Microgrid clusters can enhance the utilization rate of renewable energy and system reliability, but their traditional dispatching methods have problems such as insufficient flexibility and high computational complexity. To address this issue, this paper proposes a two-layer collaborative optimization dispatching method for microgrid clusters based on an improved SAC algorithm. Firstly, a two-layer collaborative optimization dispatching model for microgrid clusters is constructed, aiming to minimize the operation cost, where the upper layer is responsible for the global power optimization dispatching of the microgrid cluster, and the lower layer is responsible for the local power optimization dispatching of each sub-microgrid. Secondly, an improved SAC algorithm is proposed, which introduces a variational autoencoder (VAE) model into the traditional SAC algorithm to utilize the feature extraction ability of VAE to reduce the dimension of high-dimensional states and extract key features, thereby reducing the computational complexity of the algorithm. Then, a time decay factor is introduced on the basis of the prioritized experience replay mechanism to increase the adoption frequency of later-entered experiences, further improving the learning efficiency of the algorithm. Finally, the improved SAC algorithm and the alternating direction method of multipliers are used to optimize and solve the upper and lower layer models of the microgrid cluster respectively. Through the interaction mechanism of the upper and lower layers, efficient collaborative dispatching and economic operation of the microgrid cluster are achieved. Simulation results show that the proposed method outperforms Scheme 1, Scheme 2 and Scheme 3 in terms of total operation cost, load balance rate and energy utilization rate. To be exactly, the operating costs of the proposed method in this paper are reduced by14.0%, 10.2% and 6.4%, respectively; the load balance rate is increased by 9.3 percentage points, 6.4 percentage points and 3.6 percentage points, respectively; and the energy utilization rate is raised by 10.8 percentage points, 7.6 percentage points and 2.9 percentage points respectively. This result verifies the superiority of the proposed method in the optimal dispatch of microgrid clusters. Meanwhile, this method also shows good robustness in dealing with the prediction errors of wind and solar power generations.

In order to enhance the giant magnetoimpedance (GMI) performance and sensitivity of amorphous microwires, this study fabricates carbon nano-layer-coated Co-based composite amorphous microwires by using magnetron sputtering technology. The carbon layer thicknesses are set at 67, 203, 334, 467, and 601 nm, respectively. A systematic investigation is conducted on the microstructure, magnetic properties, and the influence of varying carbon layer thicknesses on the giant magnetoimpedance effect in the composite microwires. The results indicate that the composite amorphous wire with a carbon layer thickness of 334 nm exhibits optimal GMI performance, achieving a maximum GMI ratio of 522% and a corresponding sensitivity of 33.8% Oe^-1. By constructing a GMI biosensing element from this high-performance composite amorphous wire and integrating it with a bio-molecular reaction interface based on a patterned SiO₂ thin film and a PDMS microfluidic chip, ultrasensitive detection of cardiac troponin I(cTnI) is accomplished via a sandwich immunoassay. Experimental results demonstrate that the sensor exhibits a favorable linear response within the concentration range of 0.001-500 μg/L, with a detection limit as low as 0.001 μg/L, along with outstanding specificity, stability, and anti-interference capability. This study provides significant support for biomolecular detection technology based on the GMI effect in amorphous wires.

In response to the demand for efficiently boosting weak environmental energy to a usable voltage in energy harvesting systems, based on the structure of cross-coupled boost charge pumps, a high-efficiency, fast and stable boost charge pump circuit controlled by multi-phase clocks is proposed. Firstly, this circuit precisely controls the conduction timing of the charge transfer switch by adding auxiliary MOS transistors, auxiliary capacitors and using a six-phase non-overlapping clock, effectively suppressing the phenomenon of charge back flow and reducing energy loss. Then, by integrating dynamic body bias technology, conduction losses are reduced and sub-threshold leakage currents are suppressed, thereby further enhancing power conversion efficiency. Finally, the circuit is designed and verified through a simulation based on the 180 nm CMOS process. The simulation results show that under the input voltage of 0.6 V, the output voltage of the single-stage charge pump can reach 1.13 V, the maximum power conversion efficiency can reach 95.28%, the output ripple is as low as 40 mV, and the output voltage is stable within 1 μs.Therefore, it can be widely applied in low-power energy harvesting systems such as internet of things nodes.

To address issues such as sampling nonlinearity, high power consumption of capacitor array switching, and the trade-off between area and energy efficiency in high-speed and medium-to-high-precision analog-to-digital converters (ADCs), this paper proposes a 12-bit 100 MS/s successive approximation register ADC (SAR ADC) based on charge redistribution principles, implemented in 65 nm CMOS technology. A bootstrap switch is employed to mitigate input signal nonlinearity during the sampling phase, while an improved monotonic switching algorithm optimizes the switching power consumption and area of the binary-weighted capacitor array. A dynamic comparator is selected to ensure lower power consumption. Simulation results demonstrate that at a 100 MS/s sampling rate, the ADC achieves a signal-to-noise and distortion ratio (SNDR) of 64.5 dB and a spurious-free dynamic range (SFDR) of 74.53 dB at the Nyquist input frequency, corresponding to an effective number of bits (ENOB) of 10.4. With a power supply voltage of 1.2 V, the total power consumption is only 4.8 mW, showcasing excellent performance and energy efficiency competitiveness in high-speed medium-precision applications.

To address the limitations of existing defect detection methods, namely weak edge feature extraction for small targets, inadequate multi-scale perception under complex backgrounds and irregular morphologies, and limited localization accuracy, an improved YOLO11n-based PCB surface defect detection model, termed EMD-YOLO, is proposed. An Edge Enhancement (EE) module and a Multi-Scale Edge Information Enhancement (MSEI) module are introduced to strengthen the representation of edge features for minute and structurally complex defects. Additionally, a lightweight multi-scale pooling scheme is employed to construct a feature pyramid, by which the network’s multi-scale perception capability is enhanced. Furthermore, the Distance-IoU (DIoU) loss function is adopted during training to optimize bounding box regression, by which localization accuracy is further improved. Experimental results on the PKU-Market PCB dataset demonstrate that a 3.8 percentage point improvement in mean average precision (mAP@50) is achieved by EMD-YOLO compared to the baseline YOLO11n model, while nearly unchanged computational complexity is maintained, thereby significantly enhancing the model’s practicality and robustness.

In the treatment of skin cancer, early skin lesion segmentation technology is of great importance. However, in existing skin lesion segmentation models, global features of images are difficult to be extracted by CNNs. Computational overhead of the model is increased by Transformers, resulting in persistently high training costs. To solve this problem, the Local Feature Enhanced Mamba U-Net (LEM-UNet) model is proposed in this paper. Based on Vision Mamba-UNet, the decoder is reconstructed to enhance global feature extraction. A multi-scale feature extraction residual module (RMS) and a local feature extraction module (LFM) are proposed. These modules strengthen the model’s ability to segment lesions of different sizes in images and improve the model’s accuracy in capturing and extracting lesion boundary features, thus enhancing the overall segmentation effect of the model. Comparative experiments are conducted on the ISIC2017 and ISIC2018 public skin lesion segmentation datasets. The results show that compared with the original baseline model, mIoU, DSC, Acc, Spe, and Sen increase by 0.97, 0.56, 0.22, 0.20, 0.29 percentage points and 1.55, 0.91, 0.41, 0.16, 1.23 percentage points respectively. The model performs particularly well in scenarios where the contrast between the segmented object and the background is not high.

An end-to-end language-prompt multi-object tracking framework, termed enLPMOT (enhanced language-prompt multi-object tracking), is proposed to introduce high-level semantic information into the visual tracking process and to enable semantic-aware target perception and continuous association. In complex dynamic scenes, language-prompt multi-object tracking is still severely affected by frequent identity switches, heavy occlusion, and insufficient temporal consistency,which are mainly caused by insufficient diversity of supervision signals and inadequate modeling of cross-frame spatial relationships in existing methods. To address these issues, a grouped-query supervision mechanism is introduced, in which a one-to-many assignment strategy is adopted to enhance supervision diversity. An enhanced dual-path decoding structure is designed, where original queries and noise-enhanced queries are decoded in parallel and adaptively fused, so that the robustness of the model under occlusion and appearance variation is improved. Meanwhile, explicit relative geometric encoding is incorporated to model cross-frame spatial dependencies, thereby strengthening the temporal consistency of trajectories. Experimental results demonstrate that competitive performance is achieved by the proposed method on multiple evaluation metrics. Specifically, the HOTA (higher order tracking accuracy) reaches 40.30%, the DetA (detection accuracy) reaches 30.60%, and the AssR (association recall) and AssP (association precision) reach 55.77% and 83.53%, respectively.

To address the challenges of difficult cross-turn strategy evolution, emotional tone mismatch, and strategy shifts in negotiation dialogue generation, a model integrating Graph Attention Networks and emotional feedback is proposed. In this model, BART is utilised to capture global contextual semantics, and deep structural priors are extracted by constructing a dynamic graph structure containing strategic techniques and dialogue behaviours, combined with multi-head graph attention and adaptive structure-aware pooling mechanisms. To resolve the alignment problem between semantics and strategic logic, structure-aware attention and gating mechanisms are adopted to achieve adaptive information fusion. Furthermore, gradient feedback generated from an emotion auxiliary task is innovatively utilised to globally calibrate the strategy graph representation, ensuring consistency between strategic decisions and the target emotional context. Based on experimental results on the CraigslistBargain dataset, superior performance over existing baseline models is demonstrated across all core metrics, with the BLEU (bilingual evaluation understudy) score reaching 19.74% and the RC-Acc(ratio class prediction accuracy) improving to 54.21% for the negotiation outcome. Finally, the effectiveness of the graph structure modelling and emotion calibration modules in enhancing negotiation logical coherence and emotional adaptability is further validated by ablation studies and human evaluations.

The Vertex Cover (VC) problem, a classical combinatorial optimization problem in graph theory, has significant practical applications. In dynamic graph environments where edges are added incrementally, traditional static algorithms must recompute solutions from scratch at a high computational cost and with difficulty to maintain the structural stability of solutions. To address these challenges, a Hybrid Edge-Incremental Vertex Cover algorithm (HEIVC) is proposed in this paper. Starting from an initial minimal vertex cover, three complementary strategiesare integrated: rapid candidate generation, subgraph local search, and edge-independent subset partitioning combined with conflict graph-based candidate construction. Then, an adaptive module selection mechanism is designed to dynamically activate these strategies based on incremental scales and local perturbations with the objectives to maintain solution feasibility and minimality while enhancing stability and quality. Compared with the existing 2-approximation greedy algorithms and the IMVC algorithm, the results show that HEIVC achieves a significant advantage in solution size across different module combinations. These results validate the effectiveness of the proposed approach in dynamic graph incremental maintenance.

The forms and properties of the Bayesian estimation for the scale parameter in the inverse gamma distribution with known the shape parameter were investigated under the gamma conjugate prior distribution and the weighted p,q symmetric entropy loss function. Firstly, the exact form of the Bayesian estimation for the scale parameter is derived, and its admissibility and minimaxity are proved. Secondly, the forms of multi-layer Bayesian estimation, E-Bayesian estimation, and knife-cut Bayesian estimation for the scale parameter of the inverse gamma distribution are investigated based on the derived Bayesian estimation. Finally, numerical simulations with MCMC algorithm are conducted to discuss the robustness and precision of the obtained Bayesian, E-Bayesian, and knife-cut Bayesian estimation. Results indicatethat all three estimations exhibit high robustness and precision, while knife-cut Bayesian estimation has the highest precision specially.

A control strategy integrating dynamic quantization and an improved memory event-triggered mechanism is proposed for T-S fuzzy systems subject to limited network bandwidth. Firstly, in order to decrease the burden of network transmission, a dynamic quantizer is introduced to process system states. Secondly, a memory event-triggered mechanism incorporating historically transmitted data is designed to further minimize unnecessary data transmissions. Taking into account that aperiodic denial-of-service (DoS) attacks and deception attacks may occur during network communication, a hybrid attack model is established, with explicit conditions on the maximum allowable attack frequency and duration derived. A controller making full use of historical information is designed, in which the controller gain matrices and the weight matrices of the event-triggering condition are obtained using linear matrix inequality (LMI) techniques, thereby achieving co-optimization of controller gains with quantization parameters, triggering thresholds, and attack characteristics. Subsequently, a Lyapunov-Krasovskii functional is constructed, and sufficient conditions are derived to ensure the asymptotic stability of the closed-loop system while satisfying the prescribed H_∞ performance. Lastly, a numerical example is provided to illustrate the effectiveness of the proposed control strategy.

Based on daily meteorological data from 1959 to 2018, this study employed extreme climate indices, climate tendency rates, abrupt change detection, Morlet complex wavelet analysis, Pearson correlation, and wavelet coherence analysis to investigate the characteristics of extreme climate and its influencing factors in the study area. The main findings are as follows: 1) Warm extreme temperature indices showed an increasing trend, while cold extreme temperature indices exhibited a decreasing trend. The abrupt changes in extreme temperature indices occurred notably and concentratedly around1982-1994 year and1998-2006 year, with dominant cycles of 9-13 years and 44-57 years. Extreme precipitation duration indices displayed a decreasing trend, while the trends in extreme precipitation intensity indices showed spatial heterogeneity. The abrupt changes in extreme precipitation indices were insignificant and scattered, with dominant cycles of 6 years and 55-57 years. 2) Absolute extreme temperature indices showed significant correlations with multi-year average temperature (positive), elevation (negative), and latitude (negative). Extreme precipitation indices exhibited relatively significant correlations with multi-year average temperature (positive),daily precipitation (positive), and longitude (negative), displaying a trend of decreasing from west to east. 3) The AMO (Atlantic Multidecadal Oscillation)was extensively correlated with extreme temperature indices. Most extreme temperature indices lagged behind atmospheric circulation indices, which aligns with the general physical process of ocean-atmosphere driving. The weakening or reversal of phase relationships between extreme temperature indices and atmospheric circulation indices during specific periods arised from the influence of other strong climate signals.

Scientifically precise spatial distribution maps of landslide susceptibility provide an important basis for landslide disaster prevention and control, as well as the construction and planning of social infrastructure. Based on 752 sets of historical landslide data, this study comprehensively applied an information value model using frequency ratios and an AHP-GIS model to conduct a landslide susceptibility assessment, ultimately producing a 30 m-resolution landslide susceptibility zoning map for Guilin City. Furthermore,with the help of geographical detector, an explanatory power analysis was conducted on ten evaluation factors: distance to faults, precipitation, slope, elevation, population distribution, lithology, aspect,normalized difference vegetation index (NDVI), distance to water systems, and distance to roads, to assess their influence on landslide susceptibility in Guilin City. The research results indicate that:1) The spatial distribution of landslide susceptibility in Guilin exhibits a pattern of higher susceptibility in the south and lower in the north; 2) The overall level of landslide susceptibility in the city is relatively high, with extremely high and high susceptibility areas accounting for over 50% of the total; 3) Single-factor detection analysis shows that lithology has the most significant impact on landslide susceptibility, while the interaction of factors exhibits nonlinear enhancement and two-factor enhancement relationships. Among these, the interaction between lithology and precipitation has the strongest explanatory power (14.91%). The models used in this study effectively assess landslide susceptibility and improve analytical accuracy. The combined application of multiple models not only avoids the one-sidedness of a single evaluation method but also provides a methodological reference for landslide disaster assessment and prevention in other regions.

Huixian Karst Wetland, the largest karst wetland inlow-altitude regions of China, plays a crucial role in promoting the healthy circulation of ecosystems, regulating urban climate, and purifying the water quality of the Li River.Diatom indicesis used to evaluate the water ecological health status of HuixianKarst Wetland,and 16 representative sampling sites were established in July 2024 across the surface waters of the Shiziyan Underground River System (SZ), the Ancient Guiliu Canal (GL), the Muzong Lake Dispersed Discharge System Waters (ML), and the Muzong River (MR). The results indicated that: ① A total of 29 genera and 124 species of benthic diatoms were identified in the Huixian Karst Wetland, with three common dominant species across all water bodies. Nutrient-sensitive and eutrophication-preferred species, such asNitzschia palea andMelosira varians, were the primary dominant and indicator species in the wetland. ② Canonicalcorrespondence analysis (CCA) and Mantel test results indicated that total nitrogen (TN) and total phosphorus (TP) were the primary environmental factors influencing changes in the benthic diatom community in the Huixian Karst Wetland, with ammonia nitrogen (NH⁺₄-N), calcium ions (Ca²⁺), dissolved oxygen (DO), and conductivity (Cond) serving as important influencing factors. ③ Specific pollution sensitivity index (IPS), hurlimann trophic index (DI-CH), and trophic diatom index (TDI), selected through cluster analysis, non-metric multidimensional scaling (NMDS), and box plot analysis, can serve as reference biological indicators for water quality assessment in the Huixian Karst Wetland. ④ Comprehensive water quality evaluation based on the three indices revealed that the overall water quality of the Huixian Karst Wetland is poor, with only43.7% of sampling sites rated as “moderate” or “good”,while 56.3% were rated as “poor” or worse. The overall water quality status across different water bodies showed ML > MR ≈ SZ > GL.

Against the backdrop of accelerating urbanization and worsening environmental pollution, understanding the spatiotemporal evolution of ecological quality is crucial for effective ecosystem governance. Taking Yibin City as the study area, this research employs the Google Earth Engine (GEE) platform and landsat imagery to construct greenness, wetness, and heat indicators, while introducing the integrated drought index (IDI) and the remote sensing air quality index (RAQI). An improved remote sensing ecological index, namely thepollution-urbanization-based improved remote sensing ecological index (PBEI), is proposed to better capture the impacts of urbanization and air pollution. To determine index weights, both principal component analysis (PCA) and the entropy weight method (EWM) were compared. The results indicate that PCA is more sensitive to high-variance indicators and susceptible to noise, whereas EWM adaptively assigns weights based on information entropy, effectively enhancing model stability and interpretability. Compared with the traditional remote sensing ecological index (RSEI), themulti-year averageinformation entropy of the EWM-based PBEI increased byapproximately 0.26%, and the contribution rate of the first principal component in the PCA-based PBEI increased by an average of 7.41%, demonstrating its superior ability to characterize impervious surface drought patterns and transitional ecological zones. Spatially, the PCA-based PBEI better highlights areas with intensive pollution and anthropogenic disturbance, while the EWM-based PBEI strengthens the continuous expression of ecological gradients. Comprehensive analyses using the Hurst index, spatial autocorrelation, and Mann-Kendall trend test reveal that Yibin’s ecological quality exhibited a “decline-rise” pattern during 2014-2024, with significant improvement in recent years. These findings validate the applicability of the PBEI in urban ecological monitoring and provide a robust scientific reference for regional ecological assessment and sustainable urban development.

The present study was conducted to investigate the effects of different water and fertiliser regimes on greenhouse gas emissions (CH₄, N₂O, CO₂) from paddy fields. To this end, traditional water and fertiliser management was employed as the control. The experiment involved the cross-pollination of three irrigation patterns (rainfed, shallow, intermittent) with three fertilizer application rates (low, moderate, high) to create an array of interactive treatment combinations. An experiment was conducted over the course of the entire growth period of double-cropped rice as a full-season field experiment. The monitoring of greenhouse gas emission fluxes throughout the rice growth cycle is to be conducted using static chamber-gas chromatography, with the subsequent calculation and determination of their global warming potential and greenhouse gas emission intensity. The findings suggested that the cumulative N₂O emissions under all treatments in this study were lower than those under conventional water and fertiliser management. The most effective strategy for minimising CO₂ emissions through water and fertiliser management involves the implementation of shallow irrigation techniques and the moderation of fertilisation rates. The implementation of intermittent irrigation in conjunction with moderate fertilization resulted in a rice yield of 15.58 t·hm^-2. This approach led to a substantial reduction in CH₄ emissions, achieving the highest emission reduction rate among all treatments at 80.49%. The findings of this study demonstrate that a combination of intermittent irrigation and moderate fertilization has the potential to significantly reduce the global aggregate warming potential and greenhouse gas emission intensity. This approach is identified as the optimal treatment in this study for balancing rice yield with ecological emission reductions.

The stem-leaf angle (SLA) is a key component of cassava plant architecture. In this study, three cassava varieties with different SLAs were used as materials, including NANZHI199 (NZ199, large SLA), HUANAN205 (SC205, medium SLA), and XINXUAN 048 (XX048, small SLA). The effects of planting density on cassava SLA and storage root yield were investigated from the phenotypic and physiological levels. The results showed that: 1) NZ199 had the largest SLA among the three varieties. It also had drooping leaves in the middle and lower canopy, shorter internodes, and more green leaves. Under moderate dense planting (16 815 plants/hm²), its leaves turned upward and SLA decreased, indicating stronger adaptability and self-regulation under dense planting. As planting density increased, plant height, internode length and leaf area index (LAI) of NZ199 increased, whereas stem diameter, number of green leaves and leaf area decreased. NZ199 was greatly affected by dense planting during the storage root formation and expansion stages. 2) Planting density affected the changes in endogenous hormone contents at the base of cassava petioles. Among them, the contents of endogenous hormones such as abscisic acid (ABA), auxin (IAA), and gibberellin (GA) at the petiole base were negatively correlated with the cassava SLA, while the trend of brassinolide (BR) content was the opposite. 3) Withthe increase of planting density,the net photosynthetic rate (Pn), stomatal conductance (Gs) and transpiration rate (Tr) of leaves at different positions in all varieties showed a decreasing trend. However, dry matter yield, storage root yield, total starch yield and non-structural sugar yield at harvest increased.The net photosynthetic rate and chlorophyll content of the upper and middle leaves of NZ199 remained at a high level under the condition of 16 815 plants/hm².Its starch contents in storage roots and stems were significantly higher than those of SC205 and XX048. In conclusion, this study clarifies the effects of planting density on cassava plant architecture, photosynthetic characteristics, and storage root yield and quality from both morphological and physiological perspectives. These findings provide a theoretical basis for improving the production potential of cassava varieties with different plant architectures.

To investigate the regulatory effects of the activator protein PeFOC1 derived from Fusarium oxysporum on safflower seed germination and seedling growth under salt and drought stress, 250 mmol/L NaCl and 10% PEG-6000 were used to simulate salt stress and drought stress conditions, respectively. Treatments included0.1 mg/L (P1), 1 mg/L (P2), and 10 mg/L (P3) of PeFOC1, along with a control group (CK). Parameters such as germination potential, germination index, and seedling growth traits (plant height, root length, stem diameter, fresh weight, and dry weight) were analyzed. The expression of key immune-related genes (CtNPR, CtEDS1) were detected via qRT-PCR. The results demonstrated that comparedwith the untreated stress control group, treatment with 1 mg/LPeFOC1 (P2) significantly enhanced the seed germination potential under both salt and drought stress conditions, with increases of 37.15% and 61.54%, respectively. Meanwhile, P2 treatment markedly promoted root growth, with the root dry weight increasing bymore than 148% under salt stress. Analysis of immune gene expression revealed that under salt stress, 10 mg/L PeFOC1 most significantly up-regulated CtNPR and CtEDS1 (reaching 3.43-fold and 7.7-fold of CK, respectively). In contrast, under drought stress, 1 mg/LPeFOC1 specifically induced high expression of CtNPR and CtEDS1 (reaching 8.7-fold and 5.37-fold of CK, respectively). These findings demonstrate that PeFOC1 coordinately regulates safflower seed germination, seedling physiological adaptation, and immune response through a stress type-dependent concentration strategy, providing new insights and a theoretical basis for stress-resistant cultivation of safflower and the development of biological anti-stress agents.

Phoebe bournei is a precious subtropical tree species. However, the phenotypic and physiological response rules of its seedlings under high temperature, drought and their combined stress are still unclear, which restricts its cultivation regulation and germplasm utilization under extreme climates. One-year-old seedlings bred at Guanzhuang State-owned Forest Farm in Shaxian County, Fujian Province were used as the research objects. Three treatments were established in a phytotron, including 40 ℃ high temperature, 10% PEG-6000 simulated drought, and their combination, with normal growth conditions as the control. Phenotypic changes were recorded, and leaf biochemical indices including chlorophyll a, chlorophyll b, total chlorophyll, and malondialdehyde (MDA) were measured at 0, 6, 12, 24, 48, and 72 hours. The results showed that shoot tips yellowed and wilted at 12 hours. By 72 hours, stems turned brown and leaves became dry and brittle. Chlorophyll a showed the greatest decrease under combined stress. At 72 hours, chlorophyll a decreased by 36.49% compared with that of 0 hours and followed a “rise-first-then-fall” pattern. For membrane lipid peroxidation, MDA peaked at 24 hours under combined stress, increasing by 41.31% compared with that of 0 hours, and then dropped sharply, indicating acute injury followed by cell necrosis and abscission. The soluble protein content and the activities of SOD, POD and CAT all showed a trend of first increase and then decrease,and the peak value under combined stress were significantly higher than that under single stresses. Chlorophylls were significantly positively correlated with antioxidant enzymes and negatively correlated with MDA. The study indicates that,the damage of combined high temperature and drought stress to Phoebe bournei seedlings is significantly greater than that of single stress, which is mainly manifested by reactive oxygen species (ROS) burst, aggravated membrane lipid peroxidation and damage to the antioxidant system.