Data for D. punctatus and P. massoniana were obtained from the Center for Agriculture and Bioscience International (CABI: https://www.cabi.org), the Global Biodiversity Information Facility (GBIF: https://www.gbif.org/), the Chinese Virtual Herbarium (CVH: https://www.cvh.ac.cn/), and relevant published literature.

To avoid autocorrelation among samples, we further processed the distribution information of the target species. Initially, the Baidu Maps coordinate picker system was used to acquire precise longitudes and latitudes for collection points missing this information. The distribution point data in degrees, minutes, and seconds format were then converted into decimal (floating-point) numbers. Finally, to reduce spatial autocorrelation between sample points and decrease the errors in the model’s outcomes, we established a 5 km × 5 km grid (corresponding to the 2.5 arc-minute environmental data detailed in Section 2.2), retaining only one sampling point per grid and eliminating duplicate and sea-based points. The final number of sample points used for building the species distribution model was 429 for D. punctatus and 236 for P. massoniana ( Figure 1 ).

We selected 19 bioclimatic variables (BIO01-BIO19, Table 1 ) from the WorldClim database (version 2.1, spanning 1970–2000) (http://www.worldclim.org/) as the current climate data, with a spatial resolution of 2.5 arc minutes. To reduce multicollinearity among variables and prevent model overfitting, it was necessary to filter these 19 environmental variables (Morales et al., 2017). Initially, we imported the distribution data of the target species and the 19 environmental factors into MaxEnt, utilizing default parameters for pre-training, and a jackknife method to determine the contribution rate of each environmental factor ( Table 1 ). We subsequently conducted a correlation analysis of the 19 environmental factors using ArcGIS (Esri) ( Figure 2 ). If two or more environmental variables exhibited a high correlation (|r| > 0.8), we retained the variable with a higher contribution to the model and excluded those with a contribution rate of zero. Ultimately, six climatic variables were selected for modeling the potential distribution of the pine caterpillar and pine, namely, isothermality (BIO03), the minimum temperature of the coldest month (BIO06), the annual temperature range (BIO07), the mean temperature of the coldest quarter (BIO11), annual precipitation (BIO12), and the precipitation of driest month (BIO14).

Three global circulation models (GCMs) from the Sixth Coupled Model Intercomparison Project (CMIP6), namely, BCC-CSM2-MR (Beijing Climate Center Climate System Model), MIROC6 (Model for Interdisciplinary Research on Climate), and CMCC-ESM2 (Centro Euro-Mediterraneo sui Cambiamenti Climatici Earth System Model 2) were selected as the future climate models. To mitigate the uncertainty associated with reliance on a single GCM, we averaged the occurrence probabilities of these three models. Based on this, we employed the Shared Socioeconomic Pathways (SSPs) SSP1-2.6 and SSP5-8.5, representing sustainable and business-as-usual development scenarios (Riahi et al., 2017; Liu et al., 2021), respectively, applied to the years 2041–2060 (2050s) and 2061–2080 (2070s). These choices of models and SSPs aimed to explore the potential distribution changes of the pine caterpillar and pine under different future development pathways.

Aligned with our research objectives and the foundational principles of the MaxEnt model, we developed a framework to simulate the potential distribution of D. punctatus and P. massoniana and evaluate the spatiotemporal variations in their future suitable habitats under the influence of climate change. The framework is segmented into four key sections: (1) data collection; (2) data preprocessing; (3) optimization of the MaxEnt model; and (4) mapping of suitable habitats and analysis of the results ( Figure 3 ). In the first section, we compiled occurrence records for D. punctatus and P. massoniana, along with current and future environmental variables that influence their distribution. The second section involved filtering the occurrence records to remove duplicates and samples that did not fit within the range of environmental variables. We also reduced redundancy in the environmental data by analyzing the contributions and correlations of environmental variables. In the third section, two critical parameters of the MaxEnt model were optimized to enhance the accuracy of the predictions. Finally, in the fourth section, using the processed data and the optimized MaxEnt model, we simulated the potential distribution areas of both species and predicted their future distributions. We then conducted an analysis and discussion of the prediction results and key environmental factors.

During construction of the model, 75% of the geographic distribution data was used for model training, and the remaining 25% was used for model validation. To enhance reliability of the results, we established 10 replicates and employed bootstrapping as the sampling method. The Logistic function was used as the output format, and the final results represent the average of the 10 replicates. We subsequently adopted the jackknife method to assess the relative importance of each climatic factor within the suitable areas for D. punctatus and P. massoniana, thus identifying the key limiting factors affecting the distribution of these two species (Phillips et al., 2006). The accuracy of the model was evaluated by determining the area under the curve (AUC) of the receiver operating characteristic (ROC) curve. The AUC value is not influenced by specific thresholds, hence is widely used to assess the accuracy of predictive models. The AUC ranges from 0 to 1 and is directly proportional to the accuracy of a model. In general, AUC values less than 0.7 indicate poor model performance; values between 0.7 and 0.8 denote moderate performance; values between 0.8 and 0.9 indicate good performance; and values greater than 0.9 suggest excellent performance. The closer the AUC value is to 1, the better the predictive performance of the model (Zhao et al., 2021).

The MaxEnt model outputs the probability of species presence (p) in each grid cell in ASCII format, with values ranging from 0 to 1. We visualized the model’s output in ArcGIS and employed the reclassification tool to categorize suitability levels, calculating the corresponding area for each category. Based on the suitability index, the suitable areas were divided into four levels: unsuitable (p< 0.1), low suitability (0.1< p< 0.3), moderate suitability (0.3< p< 0.5), and high suitability (p > 0.5) (Yan et al., 2021).

We defined the suitable areas (assigned a value of 1) for D. punctatus and P. massoniana as spatial units with p > 0.1 and areas with p< 0.1 as unsuitable (assigned a value of 0) (Zhao et al., 2021). Based on this principle, we established potential distribution matrices for these species under current and future climate change scenarios, where 0 indicates absence and 1 indicates presence. This approach allowed us to analyze the spatial pattern changes in suitable distribution areas under future climate scenarios. Changes in areas for 2050 and 2070 were calculated based on current and projected distributions for 2050. Moreover, we defined four types of changes: newly suitable areas (matrix value changing from 0 to 1); lost suitable areas (from 1 to 0); retained suitable areas (1 remaining constant); and consistently unsuitable areas (0 remaining constant).

To further analyze the dynamics of species distribution, we simplified the distribution of D. punctatus and P. massoniana to a single centroid point and created a vector file to describe the magnitude and direction of changes in the suitable areas over time. By tracking the centroid changes under different climate scenarios, we could explore the dynamics of species distribution. Furthermore, we calculated the cosine similarity of the changes in the distribution centroids of D. punctatus and P. massoniana under the same scenarios to determine whether the migration directions of these two species were similar.

The MaxEnt model employs the ROC curve to assess the accuracy of the analytical results. The ROC curve is a tool for evaluating the performance of classification models, plotting the false positive rate on the x-axis against the true positive rate on the y-axis. The area between the curve and the x-axis, denoted as the AUC, is used to quantify the overall performance of the model. The average AUC values (obtained through 10 replicate runs of the distribution models) for D. punctatus and P. massoniana were 0.931 and 0.923, respectively, indicating that both models exhibit excellent performance, and their predictive accuracy is reliable. Furthermore, the high AUC values indicate the efficiency of the models in differentiating between suitable and unsuitable areas, thereby providing robust scientific support for our research.

We determined the key environmental variables influencing the species distribution by analyzing the percentage contributions of various environmental variables within the predictive model. Six environmental variables were selected for analysis ( Figure 4 ). Among these variables, the minimum temperature of the coldest month (bio06), annual temperature range (bio07), and annual precipitation (bio12) collectively contributed to 81.4% and 84.6% of the overall impact of D. punctatus and P. massoniana, respectively. This highlights the significance of these variables as the three main environmental driving factors in the distribution of D. punctatus and P. massoniana. The remaining contribution, accounting for less than 20%, was jointly provided by isothermality (bio03), the mean temperature of the coldest quarter (bio11), and the precipitation of the driest month (bio14).

Figure 5 reveals that D. punctatus and P. massoniana are primarily distributed in regions south of 35°N latitude in China, consistent with existing literature. The distribution areas of both species exhibit significant similarity, with their northern boundaries aligning with the Qinling Mountains and the Huai River, and their southern boundaries extending to Hainan Island. The total area of their distribution is approximately 193.98 × 10⁴ km² for D. punctatus and 191.43 × 10⁴ km² for P. massoniana, respectively. The high-suitability areas for D. punctatus are mainly concentrated in Hunan and Jiangxi Provinces, followed by Chongqing, Hubei, Henan, Anhui, Jiangsu, Zhejiang, Fujian, and Guangdong Provinces and Guangxi Zhuang Autonomous Region. In contrast, the high-suitability areas for P. massoniana are more extensive, stretching eastward to the coastal regions of China, covering the central part of Guangdong Province and the southwestern region of Guangxi Zhuang Autonomous Region, as well as the eastern part of Guizhou Province.

The impact of climate change on the distribution of D. punctatus and P. massoniana may increase the risk of pest outbreaks. Figure 6 presents the projected distribution of suitable habitats for these species under different future climate scenarios. Predictions indicate that the distribution range of both species will continue to expand in the coming decades. Particularly under the future SSP1-2.6 scenario ( Figures 6A, C ), the suitable habitat of D. punctatus is expected to expand north-eastward to northern Shandong, with the suitable habitat area reaching 224.68 × 10⁴ km² in 2050 and 235.75 × 10⁴ km² in 2070. Under the SSP5-8.5 scenario ( Figures 6B, D ), the north-eastward expansion trend of D. punctatus is even more pronounced, with the suitable habitat area estimated at 240.02 × 10⁴ km² in 2050 and 267.78 × 10⁴ km² in 2070. Moreover, under both climate scenarios, the high-suitability area for D. punctatus markedly increases, suggesting that future climatic conditions may be more conducive to the survival of this species and could increase the risk of outbreaks. In contrast, the northward expansion of P. massoniana is slower ( Figures 6E–H ), but the high-suitability area for this species continues to increase under both climate scenarios, indicating that southern regions of China will become more suitable for the growth of P. massoniana. However, the suitable habitat of D. punctatus almost entirely overlaps with the suitable area for P. massoniana, suggesting that P. massoniana may continue to suffer from the pest in the future.

Environmental variables are widely acknowledged as key factors affecting species distribution patterns. These variables impact the growth, development, and interspecies interactions of species (Raza et al., 2015). Future climate change is anticipated to markedly affect the distribution of D. punctatus and P. massoniana. In our study, the primary environmental variables influencing the distribution of D. punctatus and P. massoniana are identified as the minimum temperature of the coldest month (bio06), annual temperature range (bio07), and annual precipitation (bio12). Among these factors, those related to temperature contribute more significantly, indicating a higher sensitivity of D. punctatus and P. massoniana to temperature variations. Concurrently, annual precipitation is also instrumental in the spatial distribution modeling of these two species. This finding aligns with conclusions drawn from physiological and ecological studies of D. punctatus and P. massoniana (Zeng et al., 2010; Lei and Wang, 2024). The development of D. punctatus requires a certain amount of accumulated temperature. In the distribution areas of this species, from north to south, the number of generations completed per year increases with the rise in average annual temperature. Precipitation influences the occurrence of pest outbreaks by altering air humidity and through the washing effect on the larvae of D. punctatus. P. massoniana, preferring a light-abundant and deep-rooted environment, thrives in warm and moist climates and is typically found in regions with distinct seasons and concurrent periods of heat and rainfall (Fei et al., 2014; Wang et al., 2016; Yan et al., 2019). The distribution area of D. punctatus lies within China’s subtropical monsoon and tropical monsoon climate zones, characterized by hot summers, mild winters, minimum average temperatures above 0°C in the coldest months, and annual rainfall ranging from 1000 to 2000 mm. This region predominantly features P. massoniana as its representative vegetation. Suitable climatic conditions and extensive P. massoniana forests provide favorable conditions for outbreaks of D. punctatus. In addition to temperature and precipitation, environmental factors such as elevation, solar radiation intensity, predation competition, and extent of vegetation cover significantly influence the distribution of insects and plants (Li et al., 2023). For example, during the hatching period of the D. punctatus, larvae often disperse with the wind, with their direction of spread being influenced by the wind direction. The dispersal process is affected by wind force, wind speed, and topography (Chen, 1990). Soil pH values can impact the distribution of P. massoniana, a species that prefers acidic soils and is intolerant to saline conditions. Appropriate climatic and soil environments are essential for the formation of P. massoniana forests. Changes in P. massoniana forests can also affect the distribution patterns of the D. punctatus. However, integrating all influencing factors into a single model to simulate the potential distribution of species is a challenging task. Moreover, introducing too many variables may lead to increased multicollinearity issues, diminishing the impact of key variables. Nonetheless, the predictions of future suitable habitat migration changes in this study are consistent with the growth habits of D. punctatus and P. massoniana, and thus our results are a valuable reference for potential suitability forecasts of these two species under the context of climate change.

Under all future climate scenarios, the total distribution areas of D. punctatus and P. massoniana are projected to increase to varying degrees compared to the present, generally showing a trend of northward expansion ( Figures 7 , 8 ). By 2070, under the SSP5-8.5 scenario, predictions indicate that the suitable habitat areas for these two species will reach their maximum. Numerous studies have highlighted that climate change may cause significant shifts in species distribution patterns. Under the SSP1-2.6 scenario, by 2050, the new areas for D. punctatus are expected to include central Shandong, Henan, parts of Yunnan, and some areas of Liaoning ( Figure 8A ). By 2070, although the suitable area for D. punctatus continues to expand northward, the increase is relatively modest, and the suitable areas in the Yunnan region will have decreased slightly compared to 2050 ( Figure 8C ). Under the SSP5-8.5 scenario, the changes in suitable habitat for D. punctatus are more significant, expanding overall toward the northeastern region, covering areas from southern Gansu through Shaanxi, Shanxi, Hebei, Beijing, and Tianjin, and extending to Liaoning and parts of Jilin ( Figures 8B, D ). In future periods, the expansion area of the suitable distribution of D. punctatus generally exceeds its contraction area, indicating that future climatic conditions will be more favorable for the survival of this species. In contrast, the changes in suitable habitat for P. massoniana are relatively small but also exhibit a trend of northward expansion ( Figure 8 ). In both future climate scenarios, the changes in suitable habitat for P. massoniana are broadly consistent, showing a trend of northward expansion by 2050 ( Figures 8E, F ), with the expansion area mainly including Henan and Shandong, as well as central Shanxi and parts of Gansu in the SSP5-8.5 scenario. By 2070, the suitable habitat for P. massoniana remains relatively stable under both climate scenarios, with only a small part of the area experiencing expansion ( Figures 8G, H ). In these future climate scenarios, the stable areas of suitable habitat for D. punctatus and P. massoniana are essentially consistent, implying that P. massoniana will continue to face pest risks in the future.

Climate change is likely to cause significant changes in species distribution patterns, prompting their northward migration (Chen et al., 2022). Long-term climate observations indicate that, with the global warming trend, China’s annual average temperature is expected to rise by approximately 2.6°C, particularly under the high greenhouse gas emission scenario SSP5-8.5, where the temperature increase will be more pronounced. Concurrently, the annual average precipitation is projected to increase by 5.2%, particularly in North China and the Northwest region (Fu and Jiang, 2011). These changes are anticipated to further promote the northward migration of many species. Under current climatic conditions, the centroids of suitable areas for D. punctatus and P. massoniana are situated in Hunan Province ( Figure 9 ). Model predictions for the SSP1-2.6 scenario suggest that from the present to 2050, the centroid of the suitable habitat for D. punctatus will move 59.37 km to the northwest, and from 2050 to 2070, it will further move 31.52 km to the northeast., respectively, resulting in a total northward shift of 90.48 km. Under the SSP5-8.5 scenario, from the present to 2050, the centroid of the suitable habitat for D. punctatus is projected to move 84.85 km northwest, and from 2050 to 2070, it is expected to shift 76.54 km northeast, totaling a northward movement of 156.43 km. The suitable habitat centroid of P. massoniana also shifts northward, although the extent of migration is smaller, moving northward by 37.20 km and 73.09 km under the two future climate scenarios by 2070. This may be associated with the ecological characteristics of P. massoniana, whose needle-like leaves effectively prevent water evaporation, thus reducing its sensitivity to temperature changes. These results indicate that the suitable areas for D. punctatus and P. massoniana are expected to expand and migrate toward higher latitude regions with increasing temperature and precipitation, aligning with their preference for warm and humid environments. Global warming facilitates the spread of insects limited by low temperatures to higher latitude areas. Future trends of surface warming in China intensifying toward higher latitudes and the Tibetan Plateau, coupled with increased winter precipitation in the north, provide favorable conditions for the northward expansion of D. punctatus. Under different future climate scenarios, the movement directions and ranges of D. punctatus and P. massoniana show similarities, indicating that P. massoniana may face greater pest risks in the future.

Therefore, in the context of climate change, it is vital to effectively manage D. punctatus in various suitable habitats and to protect P. massoniana resources. Strengthening pest detection and adopting more proactive management practices are essential. Enhanced pest control measures in high-suitability areas for D. punctatus are critical, whereas in low-suitability areas, considering ecosystem conservation, reducing the use of chemical agents in favor of natural control methods is recommended. For P. massoniana, trunk injection technology can potentially be employed as an effective method to control pest infestations.

In this study, we employed the statistical method of cosine similarity to assess the similarity in the direction of centroid migration for D. punctatus and P. massoniana under different future climate scenarios and at two time points. Cosine similarity measures the similarity between two vectors by calculating the cosine of the angle between them. A cosine similarity value of 1 indicates that the two vectors are in the same direction, while a value of −1 indicates they are in opposite directions. Under the scenarios SSP1-2.6 and SSP5-8.5, from the present to 2050, the shift in distribution centroids for D. punctatus and P. massoniana showed a high degree of similarity, with a similarity index of 0.99. This indicates that the migration directions of the two species are almost identical. Furthermore, under the SSP5-8.5 scenario by 2050, their distribution centroids of D. punctatus and P. massoniana are closest, at a mere distance of 25.18 km. These findings reveal a critical insight: the period from the present to 2050 could be a high-risk phase for P. massoniana forests due to D. punctatus infestation, making this interval crucial for the control and management of D. punctatus in pine forests. This discovery holds significant implications for formulating future pest control strategies and management measures. The similarity in the shift directions of the distribution centroids for the two species between 2050 and 2070 under both scenarios also remains high, at 0.97 and 0.85, respectively. However, if proactive control measures against D. punctatus are implemented in the first phase, pest issues may be mitigated in the latter half of the century.

In this study, we focused on analyzing the potential distribution changes of D. punctatus and P. massoniana under future climate change scenarios. However, the model does not encompass all key factors that could influence the distribution of these two species, such as geographical barriers, natural enemies, human activities, and land use. Despite this, currently, no comprehensive model exists that can integrate all these factors for species distribution prediction (Guan et al., 2022). The future suitable habitats predicted in this study for the two species align with their growth habits, providing a valuable reference for understanding the habitat changes and migration directions of D. punctatus and P. massoniana in the context of climate change. Future research should consider a broader range of influencing factors to develop more comprehensive species distribution prediction models, which would enable more accurate predictions of the distribution of these two species.

The impact of climate change on ecosystems has become increasingly evident, leading to the northward spread of forest pests including Monochamus alternatus in China and the general trend of the potato pest Schrankia costaestrigalis moving toward the northeast and higher latitudes (Xu et al., 2020; Xian et al., 2023). In our study, the increases in two primary factors of climate change—temperature and precipitation—are key in influencing the spread of D. punctatus. As global climate change intensifies, global warming has significantly accelerated the northward expansion of D. punctatus. This phenomenon is concerning in that, in the not-too-distant future, D. punctatus may continue to migrate northward in search of new habitats, posing a greater threat to tree species in the northern regions. In addition, our research found that D. punctatus and P. massoniana (its primary host) share similar geographic distributions and migration trends. This finding suggests that in the future, with the further intensification of climate change, P. massoniana may face more severe pest risks. Therefore, our study underscores the importance of monitoring and managing these pest migration trends to mitigate their potential impact on ecosystems.

This study outlines three pivotal strategies for managing D. punctatus and conserving P. massoniana: (1) Given the projected suitable habitats in the current climate change scenario, enhancing surveillance and early warning systems for D. punctatus is critical. This encompasses promptly identifying and managing pest outbreaks to ensure swift and effective containment; (2) Considering the anticipated distribution patterns of D. punctatus, particularly in newly affected regions in northern China, reinforcing preventative measures is imperative. Forestry practices, in establishment and regeneration, should aim to modify habitats to create conditions less conducive to the proliferation of D. punctatus. This holistic approach is vital for curbing the spread of D. punctatus and mitigating its detrimental effects on ecosystems; (3) In P. massoniana forests already experiencing D. punctatus infestation, introducing broadleaf species and creating mixed coniferous-broadleaf forests are recommended. This strategy curbs the likelihood of extensive, high-density D. punctatus outbreaks and bolsters the overall resilience and stability of forest ecosystems. Additionally, utilizing the natural predators of D. punctatus (e.g., Trichogramma) for biological control is advisable. Cultivating and periodically releasing Trichogrammatid in suitable regions could serve as an effective natural control mechanism, preventing widespread D. punctatus infestations and thus safeguarding P. massoniana resources.