This study examines network externalities in China’s new energy vehicle industry from the perspective of knowledge division of labor. Using supply chain transaction data from 2010 to 2024, it analyzes knowledge division linkages between focal firms and supplier clusters across intra-city and inter-city scales. Intra-city network externalities are measured through value agglomeration indices, while inter-city network externalities are assessed using interlocking network analysis. The results show that: 1) intra-city externalities diffuse from national and regional centers to specialized manufacturing hubs, following hierarchical and contagious patterns with time-lagged polarization-trickle-down effects. 2) The role of localization economies gradually weakens, whereas urbanization economies first decline and then rebound, with intra-city externalities increasingly serving external markets. 3) Inter-city externalities are structured by hub cities, characterized by leapfrog cross-regional diffusion and stronger economic homophily. 4) Borrowed size effects peak and then decline, while borrowed function effects continue to strengthen. 5) Core nodes in network communities remain stable, and virtual and spatial agglomeration promote cross-regional integration as linkage motivations shift toward cognitive proximity. This study contributes a dual-scale framework for understanding urban network externalities and provides insights for improving knowledge coordination across heterogeneous cities.

In the era of the knowledge economy, cities serve as core hubs for innovation activities. The ability to effectively convert local knowledge creation into economic value and continuously introduce new technologies to diversify and upgrade its technological portfolios has become pivotal for measuring urban innovation vitality, developmental potential and urban competitiveness. This study employs panel data from 287 Chinese prefecture-level cities (excluding data of Hong Kong, Macao and Taiwan) between 2005 and 2016 to construct a multivariate regression model, examining the mechanisms through which urban capacity to convert local technological knowledge into products influences the evolutionary paths of new technology entry. The findings reveal that a city’s capacity to successfully transform local patent knowledge into marketable products acts as a critical driver for new technology entry. However, this positive impact is not uniformly distributed: the role of local technology conversion capacity in propelling cities toward leapfrogging developments into unrelated technological domains, thereby breaking path dependencies, is significantly stronger than its effect on extending into related technological fields. This driving effect exhibits heterogeneity; in industries with strong equipment-specific processes and significant differences in the technical paradigms across sub-fields, the positive role of local conversion capacity in introducing new technological varieties may be weakened or even reversed. Furthermore, interactions exist between local technology conversion capacity and external collaborative networks or internal technological relatedness. The marginal contribution of local technological conversion capacity to fostering new technology entry diminishes due to cognitive inertia or lock-in effects. These findings provide empirical support for policymaking aimed at guiding cities toward higher-level and diversified technological pathways, as well as cultivating and strengthening endogenous urban capacities for technological transformation. Specifically, policies should prioritize establishing concept validation centers, technology transfer institutions, and market-oriented trading platforms to bridge the gap between patent generation and product commercialization. Differentiated strategies are needed across industries, for sectors with high equipment specificity and divergent technical paradigms, emphasis should be placed on shared technology platforms and cross-domain knowledge exchange. While encouraging open innovation, cities must balance external technology cooperation with local capacity building, and resource-rich cities should allocate dedicated resources for exploring non-consensus technologies to prevent lock-in risks.

With the extensive penetration and integrated application of digital science and technology in the process of social and economic development, it has rapidly become a core element for the sustained growth of regional economies and the driving force as well as the key factor in the competition for resources in the future. Based on the big data from business registration in Chinese enterprises and the social and economic data of Chinese cities, this paper first uses the equity penetration method to obtain 165038 mutual investment records of digital technology enterprises in 327 cities in China from 2000 to 2021 (data excluding Hong Kong, Macao and Taiwan), and then constructs a theoretical analysis framework for the evolution of the digital technology connection network space based on the flow space theory, urban network theory, and enterprise investment behavior. Secondly, by combining complex network analysis methods such as weighted centrality, dominant flow, spatial structure index, and quadratic assignment procedure (QAP), it deeply explores the spatial structure evolution characteristics and influencing factors of China’s digital technology connection network from the three levels of node-path-pattern. The findings are as follows: 1) From the evolution characteristics of nodes, during the research period, the digital technology investment influence and control power of cities such as Beijing, Shanghai, Shenzhen and Guangzhou have been gradually increasing, and they have occupied a dominant position and played a dominant role in China’s digital technology connection network, while the capital cities in the central and western regions and a few emerging cities have the main growth engine. On the whole, China’s digital technology connection network shows a significant high-high and low-high clustering feature in space, and the digital technology investment of core cities has the ‘siphon effect’ on peripheral cities. 2) From the evolution characteristics of paths, during the research period, the evolution of China’s digital technology investment paths has moved from ‘point breakthrough’ to ‘local equilibrium’, and there is a ‘leader’ effect in regions such as the Beijing-Tianjin-Hebei region, the Yangtze River Delta, and the Pearl River Delta, and digital technology investment also shows obvious network effects and scale effects. 3) From the evolution characteristics of patterns, during the research period, China’s digital technology connection network has shown a polarized development trend, among which the local investment pattern of digital technology has gradually evolved from scattered point distribution to a trend of multiple points in local areas. Also it has a significant ‘Hu Huanyong Line’ spatial distribution feature, while the scale of digital technology investment has undergone a process of first strengthening and then weakening, and is mainly dominated by economically developed cities or a few core cities globally, and shows a diffusion and local concentration trend. 4) From the influencing factors, the effects of each influencing factor on the spatial evolution of China’s digital technology connection network have obvious phased characteristics. For example, economic development level remains an important factor affecting the form of China’s digital technology connection network, while the effects of factors such as technological innovation, informatization level, industrial structure, and financial development level vary over time.

Based on data on innovation entities and patent applications in the new energy vehicle (NEV) industry in the Pearl River Delta from 2008 to 2022, this study employs kernel density analysis, the standard deviation ellipse, and the GeoDetector model to examine the evolution of the innovation spatial pattern and its influencing factors. The main findings are as follows: 1) Enterprises are the dominant innovation entities, while universities and research institutes play supporting roles. Spatially, innovation activities are concentrated in Guangzhou and Shenzhen and gradually diffuse to surrounding areas, showing an evolutionary trend of “large-scale diffusion and small-scale agglomeration”. 2) Patent innovation exhibits significant spatial agglomeration and forms a “multi-core and belt-like clustered” spatial pattern. Innovation levels differ across industrial chain segments. The midstream segment accounts for the largest share of innovation outputs. Patent distribution in different segments presents diverse characteristics, including single-core and dual-core agglomeration. Relying on industrial parks, innovation clusters are mainly concentrated along the eastern coast of the Pearl River Delta. 3) Innovation linkages are centered on Guangzhou and Shenzhen as the primary radiation hubs, while Foshan, Dongguan, and Zhuhai act as the secondary hubs. Together, they form a spatial structure characterized as “two cores, three centers, and multiple nodes”. 4) Economic development level, government support, transportation accessibility, scientific research investment and capital support all have positive effects on NEV industrial innovation, whereas industrialization level shows a negative correlation. Among these factors, capital support (0.8991), scientific research investment (0.8057), economic development level (0.6800), and government support (0.5738) are the dominant driving factors. The interaction effects are mainly characterized by dual-factor enhancement. The new energy vehicle industry is a key vehicle for developing new quality productive forces. This study reveals innovation spatial clustering, connectivity characteristics, and determinants in the Pearl River Delta new energy vehicle industry, providing reference for optimizing innovation resource allocation, strengthening industrial chain collaboration, and promoting high-quality development.

As a strategic emerging industry, the new energy vehicle (NEV) sector exhibits complex spatial interdependencies between production and innovation. However, existing research has paid limited attention to this issue. Using Changzhou, a key NEV manufacturing hub in China, as a case study, this paper combines co-location quotient analysis with field interviews to examine the spatial distribution, co-location synergy, and evolutionary pathways of NEV production and innovation. The findings reveal significant heterogeneity in production-innovation co-location synergy across different segments of the NEV industrial chain. Core component segments (e.g., batteries) exhibit a high degree of production-innovation co-location, whereas complete vehicle manufacturing and auxiliary component segments are characterized by production-dominated agglomeration, reflecting structural and functional differentiation within the industrial chain. Accordingly, 3 localized evolutionary patterns are identified in Changzhou: a low-value-added production cluster dominated by component manufacturing; a production-chain cluster led by complete vehicle manufacturing; and a production-innovation synergy cluster driven by battery segments. Notably, the central urban area and Liyang–Jintan exhibit contrasting levels of production-innovation synergy. The central urban area centers on complete vehicle manufacturing and fosters tacit coordination through inter-firm learning and process improvements. Liyang–Jintan leverages the local embeddedness of leading battery enterprises and the development of innovation platforms to transition from externally attracted production to endogenous innovation. This has fostered an enterprise-led system integrating industrial production, technological innovation and supply-chain collaboration. The study demonstrates that production-innovation synergy depends not only on geographic proximity but also on technological characteristics of dominant segments, enterprise organizational structures, and local institutional support. Finally, the paper proposes policy recommendations to promote coordinated development of local production and innovation system.

Based on inter-provincial migration data for high-skilled talents and ordinary labor in China from 2000 to 2020, this study employs the network Moran’s I and complex network analysis to explore the temporal and spatial evolution and network characteristics of different types of labor migration. The findings reveal that: 1) Over the past two decades, economically developed coastal provinces have consistently been the primary destinations for both types of labor. Despite the pronounced spatial imbalance in their distribution, this trend has gradually diminished since 2010. 2) Both types of labor migration networks exhibit significant spatial agglomeration effects. The migration network for high-skilled talent showed an initial increase in agglomeration, followed by a decline. In contrast, the ordinary labor migration network experienced fluctuating trends. Its agglomeration first decreased, then increased, and ultimately decreased again. 3) Both migration networks exhibit “small-world” characteristics; however, the “small-world” nature of ordinary labor migration has become more pronounced over time, indicating tighter connections in migration flows. 4) Both types of labor migration are influenced by geographical proximity and path dependency, yet high-skilled talent migration is less constrained by geographical distance, allowing for long-distance migration capabilities that enable rapid responses to economic opportunities. This paper enriches the understanding of heterogeneous skilled labor migration networks within Chinese population geography. It also expands the temporal dimension of existing research. Furthermore, these findings offer robust support for decision-making. They can help policymakers optimize regional talent policies and efficiently allocate labor resources.

Enterprise relocation constitutes a pivotal mechanism for spatial restructuring of manufacturing industries. Systematic evidence on aggregate patterns and interactive drivers, however, remains scarce. Drawing on a newly proposed Migration Force Model, this study maps the spatiotemporal evolution of inter-regional relocations among Chinese manufacturing firms from 2012 to 2020 and empirically tests the underlying mechanisms. Using panel data covering 1359 migrating enterprises, the model quantifies migration force through the smoothed net difference in per-capita relocation scale between origin and destination districts. The key findings are as follows. Temporally, 2012—2017 is the peak relocation period, dominated by technology-intensive firms, whereas traditional sectors mainly completed cross-regional moves before 2018. Spatially, relocation exhibits an “east-out, west-in” pattern—departure points were concentrated while destinations were dispersed. Bidirectional flows between adjacent and conditionally similar regions were active, stabilising net migration force peaks in central districts. This study examines how production factors couple with firm demand. Labour demand strengthens the negative migration force that stems from external labour supply, while technology demand reinforces the positive force driven by specialisation intensity. Raw material cost advantages generated positive forces only when did firms exhibite responsive material demand orientations. Concerning synergy between firm capabilities and market condition, marketisation exerted positive migration force, whereas government intervention operated inversely. Industrial agglomeration and firm capability independently induced negative force, yet high-capability firms could reverse agglomeration’s negative effect. Market openness and corporate social network accessibility generated positive and negative forces respectively, with social networks amplifying the positive effect of openness. In terms of interaction between production factors and market condition, both housing prices and market size exert positive forces in isolation. Yet rising housing prices attenuate the attractiveness of large markets and eventually reverse it. Manufacturing firms’ relocation motives display heterogeneity across time, space and firm characteristics, collectively driving the spatiotemporal evolution of inter-regional migration. These findings advance an integrated analytical framework for understanding firm relocation decisions and offer empirical foundations for regional industrial policies aimed at balancing agglomeration economies with spatial equity.

In the digital era, the upgrading of urban industrial structure is not only an important symbol of high-quality development of regional economy, but also a key engine for the remodeling of national competitiveness and the construction of modern industrial system. Based on the cross-perspective of digital technology and niche theory, this paper analyzes two dimensions: “state” (static endowment of digital technology resources, including infrastructure, technology accumulation, talent reserve and institutional environment) and “potential” (dynamic evolution ability of digital technology system, including enterprise growth, cross-domain collaboration, open innovation and network effect). To construct and systematically define the conceptual framework and measurement system of “digital technology niche”. Using the panel data of 283 prefecture-level cities in China from 2009 to 2022, the two-way fixed effect model is used to empirically investigate the impact effect, transmission mechanism and regulatory conditions of digital technology niche on the upgrading of urban industrial structure. The results show that: 1) The digital technology niche significantly promotes the optimization and upgrading of local industrial structure, and this conclusion is still valid under a series of robustness tests such as changing the explained variable, controlling macro conditions and instrumental variable method. 2) The analysis of structure effect shows that the promotion effect of digital technology niche on the optimization of industrial structure is significantly stronger than that on the rationalization of industrial structure, reflecting its core advantage in promoting the industry to climb to the high end of the value chain. 3) The mechanism test shows that the optimal allocation of innovation factors and the improvement of entrepreneurial activity are important intermediary paths for the digital technology niche to promote the upgrading of industrial structure, in which the mediating effect of innovation factor allocation accounts for 40.2% of the total effect, and the mediating effect of entrepreneurial activity accounts for 28.8%. 4) The analysis of moderating effect further reveals that the improvement of digital infrastructure and the agglomeration of digital talents positively regulate the promoting effect of digital technology niche on the upgrading of industrial structure, that is, the higher the level of urban digitalization is, the more significant the industrial upgrading dividend of digital technology niche is. 5) Heterogeneity analysis shows that core cities can benefit more from digital technology niche than peripheral cities, reflecting that there are significant city-level differences in the impact of digital technology niche on industrial structure upgrading.

Under the background of innovation-driven development strategy, low-altitude economy industry has gradually become the new focus of global technology competition and an important engine for cultivating new quality productivity. This study takes the prefecture-level cities in the Yangtze River Delta from 2010 to 2021 as the research units, integrates low- altitude economy industrial chain and economic and social attribute data, and combines ArcGIS spatial analysis and spatial econometric model to explore the spatial agglomeration situation of low-altitude economy industry and its innovation effect. The results show that: 1) The spatial agglomeration scale and scope of the low-altitude economy industry in the Yangtze River Delta have continued to expand, forming a multi-center spatial agglomeration pattern centered on cities such as Shanghai, Hangzhou, Nanjing and Hefei, as well as a belt-shaped agglomeration structure in the Suzhou-Wuxi-Changzhou region; 2) Low-altitude economy industry has strong spatial unbalanced agglomeration characteristics, but the agglomeration intensity of core cities has weakened, indicating that the spatial layout of low altitude economy industry in the Yangtze River Delta has evolved in a diversified and balanced direction; 3) From the perspective of spatial spillover effect, every 1% increase in low-altitude economy industrial agglomeration will increase the innovation level of local cities by at least 0.024 4%; However, it is difficult to promote regional technical cooperation and collaborative innovation in the early stage of low-altitude economy industry development; on the contrary, it has a strong negative innovation spillover effect on neighboring cities because of the siphon effect of resources and technical barriers. From the perspective of distance spillover effect, low-altitude economy industry has certain distance attenuation characteristics on urban innovation ability, and it has strong “agglomeration shadow” and time lag effect; 4) Low-altitude economy industry has a certain degree of technology embeddedness and industrial integration, which can not only significantly enhance the technological innovation of the core cities in the Yangtze River Delta, but also provide support for the innovation and development of weak industries, which provides an effective way to narrow the regional technological development gap.

The Guangdong-Hong Kong-Macao Greater Bay Area (GBA) serves as a world-class urban agglomeration and a frontier for the national innovation-driven development strategy, and its innovation network plays a critical supporting role in the high-quality development of the regional economy. The present study, based on invention patent transfer data and socioeconomic statistical data from 2005 to 2020, and using the 61 districts of the GBA as research units, applys social network analysis, entropy-weighted TOPSIS, and geo-detector techniques to analyzing the spatiotemporal evolution characteristics of the GBA’s innovation network from 3 aspects—node characteristics, linkage characteristics, and agglomeration patterns. This study also explores the influencing factors of innovation network in the GBA evolution from four dimensions: economic development, market maturation, science-education resources, and institutional differences. The findings are as follows: 1) The heterogeneity of innovation network node capabilities is significant, with core districts such as Nanshan in Shenzhen and Yuexiu in Guangzhou playing key roles in innovation agglomeration and diffusion, while peripheral districts like Heshan in Jiangmen and Fengkai in Zhaoqing lag behind in development; 2) The intensity of innovation links shows an increasing trend, and the gravity center of the regional innovation network has shifted from a single-center to a multi-center structure, especially in the bay area urban belt, which exhibits close innovation cooperation relationships; 3) The spatial agglomeration effect of innovation interaction capabilities is significant, with high-high agglomeration areas serving as the centers of the innovation network, and the spatial center of gravity shifting from Hong Kong-Shenzhen to Guangzhou-Foshan, while low-low agglomeration areas are mainly and persistently distributed in the peripheral zones; 4) The formation and development of the innovation network are driven by multiple factors including the economy, market, scientific and educational resources, and institutional settings, with technological innovation capacity and economic output level having prominent impacts on regional innovation interaction capabilities, and also act as key factors in multi-factor synergistic effects. In addition, the administrative center effect is also an important factor affecting regional innovation interaction capability, albeit with a relatively low correlation. This study provides a theoretical foundation and empirical support for the formulation and implementation of regional innovation policies, and has positive implications for promoting the sustainable development of the GBA’s innovation network and accelerating the building of an international science and technology innovation center.

Liaoning Province, as a paradigmatic region with the highest concentration of shrinking cities in China, enhancing the development of resilience-related networks is essential for improving urban risk management capabilities. This study utilized the TOPSIS entropy weight method, an adjusted gravity model, social network analysis, and disruption simulation to investigate the correlation strength and structural characteristics of urban resilient spatial networks in Liaoning Province from 2012 to 2022. We simulated changes in network structure under various attack scenarios and subsequently identified key influencing factors using the QAP analysis method. The findings are summarized as follows: 1) Spatial correlation network of urban resilience in Liaoning Province exhibits a distinct “core-periphery” structure, with overall correlation intensity remaining relatively low. Over time, the network has transitioned from a “dual-core independent leadership” pattern to one of “dual-core coordinated development”. At the subsystem level, networks related to economic, engineering, and innovation resilience demonstrate steady improvement, whereas the social resilience network undergoes a significant decline, adversely affecting overall network stability. 2) The overall network structure has steadily strengthened, with core cities exerting significant influence and maintaining substantial control over resources. Meanwhile, peripheral regions have leveraged their geographical advantages to generate notable inter-provincial spillover effects. 3) Under random attack scenarios, the resilient interconnected network displays marked structural vulnerability, evidenced by rapid declines in key indicators such as network efficiency and clustering coefficient. In contrast, the network demonstrates greater structural robustness under targeted attacks, with these indicators declining at a more gradual pace. 4) Economic development, population size, industrial structure, and infrastructure investment exhibit positive correlations with resilient connectivity, whereas geographical proximity and ecological environment quality show negative associations. Specifically, spatial adjacency intensifies competition for resources and environmental stress, while uneven distribution of ecological assets further constrains regional resilience enhancement.

Due to the complex terrain in the areas traversed by oil and gas pipelines, geological disasters such as landslides inevitably pose a threat to the stable operation of the pipelines. Meanwhile, a single landslide susceptibility evaluation method has limitations. In order to avoid the errors brought by a single model in the landslide susceptibility evaluation and comprehensively consider various evaluation characteristics, this paper adopts the DBN (Deep Belief Networks) integrated model to comprehensively evaluate the landslide susceptibility in the study area. Aiming at the problem that the landslide susceptibility evaluation results obtained by using static factors cannot be updated for a long time and fail to reflect the actual situation of the target area, this paper combines the time-series InSAR results with the susceptibility evaluation mapping by using a correction matrix to improve the reliability, timeliness and accuracy of the landslide susceptibility evaluation, and provide technical support for the safe operation of oil and gas pipelines in the study area. The results show that the ROC curve area of the DBN integrated model reached 0.94, which can effectively combine the evaluation results of multiple models. The proportions of areas with landslide sensitivity are as follows: the low-sensitivity area accounts for 33.25%, the medium-sensitivity area accounts for 23.58%, the high-sensitivity area accounts for 22.95%, and the extremely high-sensitivity area accounts for 20.2%. Among them, the high and extremely high sensitivity areas are mainly concentrated in the northeastern and eastern parts of the study area, in the loess hilly regions. The low-sensitivity areas are mainly located in the central, eastern and southern plain regions of the study area, and the vast majority of the InSAR monitoring hazard points are distributed in the high and extremely high sensitivity areas. The InSAR correction matrix has correc-ted 1% of the evaluation units in the study area, which can effectively update the landslide sus-ceptibility evaluation results to a certain extent and improve the evaluation accuracy.

River deltas sustain dense global populations, major economic centers, and important ecosystems. However, most large river deltas in the world are threatened by land subsidence, with subsidence rates far exceeding the rate of sea-level rise. The Yellow River is the fifth largest river in the world and the second largest in China. At its estuary, the Yellow River Delta hosts the most intact warm temperate coastal wetland ecosystem in China and the under construction Yellow River Estuary National Park, endowing the region with outstanding ecological importance and national strategic value. Nevertheless, its low lying, broad, silt and mud dominated terrain is highly susceptible to land subsidence, which, when combined with sea level rise, extreme precipitation, and storm surges, poses a risk of seawater inundation to the Yellow River Delta. Existing studies on the causes of subsidence have mostly focused on tectonic movements, sediment consolidation and compaction, oil and gas resource exploitation, underground brine extraction, and urbanization. The impact of seasonal variations in shallow groundwater levels on ground deformation has rarely been quantified. Taking the Yellow River Delta as an example, this study evaluates the performance of the SBAS InSAR method using the GMTSAR and SARscape software packages to invert land subsidence from 91 Sentinel 1 images acquired between 2016 and 2024. Based on water level monitoring data from nine groundwater observation wells, the relationship between groundwater level changes and land subsidence is analyzed to reveal the contribution of groundwater level variations to surface deformation. The results show that: 1) The land subsidence results inverted by the SBAS InSAR method based on GMTSAR have an error of 8 mm/a, while those based on ENVI’s SARScape method have an error of 10 mm/a, representing an accuracy improvement of 2 mm for the former over the latter. The high subsidence rate zone is located in the coastal salt pan area where underground brine is extracted, with extreme values exceeding 360 mm/a, which is consistent with existing studies. 2) In most delta areas (located in the relatively landward region where land formation occurred earlier), groundwater level and surface elevation exhibit a significant positive correlation (correlation coefficient > 0.5, generally exceeding the 99% confidence interval), likely due to seasonal precipitation recharging groundwater, leading to a rise in both groundwater level and ground elevation. 3) In contrast, in the rapid subsidence area of coastal salt pans where underground brine is extracted, groundwater level and surface elevation show a significant negative correlation (correlation coefficient>0.5, generally exceeding the 99% confidence interval). This may be attributed to the continuous subsidence of the coastal salt pans, where underground brine is extracted for salt production, while the coastal groundwater level has been rising over the past several years. The findings of this study can provide a scientific basis for land subsidence disaster prevention and ecological protection in the Yellow River Delta.

Under the influence of global climate change and urban development, the frequency and intensity of precipitation events have been increasing, creating significant flood risks for urban areas. As important components of urban ecosystems, green spaces play crucial roles in flood regulation. However, existing research has primarily focused on linear relationships, with limited analysis of nonlinear impacts and threshold effects. Notably, in current planning practice, green space management predominantly adopts “one-size-fits-all” approaches with general recommendations to increase or decrease green space coverage, without specifying optimal amounts or configurations. The lack of threshold-based management models creates gaps between planning objectives and implementation outcomes, limiting support for precise planning and engineering practices. Therefore, this study systematically analyzes the nonlinear relationships between urban green space landscape patterns and flood regulation services, investigates the mechanisms through which green space patterns influence flood regulation services, and provides scientific evidence for sponge city construction and urban flood resilience enhancement. This study uses 110 water management units in Hangzhou as study samples and quantifies regional flood regulation services based on the SCS-CN hydrological model. A comprehensive indicator system is developed to characterize the scale, morphology, and structure of green space landscape patterns. Random forest models are integrated with Partial Dependence Plot (PDP) algorithms to analyze key influencing factors, nonlinear relationships, and threshold effects of green space landscape patterns on flood regulation services. The results of the study showed that: 1) Flood regulation services in Hangzhou show clear spatial variation, with lower values in the northeast and higher values in the southwest. Lower-performing units are concentrated in Shangcheng District, Gongshu District, and Binjiang District; medium-performing units are distributed across various districts without clear clustering; higher-performing units are concentrated in northern Yuhang District and southern Xiaoshan District. 2) The influence of different green space pattern dimensions on flood regulation follows the order: structural characteristics (28.00%) > scale characteristics (21.27%) > shape characteristics (7.34%). This indicates that optimizing green space structure is the most effective approach for enhancing flood regulation services. 3) Key green space landscape pattern factors show significant nonlinear effects on flood regulation services. Optimal flood regulation is associated with larger patch sizes, more complex shapes, and lower fragmentation levels. 4) Effective flood regulation enhancement requires maintaining green space landscape pattern factors within specific thresholds. Optimal performance is achieved when the largest patch index (LPI) exceeds 46.50%, the mean patch area (AREA_MN) exceeds 28.85 hm², the mean shape index (SHAPE_MN) exceeds 1.39, the edge density (ED) exceeds 133.52 m/hm², fragmentation (DIVISION) is below 0.66, and cohesion (COHESION) exceeds 98.36. These findings provide quantitative guidelines for urban green space planning and design. The identified thresholds offer specific targets for planners and designers to optimize green infrastructure performance for stormwater management.

In the context of new quality productivity, social media drives the reconfiguration of production factors, thereby triggers rural industrial transformation. The process, path, and interaction mechanism of “human-social media-place” in this industrial transformation urgently requires in-depth exploration. This study takes 3 internet-celebrity villages with different development levels in the desakota areas of Guangdong province (Nazhou Village, Huitong Village, and Yakou Village) as examples. From the perspective of new materialism, the study aims to analyze the process and path through which new quality productivity, represented by social media, facilitates industrial transformation in these villages. The findings reveal that: 1) The industrial transformation of internet-celebrity villages is the result of co-creation by human and non-human actors. 2) Social media influences the industrial transformation of internet-celebrity villages through two paths. First, as a social element, it empowers and enables both people and places, making them visible, discussed, and constructed. Second, as a social shaping force, it constructs virtual spaces while promoting the formation of a virtuous cycle and the conversion of virtual-real traffic between virtual and real spaces. 3) Compared with traditional rural industrial transformation, the key aspect of social media’s participation in the industrial transformation of internet-celebrity villages lies in promoting the deep integration of data with traditional production factors, facilitating the leap from traditional productivity to new quality productivity, and thus driving industrial transformation. This study responds to new materialism’s emphasis on non-human actors, validates the view that power is multi-symbiotic, and analyzes the logic of new quality productivity driving the industrial transformation of internet-celebrity villages, which holds theoretical and practical significance for rural sustainable development.

Uninhabited islands represent a strategic category of marine spatial resources characterized by high ecological sensitivity, institutional complexity, and development uncertainty. Existing studies on uninhabited island development have largely focused on policy frameworks and functional zoning, while paying limited attention to the dynamic interactions among multiple actors involved in the development process. Drawing on Actor-Network Theory (ANT), this study proposes an analytical framework centered on core actors, obligatory passage points (OPP), translation processes, non-human actors, and network stability. The framework emphasizes generalized symmetry, treating geographical conditions, institutional texts, and engineering facilities as active participants rather than passive backgrounds. Through a comparative analysis of 3 uninhabited islands in Guangdong Province—Fangji Island, Sanjiao Island, and Niutou Island—the paper examines how different development logics shape the configuration and stability of actor-networks across stages of problematization, interessement, enrollment, and mobilization. The findings indicate that uninhabited island development is not a linear process, but a relational outcome co-produced by human and non-human actors. Market-driven development is prone to instability due to non-human resistance, where aging infrastructure and shifting legal statuses of lease contracts betray the commercial logic. Government-led public welfare-tourism development exhibits non-human tightening, as the single island certificate system and high ecological restoration costs constrain translation processes, leading to long-term structural tension. In contrast, state-strategic, function-oriented development achieves high short-term stability through non-human disciplining, where rigid engineering blueprints and schedules compress human decision-making spaces. The study argues that network stability should be understood as dynamic order, a contextual and temporary effect of translation, rather than a normative ideal. By highlighting the differentiated roles of non-human actors such as typhoons, property rights, and topographical legacies, this research contributes to a deeper understanding of island governance. It offers analytical insights for the classification-based development of uninhabited islands, advocating for the construction of resilience-oriented networks capable of accommodating environmental and institutional perturbations.

Shell midden sites in the Yongjiang River Basin of Guangxi represent a typical example of Neolithic riverine terrace settlements in southern China. They provide crucial evidence for studying human physical characteristics, settlement patterns, and survival adaptations during the early to middle Holocene [(10—6) ka B.P.]. However, systematic quantitative analyses of their spatial distribution and relationship with the natural environment is still lacking, which has limited a deeper understanding of prehistoric landscape use strategies. Based on detailed archaeological data from 18 shell midden sites, combined with GIS spatial analysis and Digital Elevation Models, our study investigates quantitative patterns in site location and their reflected environmental adaptation strategies in the Yongjiang River Basin, Guangxi. Approximately 66.7% of the sites are distributed on northern bank terraces, 68.8% are oriented toward north- or east-facing slopes, and the majority are situated at elevations between 70 and 82 meters, on slopes ranging from 0° to 5°, and within one kilometer of river channels. These findings reveal a marked preference for specific topographic, geomorphic, and hydrological settings. Further analysis reveals that these communities developed an optimized locational strategy adapted to the south Asian subtropical monsoon climate by integrating multiple factors such as flood prevention (elevation control), resource accessibility (proximity to water), and micro-environmental comfort (preferring relatively cooler north-eastern slopes). Statistical analysis confirms that approximately 61% of the sites conform to this integrated strategy, indicating that prehistoric populations may have possessed the capacity for comprehensive decision-making in evaluating environmental suitability and systematically utilizing landscape resources. This study provides a new empirical case and methodological reference for deepening the understanding of human-environment relationships in south China during the early to middle Holocene.

This article first analyzes the spatial mobility characteristics of rural left-behind elderly from three dimensions: mobility distance, mobility frequency, and mobility modes. Then from the perspective of mobility, we explore how movement reshapes the agency of rural left-behind elderly in their later years. Research findings indicate that rural left-behind elderly have developed a mobility pattern characterized by short-distance daily travel and long-distance high-frequency travel. Their high-frequency mobilities primarily serve basic physiological needs and intergenerational support activities, while low-frequency mobilities is associated with specific kinship enclaves and festive events. Transportation choices for left-behind elderly exhibit diverse characteristics, with no apparent exclusion from transportation services. Mobility has become a crucial factor in reconstructing the agency of rural left-behind elderly individuals. In terms of embodiment subjectivity, the technological embeddedness within mobility enhances the physical capabilities of left-behind elderly individuals, disrupts gendered spatial orders, and through embodied practices of mobilities, revolutionizes subject cognition while fostering new bodily strategies. The relational agency emerging from mobility enables left-behind elderly individuals to reconnect their social networks and actively construct local meaning and emotional bonds. The breakthrough of inherent labels in mobility, the production of new roles, and the performance of new identities embody the agency of identity creation among left-behind elderly. This study addresses the current academic gap in research on grassroots mobility in rural areas, revealing the unique and multifaceted significance of mobility for the construction of agency among marginalized populations.

Coordinated development of economic growth quality improvement and ecological environment protection is an important component of achieving regional sustainable development. The development of low-carbon industries has become an urgent need to achieve synergy between economic quality improvement and ecological environment protection. Under-forestry-economy has the dual strategic significance of economic transformation and ecological environment protection. Improving the under-forestry-economy growth quality is conducive to the transformation of regional economy to green and low-carbon. This paper analyzes the coupling coordination mechanism of under-forestry-economy growth quality and carbon emission in the Changbai Mountain area, constructs the evaluation index system of under-forestry-economy growth quality and carbon emission evaluation index system, and uses entropy weight method, coupling coordination degree model, spatial auto-regression model and dynamic Qualitative comparative analysis (QCA) method to explore the spatial and temporal evolution characteristics and influencing factors of the coupling and coordination degree of under-forestry-economy growth quality and carbon emission in 9 counties (cities) in the Changbai Mountain area from 2012 to 2022. The conclusions are as follows: The coupling coordination degree of under-forestry-economy growth quality and carbon emission in the Changbai Mountain area from 2012 to 2022 has experienced high-speed coordination stage, transformation and adjustment stage and high-quality coordinated development stage, and is currently in the stage of high-quality coordinated development. The coupling coordination degree shows a trend of ‘rise-decline-recovery’ in time series, and presents a pattern of ‘local high-value discrete distribution’ in space, and the spatial correlation is not strong. Labor data is the main factor affecting the coupling and coordination degree of under-forestry-economy growth quality and carbon emissions in the Changbai Mountain area, and forest tending area is the core condition affecting the coupling and coordination degree. The intra-group coverage of each configuration has regional differences. According to the orientation of influencing factors, the research area is divided into 2 categories: laborer-labor data-oriented and labor data-oriented, and corresponding development suggestions are proposed for the research results.

Inland aquaculture, due to its strong regional characteristics and weak adaptability, is confronted with a series of problems such as fragmented scale, ecological overloading, unclear production capacity, indistinct advantages, and spatial imbalance. In view of this, taking the Poyang Lake Basin as an example, this article explores the type structure, spatial pattern, location advantages and spatial adaptability of inland aquaculture based on the data of the first national census of aquaculture germplasm resources and methods such as spatial autocorrelation and spatial lag analysis, so as to clarify the resource inventory and propose a spatial optimization plan for industrial upgrading. The results show that: 1) The aquaculture in the Poyang Lake Basin is dominated by major freshwater fish, shrimp and crab species, with shellfish, amphibious reptiles, and other characteristic economic species as supplementary components. The spatial pattern features a main agglomeration center in the Poyang Lake core area and a secondary hotspot in northeastern Jinanxi. 2) The comprehensive advantageous areas for aquaculture are distributed in a concentric pattern around Poyang Lake and are also scattered along the tributaries. 3) There is a certain spatial mismatch between the aquaculture pattern and the degree of locational advantage. The center of gravity of aquaculture is 70.54 km away from the center of gravity of the comprehensive locational advantages. In the negatively high-mismatch areas (e.g., Dexing, Nanchang), high-density aquaculture faces the risk of resource and environmental overloading and has not fully leveraged the market and technological advantages of the large metropolitan areas. In the positively high-mismatch areas (e.g., Duchang, Gan County), the characteristic resources of the mountainous areas and the aquaculture production capacity await further development. In the future, aquaculture in Poyang Lake should be focused on ecological scale, factory intensive and regional characteristic transformation around the ecological fishery sector in northern Jiangxi, intensive fishery sector in northeast Jiangxi, and mountain stream characteristic fishery sector in southern Jiangxi.