losses, whereas South Asia, Africa, and Oceania generally showed gains. Regional models demonstrated superior predictive power (R² up to 0.8949) compared to the global model, emphasizing localized driver effects such as coastline accessibility, protected area status, and agricultural suitability. The coexistence of mangrove gain and loss within similar areas highlights complex, non-linear ecosystem dynamics. These findings enhance understanding of mangrove change mechanisms and offer critical insights to inform targeted conservation and climate adaptation strategies worldwide.

Coastal reclamation poses a significant threat to the ecological integrity and sustainability of China’s coastal zone, prompting the implementation of stricter regulatory controls in recent years. However, a comprehensive understanding of long-term reclamation dynamics and the effectiveness of conservation initiatives over the past few decades remains limited. This study first quantified reclaimed areas and analyzed their spatial variability across multiple spatiotemporal scales using remote sensing-based land use and land cover change dataset (1990−2020). We then evaluated long-term trends in reclamation activities in relation to economic growth and policy interventions, which were hypothesized as primary drivers of reclamation dynamics. Our results indicate that land reclamation occurred extensively across all coastal regions, with the largest cumulative reclaimed areas concentrated in Shandong, Liaoning, and Jiangsu provinces. Each of these provinces exceeded 1,800 km², collectively accounting for approximately 11,592.0 km² of reclaimed land from 1990 to 2020. Reclamation expanded rapidly during 2000−2012 (40.9 km²/yr), followed by a pronounced decline post-2012 (−38.0 km²/yr). This transition coincided with the implementation of policy-driven conservation measures, including the establishment of protected areas. Nevertheless, substantial spatial heterogeneity in reclamation patterns persisted, which reflected the influence of local development priorities and the variable effectiveness of regional conservation strategies. These findings suggest that China’s coastal zone is undergoing a transformative shift from land-dependent development toward greener and more sustainable pathways. This study provides robust scientific evidence to support coastal management and offers critical insights for policymakers and stakeholders addressing coastal sustainability challenges at both regional and global scales.

The water‒food‒ecology nexus involves complex interdependencies, but existing studies often overlook multi-sectoral bidirectional feedback and cross-border effects. Using geographically and temporally weighted regression and spatial simultaneous equations, we analyse water‒food‒ecology interactions across 335 Chinese cities (2000-2022). Our results reveal persistent spatial disparities. Cities with high-performance in all sectors remain scarce (<5 %), with high-performance cities of water clustered in Southwest China, food in Northeast China, and ecology in key ecological protection zones. We find the bidirectional promotion between sectors accounted for 16.6 %. The degree of water resource exploration and utilization, serving as a critical leverage point, shows bidirectional promotion with the municipal solid waste harmless treatment rate in ecology sector. Additionally, sectoral conflicts, particularly resource competition between per capita water consumption and ecological water use efficiency, and land competition between agriculture and forests, intensify with urban development. Our further cross-border analysis reveals asymmetric interactions: high-income cities’ water overuse constrains ecological protection in lower-middle-income cities, whereas food production in the latter enhances ecosystems in the former. These dynamics highlight the need for differentiated governance that targets leverage points such as optimization of water resource exploration and utilization as well as cross-regional compensation mechanisms to address negative spillover effects. Our framework advances water‒food‒ecology nexus management by integrating multi-sector feedback and spatial spillovers, providing a template for global sustainability strategies.

Climate change is intensifying extreme precipitation events and flooding, posing unprecedented challenges to agricultural sustainability. Here we evaluate a climate-adaptive agricultural innovation, the aquaculture-planting system implemented in China’s Jianghan Plain, by integrating multi-temporal Landsat imagery, long-term precipitation records, field-collected sediment samples, and socioeconomic statistics to assess its spatial-temporal expansion, flood resilience, ecological health, and economic performance. Over the past 30 years, this region has transformed 4,386 km² of flood-prone farmland into an integrated network of aquaculture zones that function as artificial wetlands, reaching a total area of 5,678 km² by 2023. Analysis of meteorological data reveals that extreme precipitation events now constitute 60 %-70 % of total summer rainfall in the region, up from 30 % in the 1960s. The aquaculture-planting system demonstrates remarkable flood resilience, with capacity to retain approximately 568 million cubic meters of floodwater. Moreover, the system yields significant economic benefits, generating combined net returns that significantly exceed traditional crop farming, with the aquaculture component alone yielding 63,000 CNY ha⁻¹ compared to traditional crop farming’s 25,200 CNY ha⁻¹. Sediment analysis shows that the system maintains ecological health through nutrient recycling while keeping heavy metal concentrations within safe limits. Our findings suggest that the aquaculture-planting system offers a viable model for flood-prone agricultural regions seeking to enhance climate resilience while promoting sustainable development.

Health GeoAI—the integration of artificial intelligence with geographically contextualized health data—offers transformative potential for precision public health. Yet its rapid expansion, often driven by algorithmic performance, risks reinforcing spatial inequities, obscuring decision pathways, and generating environmental externalities. This study introduces a forward-looking framework for Responsible Health GeoAI that embeds geographical equity, accountability, and environmental sustainability as core design imperatives rather than peripheral considerations. Building on advances in foundation models and multimodal learning, the framework establishes two measurable boundaries—an equity floor ensuring subgroup fairness and calibration, and a carbon ceiling constraining computational and energy costs. These operational principles align GeoAI innovation with the broader goals of fairness, transparency, and sustainability. By situating GeoAI as a socio-technical system and integrating spatial validation, participatory governance, and carbon accountability, this study provides a structured pathway for developing GeoAI that is not only intelligent but also equitable, explainable, and environmentally responsible. The framework offers strategic insights for the institutionalization of responsible AI in global health and sustainability policy.

Livelihood diversification is an effective strategy for maintaining household welfare under conditions of vulnerability. However, a comprehensive evaluation of their effects, determinants, and mechanisms on livelihood diversification and household welfare remains lacking in Xizang. This research used datasets on averages, trend, and variability of mean annual temperature, mean annual precipitation, and normalized difference vegetation index (NDVI) to assess vulnerability context. Data on livelihood sub-capitals, disposable income, and daily expenditure were collected using semi-structured questionnaires, with 796 small-scale households randomly selected from agricultural (215), agro-pastoral (278), and pastoral (303) zones. Zonal cross-variances in livelihood capital, income diversification, and household welfare were examined using one-way analysis of variance. Meanwhile, their interrelationships and influential pathways were further explored using bivariate regression, Pearson correlation with Mantel test, and piecewise structural equation modelling. Our results indicated that income diversification is highest in the pastoral (0.51), followed by the agro-pastoral (0.44) and agricultural zones (0.29, p < 0.001). Household welfare peaks in the agro-pastoral zone (0.23), followed by the agricultural (0.17) and pastoral (0.13) zones (p < 0.001). Income diversification positively influences welfare levels among agro-pastoral households across all zones. Mean annual temperature directly influences household welfare in agro-pastoral and pastoral zones. Meanwhile, Mean annual precipitation, NDVI, and mean annual precipitation trendindirectly affect income diversification via shifting human and physical capitals, with these indirect pathways varying across zones. Therefore, it suggests that policymakers should assess bioclimatic conditions and capital allocations to tailor interventions to meet region-specific needs, and that focusing on factors that motivate households to diversify their livelihoods is essential for developing targeted interventions.