Multi-graph spatial–temporal graph convolutional networks for predicting the spread of Dendroctonus valens in China

Abstract

The red turpentine beetle (Dendroctonus valens) is the most widespread pine pest in North America. Since its invasion into China in 1998, it has killed over 10 million pine trees, causing severe ecological and economic losses, and has been designated as a high-priority managed invasive species. Nevertheless, characterizing its spatiotemporal spread remains challenging, as fundamental processes of growth, dispersal, and establishment are governed by nonlinear responses to environmental conditions. To enable precise prevention and control, this study first analyzes the historical spatiotemporal patterns of D. valens in China and identifies the key factors influencing its spread. Furthermore, we propose a spatiotemporal graph convolutional network-based risk prediction framework (MG-STGCN) for D. valens, integrating multi-dimensional graph structures with a spatial attention structure (SAS) and a gated recurrent unit (GRU) to generate county-level risk predictions across China. The predicted risk distribution exhibits a clear spatial pattern concentrated east of the Hu Huanyong Line, forming a continuous belt across Northeast and North China. High-risk areas are primarily clustered in southern Northeast China and adjacent regions of North China, while only limited and scattered risk signals appear in western transitional zones. Compared with the current distribution, the overall high-risk clusters remain relatively stable, but a slight northeastward shift can be observed, accompanied by emerging signals of westward penetration. The proposed MG-STGCN achieves the best performance among all compared models, with a recall of 89.57%, precision of 90.68%, and an F1-score of 90.12%. The improvements are mainly attributed to the complementary effects of the temporal modeling module and the spatial attention mechanism, which enhance the detection of outbreak signals while improving the discrimination of non-outbreak regions under highly imbalanced conditions. Attention-based analysis reveals that pest spread is dominated by local spatial continuity, with limited long-distance interactions. Different graph structures capture complementary transmission mechanisms, where spatial and wind-driven graphs enhance sensitivity to spread, while host-related connectivity improves specificity by constraining ecologically feasible pathways, resulting in a balanced predictive performance. This framework provides a reliable tool for large-scale pest risk assessment, and supporting more targeted surveillance and management strategies.

Graphical abstract

Keywords

Dendroctonus valens / Spatiotemporal dynamic / Multi-graph / Spatial–temporal graph convolutional networks / Forest protection

References

[1] Anderegg WRL, Hicke JA, Fisher RA, Allen CD, Aukema JE, Bentz B, Hood S, Lichstein JW, Macalady AK, McDowell NG, Pan Y, Raffa K, Sala A, Shaw JD, Stephenson NL, Tague CL, Zeppel M. Tree mortality from drought, insects, and their interactions in a changing climate. New Phytol, 2015, 208(3): 674-683.

[2] Biondi M, D’Alessandro P, De Simone W, Iannella M. DBSCAN and GIE, two density-based “grid-free” methods for finding areas of endemism: a case study of flea beetles (Coleoptera, Chrysomelidae) in the Afrotropical region. InSects, 2021, 12(12. ArticleID: 1115

[3] Bjørnstad ON, Peltonen M, Liebhold AM, Baltensweiler W. Waves of larch budmoth outbreaks in the European Alps. Science, 2002, 298(5595): 1020-1023.

[4] Chapman JW, Reynolds DR, Wilson K. Long-range seasonal migration in insects: mechanisms, evolutionary drivers and ecological consequences. Ecol Lett, 2015, 18(3): 287-302.

[5] Chen P, Shen C, Tao Z, Qin W, Huang W, Siemann E. Deterministic responses of biodiversity to climate change through exotic species invasions. Nat Plants, 2024, 10: 1464-1475.

[6] Chen T, Guestrin C (2016) XGBoost: a scalable tree boosting system. In: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD’16), pp. 785–794.https://doi.org/10.1145/2939672.2939785

[7] Cho K, Van Merriënboer B, Gulcehre C, Bahdanau D, Bougares F, Schwenk H, Bengio Y (2014) Learning phrase representations using RNN encoder–decoder for statistical machine translation. Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP): 1724–1734. https://doi.org/10.3115/v1/D14-1179

[8] Cutler DR, Edwards TCJr, Beard KH, Cutler A, Hess KT, Gibson J, Lawler JJ. Random forests for classification in ecology. Ecology, 2007, 88(11): 2783-2792.

[9] de Lima CL, da Silva ACG, Moreno GMM, da Silva CC, Musah A, Aldosery A, Dutra L, Ambrizzi T, Borges IVG, Tunali M. Temporal and spatiotemporal arboviruses forecasting by machine learning: a systematic review. Front Public Health, 2022, 10: 90007.

[10] Ding C, Sun S, Zhao J. An ELICIT information-based ORESTE method for failure mode and effect analysis considering risk correlation with GRA-DEMATEL. Inf Fusion, 2023, 89: 527-536.

[11] Dormann CF, Schymanski SJ, Cabral J, Chuine I, Graham C, Hartig F, Kearney M, Morin X, Römermann C, Schröder B, Singer A. Correlation and process in species distribution models: bridging a dichotomy. J Biogeogr, 2012, 39(12): 2119-2131.

[12] Early R, Bradley BA, Dukes JS, Lawler JJ, Olden JD, Blumenthal DM, Gonzalez P, Grosholz ED, Ibañez I, Miller LP. Global threats from invasive alien species in the twenty-first century and national response capacities. Nat Commun, 2016, 7(1. ArticleID: 12485

[13] Elith J, Leathwick JR. Species distribution models: ecological explanation and prediction across space and time. Annu Rev Ecol Evol Syst, 2009, 40(1): 677-697.

[14] EPPO Reporting Service (2019) https://gd.eppo.int/reporting/article-6529 [accessed on 05.2019].

[15] Erbilgin N, Mori SR, Sun JH, Stein JD, Owen DR, Merrill LD, Campos Bolaños R, Raffa KF, Méndez Montiel T, Wood DL, Gillette NE. Response to host volatiles by native and introduced populations of Dendroctonus valens (Coleoptera: Curculionidae, Scolytinae) in North America and China. J Chem Ecol, 2007, 33(1): 131-146.

[16] Escobar LE, Craft ME. Advances and limitations of disease biogeography using ecological niche modeling. Front Microbiol, 2016, 7. ArticleID: 1174

[17] Evans TG, Diamond SE, Kelly MW. A replicated climate change field experiment reveals rapid evolutionary response in an ecologically important soil invertebrate. Glob Change Biol, 2016, 22(7): 2215-2226.

[18] Fettig CJ, Klepzig KD, Billings RF, Munson AS, Nebeker TE, Negrón JF, Nowak JT. Reconstruction of fire history (1680–2003) in Gaspesian mixedwood boreal forests of eastern Canada. For Ecol Manage, 2007, 238(1-3): 24-53.

[19] Fritz C, Dorigatti E, Rügamer D. Combining graph neural networks and spatio-temporal disease models to improve the prediction of weekly COVID-19 cases in Germany. Sci Rep, 2022, 12(1. ArticleID: 3930

[20] Gao B, Yu L, Ren L, Zhan Z, Luo Y. Early detection of Dendroctonus valens infestation with machine learning algorithms based on hyperspectral reflectance. Remote Sens, 2022, 14(6): 1373.

[21] González-Hernández A, Morales-Villafaña R, Romero-Sánchez ME, Islas-Trejo B, Pérez-Miranda R. Modelling potential distribution of a pine bark beetle in Mexican temperate forests using forecast data and spatial analysis tools. J for Res, 2018, 31(2): 649-659.

[22] Guo S, Lin Y, Feng N, Song C, Wan H (2019) Attention based spatial-temporal graph convolutional networks for traffic flow forecasting. In: Proceedings of the AAAI Conference on Artificial Intelligence 33(01): 922–929. https://doi.org/10.1609/aaai.v33i01.3301922

[23] He SY, Ge XZ, Wang T, Wen JB, Zong SX. Areas of potential suitability and survival of Dendroctonus valens in China under extreme climate warming scenario. Bull Entomol Res, 2015, 105(4): 477-484.

[24] Hou Z, Dong Y, Shi F, Xu Y, Ge S, Tao J, Ren L, Zong S. Seasonal shifts in cold tolerance and the composition of the gut microbiome of Dendroctonus valens LeConte occur concurrently. Forests, 2021, 12(7): 888.

[25] Hu H, Zhou H, Wang C, Li Y, Li Y, Fang G. Quantitative analysis of the spread pattern of pine wilt disease from the Yangtze River Basin in China. Pest Manag Sci, 2026, 82: 6758-6774.

[26] Jarnevich CS, Stohlgren TJ, Kumar S, Morisette JT, Holcombe TR. Caveats for correlative species distribution modeling. Eco Inform, 2015, 29: 6-15.

[27] Jia X, Willard J, Karpatne A, Read J, Zwart J, Steinbach M, Kumar V. Physics-guided machine learning for scientific discovery. Nat Rev Earth Environ, 2020, 1: 259-271.

[28] Jones KL, Shegelski VA, Marculis NG, Wijerathna AN, Evenden ML. Factors influencing dispersal by flight in bark beetles (Coleoptera: Curculionidae: Scolytinae): from genes to landscapes. Can J for Res, 2019, 49(9): 1024-1041.

[29] Kuhn M, Johnson K. Applied Predictive Modeling, 2013. New York, Springer.

[30] Kuo FY, Wen TH, Sabel CE. Characterizing diffusion dynamics of disease clustering: a modified space–time DBSCAN (MST-DBSCAN) algorithm. Ann Assoc Am Geogr, 2018, 108(4): 1168-1186.

[31] Li P, Zhang T, Jin Y. A spatio-temporal graph convolutional network for air quality prediction. Sustain, 2023, 15(9): 7624.

[32] Li X, Zhang Y, Huang J, Fang G, Liu D. Spatiotemporal forecasting of pine wilt disease in China using a graph convolutional-LSTM network. Ecol Indic, 2023, 147: 109939

[33] Li J, Zhang B, Jiang J, Mao Y, Li K, Liu F. Machine learning provides insights for spatially explicit pest management strategies by integrating information on population connectivity and habitat use in a key agricultural pest. Pest Manag Sci, 2024, 80(10): 4871-4882.

[34] Liu Y, Gao B, Bian L, Ren L, Luo Y. Invasion of Red Turpentine Beetles led to the increase of native trunk-boring beetles in Chinese pine stands. For Ecol Manage, 2024, 557. ArticleID: 121758

[35] Lu X, Huang J, Li X, Fang G, Liu D. The interaction of environmental factors increases the risk of spatiotemporal transmission of pine wilt disease. Ecol Ind, 2021, 133. ArticleID: 108394

[36] Lu X, Huang J, Li X, Fang G, Liu D. The interaction of environmental factors increases the risk of spatiotemporal transmission of pine wilt disease. Ecol Indic, 2021, 132. ArticleID: 108394

[37] Mo X, Zhao X, Zhao J, Huang J, Fang G. Risk prediction of pine wilt disease based on graphical convolutional network in China. Pest Manag Sci, 2025, 81(10): 6550-6559.

[38] Munro HL, Montes CR, Gandhi KJK. A new approach to evaluate the risk of bark beetle outbreaks using multi-step machine learning methods. For Ecol Manage, 2022, 520. ArticleID: 120347

[39] Ning H, Tang M, Chen H. Impact of climate change on potential distribution of Chinese White Pine Beetle Dendroctonus armandi in China. Forests, 2021, 12(5): 544.

[40] Poland TM, Rassati D. Improved biosecurity surveillance of non-native forest insects: a review of current methods. J Pest Sci, 2019, 92(1): 37-49.

[41] Raffa KF, Aukema BH, Bentz BJ, Carroll AL, Hicke JA, Turner MG, Romme WH. Cross-scale drivers of natural disturbances prone to anthropogenic amplification: the dynamics of bark beetle eruptions. Bioscience, 2008, 58(6): 501-517.

[42] Reichstein M, Camps-Valls G, Stevens B, Jung M, Denzler J, Carvalhais N, Prabhat, . Deep learning and process understanding for data-driven Earth system science. Nature, 2019, 566: 195-204.

[43] Saito T, Rehmsmeier M. The precision-recall plot is more informative than the ROC plot when evaluating binary classifiers on imbalanced datasets. PLoS ONE, 2015, 10(3. ArticleID: e0118432

[44] Smith RH. Red turpentine beetle, 1971. Washington, DC, U.S. Department of Agriculture, Forest Service

[45] Sun JH, Miao ZW, Zhang ZN, Zhang ZN, Gillette NE. Red turpentine beetle, Dendroctonus valens LeConte (Coleoptera: Scolytidae), response to host semiochemicals in China. Environ Entomol, 2004, 33(2): 206-212.

[46] Sun JH, Lu M, Gillette NE, Wingfield MJ. Red turpentine beetle: innocuous native becomes invasive tree killer in China. Annu Rev Entomol, 2013, 58: 293-311.

[47] Taerum SJ, Duong TA, de Beer ZW, Gillette N, Sun JH, Owen DR, Wingfield MJ. Large shift in symbiont assemblage in the invasive red turpentine beetle. PLoS ONE, 2013, 8(10. ArticleID: e78126

[48] Turbelin AJ, Cuthbert RN, Essl F, Haubrock PJ, Ricciardi A, Courchamp F. Biological invasions are as costly as natural hazards. Perspect Ecol Conserv, 2023, 21: 143-150.

[49] Wang HB, Zhang Z, Kong XB, Liu SC, Shen ZR. Preliminary deduction of potential distribution and alternative hosts of invasive pest, Dendroctonus valens (Coleoptera: Scolytidae). Sci Silvae Sin, 2007, 43(10): 71-76.

[50] Wang Z, Liang L, Wang HM, Decock C, Lu Q. Ophiostomatoid fungi associated with Ips bark beetles in China. Fungal Divers, 2024, 129(1): 283-364.

[51] Wang X, Cao JZ, Zhao TH, Zhang B, Chen GZ, Li ZH, Chen HL, Tu W, Li QQ. ST-Camba: a decoupled-free spatiotemporal graph fusion state space model with linear complexity for efficient traffic forecasting. Inf Fusion, 2025, 126. ArticleID: 103495

[52] Wu Z, Pan S, Long G, Jiang J, Chang X, Zhang C (2020) Connecting the dots: Multivariate time series forecasting with graph neural networks. In: Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining (KDD ’20), pp. 753–763. https://doi.org/10.1145/3394486.3403118

[53] Xiao YF, Wang MJ, Li XY, Chen F, Wang JS, Zhang HW, Cidan P. Analysis on the potential distribution area of Dendroctonus micans in Tibet. Forest Grassl Resour Res, 2020, 11(2): 141-145.

[54] Xu W, Xiao Y, Zhang J, Yang W, Zhang L, Hull V, Wang Z, Zheng H, Liu J, Polasky S. Strengthening protected areas for biodiversity and ecosystem services in China. Proc Natl Acad Sci U S A, 2017, 114(7): 1601-1606.

[55] Yan ZL, Sun JH, Don O, Zhang ZN. The red turpentine beetle, Dendroctonus valens LeConte (Scolytidae): an exotic invasive pest of pine in China. Biodivers Conserv, 2005, 14(7): 1735-1760.

[56] Yang ZQ, Wang XY, Zhang YN. Recent advances in biological control of important native and invasive forest pests in China. Biol Control, 2014, 68: 117-128.

[57] Yu B, Yin H, Zhu Z (2018) Spatio-temporal graph convolutional networks: a deep learning framework for traffic forecasting. In: Proceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence (IJCAI), pp. 363–3640. https://doi.org/10.24963/ijcai.2018/505

[58] Zhan Z, Yu L, Ren L, Liu Y, Lu Z, Luo Y. Infestation patterns of incipient red turpentine beetle populations in fire-affected, logged and undisturbed forest stands of northern China. For Ecol Manage, 2022, 521. ArticleID: 120424

[59] Zhang LY, Chen QC, Zhang X. Studies on the morphological characters and bionomics of Dendroctonus valens LeConte. Sci Silvae Sin, 2002, 38(4): 95-99.

[60] Zhang J, Zheng Y, Qi D. Deep spatio-temporal residual networks for citywide crowd flows prediction. Proceed AAAI Confer Artific Intell, 2017, 31(1): 1655-1661.

[61] Zhang X, Zhang X, Liu J, Wu B, Hu Y. Graph features dynamic fusion learning driven by multi-head attention for large rotating machinery fault diagnosis with multi-sensor data. Eng Appl Artif Intell, 2023, 125. ArticleID: 106601

[62] Zhao JX, Yang ZQ, Ren XH, Liang XM. Biological characteristics and occurring law of Dendroctonus valens in China. Sci Silvae Sin, 2008, 44(2): 99-105.

[63] Zhou Y, Guo S, Wang T, Zong S, Ge X. Modeling the pest-pathogen threats in a warming world for the red turpentine beetle (Dendroctonus valens) and its symbiotic fungus (Leptographium procerum). Pest Manag Sci, 2024, 80(7): 3423-3435.

[64] Zhou HW, Ji HW, Wu YX, Zhao P. Improve the detection model of tree decay by Dioryctria based on the YOLOv8. For Eng, 2025, 41(1): 126-137.

Funding

National Key Research and Development Program of China(2023YFC2604801)

RIGHTS & PERMISSIONS

The Author(s)