Beijing’s One Million Mu Plain Afforestation Project involves planting large areas with the exotic North American tree species Fraxinus pennsylvanica Marsh (ash). As an exotic tree species, ash is very vulnerable to infestations by the emerald ash borer (EAB), a native Chinese wood borer pest. In the early stage of an EAB infestation, attacked trees show no obvious sign. Once the stand has reached the late damage stage, death occurs rapidly. Therefore, there is a need for efficient early detection methods of EAB stress over large areas. The combination of unmanned aerial vehicle (UAV)-based hyperspectral imaging (HI) with light detection and ranging (LiDAR) is a promising practical approach for monitoring insect disturbance. In this study, we identified the most useful narrow-band spectral HI data and 3D LiDAR data for the early detection of EAB stress in ash. UAV-HI data of different infested stages (healthy, light, moderate and severe) of EAB in the 400–1000 nm range were collected from ash canopies and were processed by Partial Least Squares–Variable Importance in Projection (PLS-VIP) to identify the maximally sensitive bands. Band R678 nm had the highest PLS-VIP scores and the most robust classification ability. We combined this band with band R776 nm to develop an innovative normalized difference vegetation index (NDVI_(776,678)) to estimate EAB stress. LiDAR data were used to segment individual trees and supplement the HI data. The new NDVI_(776,678) identified different stages of EAB stress, with a producer’s accuracy of 90% for healthy trees, 76.25% for light infestation, 58.33% for moderate infestation, and 100% for severe infestation, with an overall accuracy of 82.90% when combined with UAV-HI and LiDAR.

The long-lasting outbreak of the pine shoot beetle (PSB, Tomicus spp.) threatens forest ecological security. Effective monitoring is urgently needed for the Integrated Pest Management (IPM) of this pest. UAV-based hyperspectral remote sensing (HRS) offers opportunities for the early and accurate detection of PSB attacks. However, the insufficient exploration of spectral and structural information from early-attacked crowns and the lack of suitable detection models limit UAV applications. This study developed a UAV-based framework for detecting early-stage PSB attacks by integrating hyperspectral images (HSIs), LiDAR point clouds, and structure from motion (SfM) photogrammetry data. Individual tree segmentation algorithms were utilized to extract both spectral and structural variables of damaged tree crowns. Random forest (RF) was employed to determine the optimal detection model as well as to clarify the contributions of the candidate variables. The results are as follows: (1) Point cloud segmentation using the Canopy Height Model (CHM) yielded the highest crown segmentation accuracy (F-score: 87.80%). (2) Near-infrared reflectance exhibited the greatest decrease for early-attacked crowns, while the structural variable intensity percentile (int_P50-int_P95) showed significant differences (p < 0.05). (3) In the RF model, spectral variables were predominant, with LiDAR structural variables serving as a supplement. The anthocyanin reflectance index and int_kurtosis were identified as the best indicators for early detection. (4) Combining HSI with LiDAR data obtained the best RF model accuracy (classification accuracy: 87.31%; Kappa: 0.8275; SDR estimation accuracy: R² = 0.8485; RMSE_cv = 3.728%). RF integrating HSI and SfM data exhibited similar performance. In conclusion, this study identified optimal spectral and structural variables for UAV monitoring and improved HRS model accuracy and thereby provided technical support for the IPM of PSB outbreaks.

The outbreak of Pine Shoot Beetle (PSB, Tomicus spp.) posed a significant threat to the health of Yunnan pine forests, necessitating the development of an efficient and accurate remote sensing monitoring method. The integration of unmanned aerial vehicle (UAV) imagery and deep learning algorithms shows great potential for monitoring forest-damaged trees. Previous studies have utilized various deep learning semantic segmentation models for identifying damaged trees in forested areas; however, these approaches were constrained by limited accuracy and misclassification issues, particularly in complex forest backgrounds. This study evaluated the performance of five semantic segmentation models in identifying PSB-damaged trees (UNet, UNet++, PAN, DeepLabV3+ and FPN). Experimental results showed that the FPN model outperformed the others in terms of segmentation precision (0.8341), F1 score (0.8352), IoU (0.7239), mIoU (0.7185) and validation accuracy (0.9687). Under the pure Yunnan pine background, the FPN model demonstrated the best segmentation performance, followed by mixed grassland-Yunnan pine backgrounds. Its performance was the poorest in mixed bare soil-Yunnan pine background. Notably, even under this challenging background, FPN still effectively identified diseased trees, with only a 1.7% reduction in precision compared to the pure Yunnan pine background (0.9892). The proposed method in this study contributed to the rapid and accurate monitoring of PSB-damaged trees, providing valuable technical support for the prevention and management of PSB.

National and provincial surveillance of pine wood-boring pests in China is designed to detect damaging taxa, map occurrence, assess risk and loss, and support early warning, zoning and control decisions. Province-scale comparisons of alternative monitoring devices remain rare, especially under the operational conditions required by such programs in climatically and topographically heterogeneous forests. Using data from the 2025–2026 systematic survey of pine wood-boring pests in Yunnan Province, China, we integrated several monitoring datasets to compare baited logs, lure traps, and flight-intercept traps. The harmonized database comprised 2603 standard monitoring subcompartments and 3519 installed sites, including 4080 baited-log piles, 4807 lure-trap units, and 373 flight-intercept traps. Main performance analyses focused on active sites with at least one collection event (570 baited-log sites, 63 flight-intercept sites, and 496 lure-trap sites), whereas installed site summaries were retained to characterize operational coverage. Because the study was observational and the three devices have different sampling mechanisms, we interpreted detection probability as the primary early warning metric, and catch, operational taxon richness, standardized yield, and cost metrics as supporting indicators of diagnostic and operational return. Site-level comparisons were complemented with paired analyses of 21 co-located subcompartments, a more comparable subset defined within county × elevation band × host group strata represented by all three methods, county-clustered regression, and a taxonomic-resolution sensitivity analysis. Lure traps consistently had the highest detection probability (0.73), the greatest cumulative catch (8617 individuals), and the broadest operational taxonomic coverage (45 operational taxa). In county-clustered models, lure traps had higher odds of detection (odds ratio = 11.25, 95% CI: 5.64–22.43) and higher catch rates (incidence rate ratio = 5.97, 95% CI: 2.26–15.76) than baited logs after adjustment for elevation band, host group, and collection effort. The same ranking persisted in the more comparable subset and after exclusion of unresolved family-, subfamily-, genus-, and unknown-level records. Standardized yield peaked at 1500–2200 m. Scenario-based costing showed that lure traps had the lowest cost per captured and resolved captured individual, whereas detection cost estimates were interpreted together with absolute detections and operational taxonomic output. Overall, the results support a tiered surveillance architecture in which lure traps serve as the primary routine early warning tool, baited logs provide targeted complementary information, and flight-intercept traps are reserved mainly for exploratory or faunistic surveys.

Acoustic detection technology has emerged as a promising, non-destructive and continuous monitoring method for pest early detection at the single tree level. However, field application still encounters problems, especially under complex infestation scenarios, i.e., co-infestations by multiple pest species. This study aims to develop a novel acoustic-based recognition model for detecting forest wood-boring pests, specially designed to enhance monitoring accuracy under complex infestation scenarios. We collected feeding vibration signals from four wood-boring pests: Semanotus bifasciatus, Phloeosinus aubei, Agrilus planipennis, and Streltzoviella insularis. Three infestation scenarios were designed: single-species, co-infestation without mixed signals, and co-infestation with mixed signals. Three machine learning (ML) models (Random Forest, Support Vector Machine, and Artificial Neural Network) based on seven acoustic feature variables, and three deep learning (DL) models (AlexNet, ResNet, and VGG) using spectrograms were employed to classify the signals. Results showed that ML models achieved perfect accuracy (OA: 100%, Kappa: 1) in single-species scenarios but declined significantly under co-infestation scenarios with mixed signals. In contrast, DL models, particularly ResNet, maintained high accuracy (OA: 85.0–88.75%) and effectively discriminated mixed signals. In conclusion, this study demonstrates the superiority of spectrogram-based DL models for acoustic detection under complex infestation scenarios and provides a foundation for developing a general, real-time detection model for integrated pest management in forest ecosystems.

Climate change may alter grassland pest distributions, yet province-wide comparisons based on extensive field surveys remain limited in Yunnan, China. We analysed 10,342 census and special-survey records, retaining 4281 species–climate-cell records for nine pests. Five WorldClim predictors entered ten-algorithm ensembles weighted by the true skill statistic (TSS), evaluated using nested spatial cross-validation. Three climate models projected SSP1-2.6, SSP2-4.5, and SSP5-8.5 across four 20-year periods spanning 2021–2100. Predictive performance varied among species: area under the receiver operating characteristic curve (AUC) ranged from 0.591 to 0.778, and TSS from 0.198 to 0.365. Pseudoxya diminuta had the highest AUC (0.778), Patanga japonica the highest TSS (0.365), and the 27-cell Holotrichia parallela model the lowest AUC (0.591). Warm-quarter precipitation ranked first for six species, with mean permutation-induced AUC decreases of 0.054–0.145. Baseline threshold-defined suitable area was largest for Oxya chinensis (235.3 × 10³ km²). By 2081–2100 under SSP5-8.5, consensus suitable area declined by 63.4% for P. japonica and 54.6% for P. diminuta. These species-specific climatic associations and projected changes provide a basis for coordinated provincial surveillance, supported by survey-derived county–township revisit priorities.