Environmental and Earth Sciences

Abstract: High-density cities increasingly treat parks, greenways, waterways, streetscapes, and urban landscapes as a connected green infrastructure (UGI) system supporting climate adaptation, biodiversity, health, and recreation. Yet these functions depend on how land-use planning, project delivery, and long-term management are connected. This study uses a qualitative, mechanism-oriented comparison to examine how Singapore governs networked urban landscapes and how these functions relate to Shanghai's policy transition. A 2025 professional study visit provided contextual evidence alongside policy and planning documents, statistics, institutional materials, and peer-reviewed research. Five interrelated mechanisms were identified in Singapore: nested planning and implementation rules; networked multifunctional land use; project-level interagency co-production; professional stewardship linked to operational data; and sustained participation combined with differentiated access. Shanghai's recent plans address several corresponding functions, including life-cycle park management, integrated spatial data, ecological monitoring, and collaborative governance, but published evidence does not establish how consistently they operate across districts, projects, and asset owners. The comparison supports conditional rather than direct transfer: governance functions should be adapted to local authority, spatial and climatic conditions, operating capacity, and accountability. The resulting implementation-chain framework links spatial connectivity with institutional continuity and helps diagnose where urban landscape and greenway objectives may become disconnected from long-term operation.

Abstract: The era of high-quality urban development demands refined, digitalized, and visualized scientific research, and 3D point cloud acquisition technology has emerged as a key tool for tree information collection, surveys, assessment, monitoring, and management, which facilitates the innovative development of the landscape architecture industry. While existing applications primarily focus on batch acquisition of point clouds for tree communities, scattered trees in fields such as ancient and famous tree conservation still rely on backpack/handheld LiDAR devices. These devices often require multiple circular scans around individual trees to ensure point cloud completeness, but motion distortion resulting from such multi-circular mobile acquisition has been generally overlooked. Based on Popper’s falsificationism, this study employs three methods: tree parameter comparison, optical porosity analysis, and CloudCompare-based point cloud distance comparison. These methods are used to observe that motion distortion significantly impairs the accuracy of individual tree point cloud data. Results indicate that motion distortion increases point cloud volume and Euclidean distance errors. Notably, a greater number of collection circles induces random variations in tree morphological parameters and a reduction in point cloud optical porosity. Consequently, tripod-mounted LiDAR is recommended over handheld/backpack devices. Significantly, this study is the first to highlight the potential substantial impact of accumulated point cloud motion distortion on landscape tree point cloud data acquisition. It provides a theoretical foundation for enhancing the precision of dynamically acquired tree point cloud data, which in turn improves the accuracy of tree analysis, monitoring, and assessment studies based on tree point cloud data.

Abstract:

High-altitude lakes are increasingly exposed to nutrient enrichment, organic loading and wastewater-derived contamination, while predictive performance can be overstated when temporally ordered observations are randomly partitioned or target-defining measurements are reused as predictors. We analyzed 177 consecutive monthly water-quality records from the Inner Bay of Lake Titicaca, Peru (January 2011–September 2025), integrating Carlson’s composite trophic state index (CTSI), Hamed–Rao modified Mann–Kendall tests, block-bootstrap Sen slopes, Pettitt change-point detection with 12-month block permutation, nutrient stoichiometry, correlation, principal component analysis, leakage-aware contemporaneous classification and one-month-ahead forecasting. Total phosphorus was treated as elemental P and phosphate as PO4. The bay remained chronically hypereutrophic (mean CTSI 74.08 ± 4.07; 83.1% of months). BOD5 increased by 0.548 mg L⁻¹ yr⁻¹, chlorophyll a by 3.254 mg m⁻³ yr⁻¹, total suspended solids by 0.752 mg L⁻¹ yr⁻¹ and conductivity by 13.13 µS cm⁻¹ yr⁻¹, whereas total phosphorus declined by 0.085 mg P L⁻¹ yr⁻¹. Block-supported shifts occurred in chlorophyll-a in November 2016 (21.81 to 54.18 mg m⁻³) and BOD5 in June 2018 (6.54 to 11.77 mg L⁻¹). After removing target-defining variables and preserving temporal order, the best trophic-state classifier had balanced accuracy 0.575 and MCC 0.257. Organic-pollution classification had balanced accuracy 0.624 but sensitivity only 0.271, whereas the fecal-indicator model was unstable (MCC 0.130; ROC-AUC 0.460). Persistence was the best BOD5 forecast (RMSE 4.193 mg L⁻¹; R² 0.390). The best CTSI model explained only 4.8% of future variance, and all thermotolerant-coliform forecasts had negative out-of-time R². Persistent ecological degradation was therefore evident, but monthly observations alone were insufficient for deployment-ready early warning. Higher-frequency sensing, hydrometeorological and wastewater-load covariates, spatial replication, direct microbiological measurements and prospective validation are required.

Abstract: The rhizosphere microbiome plays a crucial role in the development, health, and therapeutic properties of Panax ginseng. Therefore, we evaluated the spatial trends in bacterial communities using a comprehensive meta-analysis of 359 sequenced rhizosphere samples from seven NCBI Sequence Read Archive BioProjects. The SILVA reference database was used to compare the Chinese and Korean sequencing datasets, focusing on quality filtering, denoising, amplicon sequence variants, and taxonomy assignment. The bacterial community structure was characterized, and differentially abundant taxa were identified using the Kruskal-Wallis test with false discovery rate correction. Co-occurrence network analysis illustrated region-specific microbial relationships and community organization, whereas functional profiles were projected to assess potential metabolic differences. Despite regional and technical variations, 37 bacterial species consistently emerged as conserved core microbiome members. Regional analyses revealed distinct bacterial drivers, such as Ligilactobacillus, Romboutsia, Nitrospira, and Bradyrhizobium in the Korean samples and Sphingoaurantiacus and Holophaga in the Chinese samples. Functional predictions showed that plant-associated defense mechanisms, nutrient mobilization, and carbon cycling varied geographically. Network studies demonstrated that community structures differed across regions. Overall, these findings indicate that although a conserved core community exists, regional differences significantly influence the composition, functional potential, and ecological relationships of the P. ginseng rhizosphere.

Abstract: The decline in ecosystem services in landscapes managed by rural communities stems largely from inappropriate human intervention. Therefore, this study examines a case within this context, considering the current challenges faced by these communities and the transitions they are beginning to undergo to better co-produce ecosystem services within their territories. The methodology employed involved the analysis of satellite imagery to study the dynamics of land-use and vegetation changes, as well as the examination of three ecosystem services: water regulation, provisioning and cultural - recreational. Participatory research was also conducted through interviews, focus groups, and direct observation. A notable dynamic of land-use change was identified, including the transformation of natural vegetation cover, a network of human activities that mutually affect one another and reduce the landscape’s potential to provide services, and a tacitly exercised power structure that manifests itself through ecological impacts. Faced with this situation, residents are debating the need not only to implement ecological measures but also to expand their networks and seek alternatives - such as glocal strategies - that will allow them to place their products in commercial chains that are more advantageous for them.

Abstract: Coarse cereals are indispensable to both national food security and the implementation of the Healthy China Initiative, underscoring the urgent need to promote their green and lowcarbon production. Using panel data from Shanxi Province spanning 2010–2020, this study first employed water footprint (WF) and carbon footprint (CF) accounting methods, together with a coupling coordination degree (CCD) model, to characterize the spatiotemporal dynamics of WF, CF, and their CCD for six major coarse cereal crops across the province. On this basis, the entropyweighted TOPSIS model was applied to assess the spatiotemporal patterns of comprehensive production potential (CPP), while exploratory spatial data analysis (ESDA) was used to detect spatial autocorrelation. Subsequently, path analysis was conducted to identify the key determinants of CPP. The results indicated that: (1) both total WF and CF per unit yield exhibited declining trends with pronounced spatial heterogeneity, generally higher in northern Shanxi, relatively balanced in the central region, and lower in the south. (2) the CCD between WF and CF showed a sustained upward trajectory over the study period.(3) the CPP of all selected coarse cereal crops increased significantly. and (4) effective irrigated area exerted the strongest positive effect on CPP, whereas the net application rate of agricultural chemical fertilizers and the proportion of leguminous coarse cereals imposed notable negative effects. Accordingly, we propose the following targeted strategies: expanding watersaving effective irrigated areas steadily, strictly curbing fertilizer overuse, tailoring dominant coarse cereal varieties to local agroecological conditions, and optimizing the regionspecific development of characteristic coarse cereals.

Abstract: Wetland vulnerability assessment is essential for identifying interacting pressures, detecting potential changes in ecological character, and supporting adaptive management of Ramsar Sites. This study develops an ecological character-based vulnerability assessment framework and applies it to the Ramsar Site Bau Sau (Crocodile Lake, Ramsar Site No. 1498) in Cat Tien National Park, Vietnam. The framework integrates 26 indicators within six criterion groups and three components: exposure/pressure (E), sensitivity (S), and adaptive capacity/resilience (AC). Indicators were selected to explicitly link climatic, hydrological, ecological, and anthropogenic pressures with the ecosystem components, processes, and functions that define wetland ecological character. The assessment yielded E = 0.500, S = 0.604, and AC = 0.688, resulting in an overall vulnerability index of V = 0.472, classified as moderate. Hydrological variability, inundation dynamics, and hydrological-environmental sensitivity emerged as the principal drivers of vulnerability, whereas relatively strong conservation, management, and monitoring capacity reduced overall vulnerability. The findings indicate that maintaining hydrological processes and ecological character should be central to adaptive management at Bau Sau. More broadly, the proposed framework provides a transparent and adaptable approach for linking Ramsar ecological-character principles with quantitative vulnerability assessment and management prioritization in data-limited wetlands.

Abstract: Pollution of coastal territories by marine litter is a pressing issue not only for large cities but also for remote and uninhabited regions such as the Arctic. At the same time, the scale of pollution is growing annually, while existing environmental monitoring methods remain labour-intensive. For the first time, an assessment of the volume and composition of marine litter was conducted for Arctic coastlines using high-resolution airphotos. The method is based on training the YOLOv8 (2023) convolutional neural network in the OBB configuration to recognise plastic, metal, fishing nets, wood, and other types of marine litter in images. Based on a comparative analysis, it was established that machine learning methods combined with remote sensing data represent a new high-precision tool for environmental monitoring, which can significantly contribute to reducing the anthropogenic pressure on the ecosystems of Arctic territories.

Abstract: Prosopis chilensis (Mesquite) is one of the world’s most aggressive invasive woody species, and its ecological impacts have drawn increasing scientific attention, particularly in arid and semi-arid ecosystems. In Sudan, it is rapidly reshaping Doum palm (Hyphaene thebaica) forests, yet its effects on native forest ecosystems remain poorly quantified. This study aimed to: (i) assess tree species composition and structure, (ii) evaluate the impact of P. chilensis invasion on species richness and native species abundance in H. thebaica forests. A stratified random sampling approach was employed across 59 plots within three forests reserves, namely Alhelgi, Geli and Um Bashim. P. chilensis showed overwhelming dominance in Alhelgi and Geli forests, with highest densities of 131 and 60 trees ha⁻¹, and accounting for 83% and 86% of total tree density, and relative abundance of 63% and 71% of total tree stems, respectively. In contrast, P. chilensis was absent in Um Bashim forest. H. thebaica was the dominant native tree species, with the highest Importance Value Index (IVI) of 137. A total of nine species, belonging to seven genera and six families, were recorded across the studied forests. Within the invaded forests, Pearson’s correlation confirmed that increasing of P. chilensis abundance significantly reduced species richness (R= -0.34, p=0.021). While Kendall’s tau confirmed a significant decline in native species abundance at the community level (τ = −0.22, p = 0.036). At the individual species level, this negative effect was not statistically significant for most species, except for Vachellia nilotica, which showed a significant decline (τ = −0.28, p = 0.022). In addition, the findings reveal P. chilensis as a formidable and accelerating threat to Doum palm forests’ integrity. The study recommends integrated management strategies, including the control of P. chilensis and restoration of native vegetation, to safeguard these dryland ecosystems.

Abstract: Fishing Refuge Zones (FRZs) seek to promote the recovery of exploited populations and associated ecosystems, although their effectiveness depends on various factors, such as habitat condition, the reduction of local stressors, and management implementation. We evaluated changes in the reef fish community of the Akumal FRZ, in the Mexican Caribbean, ten years after its establishment, with emphasis on commercial and herbivorous fishes. In 2025, we revisited 23 shallow and deep reef front sites surveyed in 2015 using visual censuses and benthic video transects. Fish biomass, abundance, and richness were analyzed using generalized linear mixed models; size structure using kernel density estimates; and the relationship between fish biomass and benthic condition using redundancy analysis. Commercial species showed no significant differences in biomass, richness, or size structure, but had lower abundance in 2025. In contrast, herbivores showed higher biomass and a shift toward a larger modal size, although with lower abundance and richness. These changes occurred on a reef with lower coral cover in 2025 and a predominance of brown algae in both periods. The results show a partial and functionally heterogeneous response, in a context of persistent degradation and limitations in management implementation and continuity.

Abstract: Grassland degradation is a global ecological crisis that profoundly alters aboveground vegetation and soil properties, yet its impacts on soil microbial co‑occurrence networks remain poorly understood. Here, we investigated soil bacterial and fungal communities along a well‑defined five‑stage degradation gradient spanning from non‑degraded Leymus chinensis grassland to extremely degraded bare saline patches in the Songnen meadow steppe of northeastern China, using Illumina MiSeq sequencing and co‑occurrence network analysis. Our results revealed that bacterial α‑diversity exhibited a unimodal (hump‑shaped) response peaking at the moderately degraded KA stage, whereas fungal diversity declined monotonically along the gradient, indicating greater sensitivity of fungi to degradation stress. Both bacterial and fungal community compositions shifted directionally with degradation, driven primarily by soil alkalization (pH) and electrical conductivity (EC). Network complexity followed a unimodal pattern for both kingdoms, maximizing at KA and collapsing at AS, suggesting a critical ecological threshold between moderate and severe degradation, while the increased proportion of positive correlations under severe degradation implied enhanced microbial cooperation in response to environmental stress. Structural equation models further revealed distinct regulatory pathways: bacterial networks were governed by both direct environmental filtering and indirect diversity‑mediated effects, whereas fungal networks responded more strongly to direct pH/EC constraints and compositional shifts. Our findings demonstrate that microbial networks exhibit nonlinear threshold responses to grassland degradation, with fungi serving as more sensitive bioindicators than bacteria, and highlight the importance of integrating network‑level properties into degradation monitoring and restoration frameworks.

Abstract: This article presents a design-led environmental media prototype for translating saproxylic beetle evidence into a web-based narrative about slow forest degradation in Crete. The prototype uses five interface scenes to convert ecological constructs, including deadwood continuity, decay-stage diversity, microclimate refuge, drought-stress coupling, pathogen/vector risk, and fragmentation, into comparison, gradient, threshold, and reachability tasks. The contribution is methodological rather than ecological: it proposes an evidence-to-interface translation matrix and an anti-cutesification/anti-spectacle grammar for making proxy variables, uncertainty cues, and versioned sources inspectable in public-facing environmental media. A small mixed-background reader-response session (N = 18) is used as a boundary test of readability. The session suggests that many readers could retell the deadwood-removal inference and distinguish structural loss from simple character appeal, while also revealing recurring misreadings around proxy status, scenario precision, and particle-haze risk cues. The article therefore positions the prototype as a proof-of-concept for auditable environmental-media translation, with broader claims about public use, institutional uptake, and long-term durability reserved for future comparative and longitudinal testing.

Abstract: Northeast India sits within the Indo-Burma and Eastern Himalaya biodiversity hotspots, yet its forests face accelerating loss from extractive industries, infrastructural expansion, and weak governance. Although global conservation discourse increasingly recognises community-based approaches, the contributions of vernacular grassroots organisations from frontier regions remain insufficiently documented in peer-reviewed scholarship. This article addresses that gap. It examines the conservation praxis of Nature's Beckon, an independent activist organisation founded in Dhubri, Assam, in 1982, and the work of its founder-director Soumyadeep Dutta, an Ashoka Fellow and naturalist-writer. The study adopts a qualitative, analytical methodology that integrates historical and ecocritical approaches. Data are drawn from organisational records, Dutta's published Assamese-language corpus, news media, government notifications, and peer-reviewed scientific literature on protected-area outcomes. Four case studies are analysed: the Chakrashila Wildlife Sanctuary movement (1983–1995); the Dehing Patkai rainforest campaign (1992–2021); primate conservation focused on the golden langur (Trachypithecus geei) and western hoolock gibbon (Hoolock hoolock); and the Village Sanctuary institutional model. Findings indicate that Nature’s Beckon has played a documented and substantial role in measurable conservation outcomes, including the gazettement of two protected areas with a combined area of approximately 277.22 km². The article advances an ecology of the margins framework to characterise the organisation's distinctive integration of scientific documentation, vernacular advocacy, and community institution-building. Practical policy implications include formal recognition of community-conserved areas, integration of vernacular knowledge into protected-area management, replication of the eco-emissary training framework, and alignment of conservation funding cycles with the multi-decadal time horizons that durable place-based conservation requires.

Abstract: Stachys fontqueri is a strict endemic species of the Moroccan Rif that depends on specific ecological conditions. To understand the effect of climate change on the potential evolu-tion of the species' geographic range under current and future climatic scenarios, ecologi-cal niche modelling was performed using MaxEnt algorithm based on five bioclimatic variables. To model the effect of climate change, four climatic scenarios were used, namely CSM2- SSP1-2.6, CSM2-SSP5-8.5, MIROC6-SSP1-2.6, and MIROC6-SSP5-8.5, for the period 2061–2080. The results demonstrated high model performance, with an AUC ranging from 0.921 to 0.930 and a TSS from 0.81 to 0.83. Three bioclimatic variables contributed significantly to determining the suitable potential distribution area of the species, namely Precipitation Seasonality (Bio15), Temperature Annual Range (Bio7), and Annual Mean Temperature (Bio1). The suitable area covered 3,244 km² under the current climate and is projected to decrease by 23.25% to 29.59% under future climate scenarios. This contraction of suitable habitat due to climate change could be exacerbated by human activities, there-by requiring urgent in situ and ex situ conservation measures to ensure the species' resilience.

Abstract: Accurate monitoring of forest carbon stocks represents a critical prerequisite for achieving carbon neutrality. However, conventional remote sensing‑based estimation methods frequently overlook forest heterogeneity, causing systematic overestimation or underestimation. To address this gap, we propose a novel forest carbon stock esti-mation framework that integrates two complementary strategies: (1) forest type‑specific modeling to account for forest heterogeneity, and (2) hyperparameter op-timization to enhance Random Forest model performance. Using ground‑measured carbon stocks and a CCDC‑derived forest vegetation classification map for Hangzhou City, China, we built forest‑type‑specific Random Forest models based on ICESat‑2 canopy height metrics and optimized each model via hyperparameter tuning. The re-sults show that the 70th-90th percentiles and the mean canopy height are relatively highly correlated with carbon stock. Forest‑type‑specific modeling improves estima-tion accuracy, yielding R² gains of 0.10–0.17 and reduced RMSE by 2.28–7.43 Mg C/ha over the non‑stratified model. Integrating forest classification and hyperparameter op-timization strategies improved model R² by 0.16–0.23 and lowered RMSE by 3.05–8.20 Mg C/ha. Overall, this study demonstrates that accounting for forest heterogeneity and applying hyperparameter optimization can significantly enhance the accuracy of for-est carbon stock estimation.

Abstract: Korla fragrant pear diseases and pests detection faces challenges such as significant object scale variation, multi-organ target confusion, and limited computational resources for real-time inference on edge devices. To address these issues, this study proposes a lightweight object detection model, KFP-YOLO, based on YOLO26n. A multi-organ dataset, the Korla fragrant pear diseases and pests dataset (KFP-PDD), was collected, covering pear leaves, fruits, and flowers and containing 11 classes including healthy and diseased samples. To reduce computational cost while maintaining effective feature representation, an ADown lightweight downsampling module is introduced. A C3-PD feature extraction module is designed by integrating Partial convolution and SE attention to reduce redundant computation and enhance feature representation capability. Furthermore, a CFA feature enhancement module is proposed, which incorporates coordinate attention into the multi-scale feature fusion process to improve spatial information modeling and fine-grained feature representation. Experimental results show that, compared with YOLO26n, KFP-YOLO reduces parameters and GFLOPs by 23.0 % and 23.7 %, respectively, while achieving an inference speed of 278.97 FPS. Meanwhile, mAP@0.5 reaches 94.65 %, with only a 0.44 percentage point drop. Ablation studies verify the effectiveness and synergistic optimization of each proposed module. Deployment experiments on the Jetson AGX Orin platform demonstrate strong real-time performance and edge computing adaptability, indicating that the proposed method offers an efficient solution for intelligent orchard diseases and pests monitoring.

Abstract: Microplastic pollution has become an emerging threat to marine biodiversity, yet the developmental-stage-specific responses of endangered marine species remain poorly understood. Tachypleus tridentatus, a threatened marine arthropod of high ecological and conservation value, is exposed to microplastics in coastal habitats. In this study, five-instar-old and six-instar-old juvenile T. tridentatus were exposed to polystyrene microplastics (PS-MPs) with a diameter of 6 μm at environmentally relevant concentrations (0, 102, 104 particles/L) for 21 days. Fluorescence imaging revealed the accumulation of PS-MPs within the intestinal tract, with distinct retention characteristics among developmental stages. Transcriptome analysis revealed significant differential gene expression after exposure, with affected pathways related to energy metabolism, immune regulation, stress response, signal transduction, and development. Younger juveniles exhibited stronger transcriptional disturbances in metabolic and stress-related pathways, whereas older juveniles showed more coordinated regulation of homeostatic and adaptive genes. These findings suggest that juvenile T. tridentatus employs age-dependent molecular response strategies when confronted with microplastic exposure, reflecting developmental differences in stress sensitivity and adaptive capacity. Our study provides new insights into the molecular mechanisms underlying microplastic responses in an endangered marine species and highlights the importance of incorporating developmental-stage variation into ecological risk assessments and conservation strategies for horseshoe crabs in polluted coastal ecosystems.

Abstract: Accurate identification of suitable habitats is essential for the ecological, economic, conservation, and sustainable management of Almaciga (Agathis philippinensis). This study employed a geospatial Maximum Entropy (MaxEnt) species distribution modeling approach to predict the habitat suitability of Almaciga and identify the environmental factors influencing its distribution. Twenty environmental variables were initially evaluated using Pearson correlation analysis, hierarchical clustering, and Variance Inflation Factor (VIF) to minimize multicollinearity. The survey occurrences were split into 80% for training and 20% for testing. The final predictor set consisted of the maximum temperature of the warmest month (BIO5) and elevation (ELEV). Model performance was evaluated using the Area Under the Receiver Operating Characteristic Curve (AUC) with 99% and 98% accuracy for training and testing, respectively. The elevation was the most influential variable, contributing 90.5% to the model and accounting for 93.8% permutation importance. The suitability maps revealed that highly suitable areas (i.e., 331.27 km2) are concentrated in mountainous regions categorized by higher elevations and cooler temperatures. These findings provide valuable spatial information for conservation planning, habitat restoration, sustainable forest management, and climate adaptation strategies for Almaciga across the Philippines.

Abstract: Frequent hydration–desiccation cycles in Antarctica impose strong selective pressures on bryophytes physiology and associated microbiota. Here, we investigated physiological and transcriptional responses of the Antarctic moss Polytrichastrum alpinum and its associated microbial communities under controlled hydration, desiccation and rehydration conditions. Desiccation induced strong but reversible physiological and molecular changes, with most gene expression patterns returning toward hydrated states upon rehydration, indicating efficient recovery of metabolic activity. This response involved coordinated regulation of stress-associated gene groups, including LEA proteins, heat-shock proteins and components of ABA and calcium signalling. Furthermore, field-collected samples exhibited transcriptional profiles similar to desiccated material, suggesting that this species naturally operates in a persistently water-limited state. Associated bacterial and fungal communities showed coordinated shifts in both composition and functional activity during the experiment, particularly in pathways related to carbon and amino acid metabolism. These results highlight the resilience of Antarctic P. alpinum and its microbiome under increasing environmental variability driven by climate change.

Abstract: The rapid expansion of artificial intelligence raises critical questions about its environmental implications in an increasingly globalized world. In this study, we explore the impact of Artificial Intelligence (AI) on environmental sustainability and its role in globalization, which bears significant implications for global sustainability in the digital era. We used the comprehensive evaluation index of the artificial intelligence index established by the Principal Component Analysis (PCA) method on a sample of 62 countries globally, from 2000 to 2022. Our study is based on the estimation of the Autoregressive Distributed Lag (ARDL) model, which distinguishes between short- and long-term dynamics. We noted that the overall level of global AI shows an upward trend. The research results show that AI has a positive and significant inhibitory effect on ecological footprints and greenhouse gas emissions in the long term. In addition, affected by globalization, the effects of AI on environmental sustainability show nonlinear characteristics. AI’s marginal effect on reducing environmental degradation increases. These findings emphasize the important role of AI in environmental governance and provide a new and comprehensive perspective for policymakers. Policymakers should prioritize AI R&D investment, promote cross-sectoral AI integration and strengthen international cooperation to maximize environmental benefits.