INTRODUCTION
Orthosia songi Chen et Zhang (Lepidoptera:Noctuidae) is a pest that feeds on the leaves and fruits of eucommia, causing serious damage to eucommia, with a specialized diet, high leaf feeding rate, rapid spread and long damage cycle affecting the normal development of the plant in light cases and causing the death of patches of eucommia in heavy cases, resulting in significant ecological damage and economic losses to the industry, and is now one of the most threatening pest disasters in the eucommia industry (DU & DU, 2020). In recent years, O. songi has caused massive outbreaks in major eucommia production areas such as Sichuan, Guizhou, Hunan, Shaanxi and Henan. In Hanzhong planting area of Shaanxi Province, more than 20 million plants and 500 hm2 were affected. Currently. Few studies have been conducted on O. songi, mainly focusing on the study of biological characteristics, occurrence patterns and control techniques, spatial sampling of larvae and control methods (ZHAO et al., 2015). As for monitoring and early warning, short-term warnings and risk levels at small scales are dominant, but long-term effective large-scale forecasts are lacking.
Species distribution models are often based on statistical models and environmental variables that extrapolate potential and actual distribution data of species in space and time, and are applied in research areas such as conservation of endangered species, invasive alien species, habitat planning and future mechanisms of climate and species distribution response (NGAREGA et al., 2021). MaxEnt has the advantages of good performance phenotypes and accurate prediction results, and is therefore widely used for species distribution prediction (QIAO et al., 2015). In recent years, the prediction of potential distributions of agricultural and forestry pests using MaxEnt models has become a hot research topic with remarkable results (ABOU-SHAARA et al., 2021). The purpose of this paper is to simulate the potential distribution and change trend of O. songi in China under past and future climatic conditions, and to find the key environmental variables affecting O. songi, to provide important references for the monitoring and control of O. songi.
MATERIALS AND METHODS
The source of occurrence points
Species distribution data were derived from the Chinese forest pest dataset (NATIONAL FORESTRY AND GRASSLAND ADMINISTRATION FOREST AND GRASSLAND PEST CONTROL STATION, 2020). Using ‘Orthosia songi Chen et Zhang (Lepidoptera: Noctuidae)’ as the keyword. Using ArcGIS annotations (latitude and longitude, based on the map of China (http://bzdt.ch.mnr.gov.cn/), excluding repeat or similar distribution points (two points are less than 5 km apart, take one point), and eventually get 73 distribution points.
Environmental variables
Climate data were primarily sourced from the Global Climate Interpolation Data Network (http://www.worldclim.org/). In the Sixth Assessment Report (AR6) of the International Panel on Climate Change (IPCC), 4 representative enrichment pathways (SSPs126, SSPs 245, SSPs 370 and SSPs 585) were established to represent possible future emission scenarios of greenhouse gases. Four scenarios of CO2 emissions (SSPs126, SSPs245, SSPs370 and SSPs585) were selected for the past (1970-2000) and future (2021-2040, 2041-2060, 2061-2080 and 2081-2100). Climate data include 19 bioclimatic variables (Bio01-Bio19) and 3 meteorological data (monthly precipitation, monthly maximum temperature and monthly minimum temperature), and the BCC-CSM2-MR model (BCC-CSM2-MR, developed by Beijing Climate Center, has shown good performance in simulating the Chinese climate) was chosen for the global climate model (XIN et al., 2019). Topographic data were sourced from the global SRTM data network (http://srtm.csi.cgiar.org/) and included 3 factors: slope, aspect and elevation. A spatial resolution of 30s (accuracy of 1 km×1 km) was used for all environmental variable data (Table 1).
3 methods (PCA correlation analysis, Spearman correlation analysis and folding knife method analysis) are combined in this paper to select environmental variables that have low correlation but are significant. Referring to the method of 22 environmental variables from different time periods were divided and selected, and finally 8 environmental variables were selected to be applied to the model calculations under past and future climate conditions, respectively.
Maximum entropy model optimization
MaxEnt includes regular multipliers (RM) and factor types (FC), which are used to optimise the model.RM promotes smoothing and minimises overfitting of the model, while FC ensures that the response curve corresponds to each variable (ZHU et al., 2018). FC has 5 main categories, including Linear features (L), Product features (P), Quadratic features (Q), Threshold features (T), and Hinge features (H). During parameter optimization, the RM value is set between 0.5 and 4 in increments of 0.5, and eight FC (L, LQ, LQH, LQHP, LQHPT, QHP, and QHPT) combinations are set. The two parameters were combined to construct the candidate model. The training data was 75% of the sample data randomly selected, and the test data was the remaining 25% of the sample data, and the repetition set at 10 times. ENMeval Tools was used to calculate the standardized Akaike Information Criterion Coefficient (AICc), which has criteria for evaluating the models, and the parameter set with the lowest AICc score was selected as the final model.
Classification of potentially suitable areas
The predicted areas were classified by ArcGIS 10.5 (Jenks’ natural breaks method) into 4 different levels: high potentially suitable areas (0.5 ≦ P < 0.1), medium potentially suitable areas (0.3 ≦ P < 0.5), low potentially suitable areas (0.1 ≦ P < 0.3) and unsuitable areas (P < 0.1).The area of the different habitat zones was also calculated.
Model evaluation
The area under the subject operating characteristic (ROC) curve, or AUC value, is widely used to evaluate the performance of species distribution models. The AUC ranges from 0-1, the closer the value is to 1, the higher the accuracy of the model (0.5 < AUC < 0.6, model was worse; 0.6 < AUC < 0.7, model was as poor; 0.7 < AUC < 0.8, model was good; 0.8 < AUC < 0.9, model was better, and 0.9 < AUC < 1.0, model was best). However, AUC values have some drawbacks in practical use and have received serious criticism (LOBO et al., 2008). To reduce the error arising from the AUC values, using Niche Analyst software (version 3.0; Nichea.sourceforge.net/) to calculate the AUC values for the local area ROC (P-ROC) (PETERSON et al., 2008). We set the error rate at 0.05 (E=0.05) and used the AUC ratio to judge the credibility of the model, as well as the minimum training presence omission rate (ORmtp) and the 10 percentile training presence omission rate (OR10) to verify whether the model was overfitted.
Calculate the offset of the mass point
The model with the highest accuracy and feasibility was selected, and ArcGIS 10.5 was used to extract the high, medium and low habitat areas of the moth in the past and in the future, which were combined into a total habitat area, and the centre of mass was labelled according to the mass point formula.
RESULTS
Screening of distribution sites
73 distribution points of O. songi were finally selected, including Zhejiang Province (1), Anhui Province (4), Fujian Province (1), Jiangxi Province (1), Henan Province (13), Hubei Province (8), Hunan Province (6), Guangxi Province (1), Sichuan Province (4), Guizhou Province (14), Yunnan Province (2), Shaanxi Province (15) and Gansu Province (8), collated in Excel and saved in csv format (Figure 1A).
Analysis of environmental variables
The 22 environmental variables under past and future time were eliminated and screened by three methods (PCA, Peason correlation analysis and knife-cut method) respectively, and finally 8 environmental variables with low correlation and high contribution were obtained, and the optimal range of the main environmental variables was classified according to the response curve (Figure 1B, 1C and Figure 1D and 1E). Of the 8 variables under 1970-2000, Tmin01 (72%), Prec02 (6.5%), Bio03 (4.1%), Bio04 (3.8%), Tmax01 (3.5%), Slope (2.8%), Aspect (2.6%), Tmax04 (2.4%), Bio02 (2.3%) and Tmax06 (0.2%) had optimum ranges of -10-5 ºC (-0.89 ºC), 2-67mm (18.8mm), 24-42% (29.1%), 528-1065 (788.2), 1-14 ºC (7.2 ºC), 0-20º (4.1º), 0-325 (143.8), 16-26 ºC (20.3 ºC), 6-14 ºC (8.8 ºC) and 20-32 ºC (27.3 ºC); of the 8 variables in 2021-2100, Bio02 (34.8%), Tmin10 (18.1%), Bio04 (13.9%), Tmax06 (11.6%), Prec09 (4.8%), Bio03 (4.8%), Prec03 (4.7%), Tmax03 (3.9%), Slope (1.9%) and Tmax10 (1.5%) had optimum ranges of 7-12 ºC (8.9 ºC), 10-19 ºC (14.6 ºC), 562-1027 (790.3), 26-36 ºC (31 ºC), 68-231mm (131.6mm), 24-37% (30.3%), 17-157mm (53.5mm) ,12-21 ºC (16 ºC), 0-25º (4.5º) and 18-28 ºC (23.7 ºC). (Figure 1F and 1G).
Classification of suitable areas
The value of AICc was minimized when FC = LQ and RM = 0.5, and the simulation results were reclassed into 4 classes of suitable areas (Figure 1H and 1I).
Under 1970-2000, the highly suitable areas were mainly in Henan, Hubei, Guizhou, Chongqing, Sichuan and Shaanxi Provinces, with local distribution in Shanxi, Hunan and Gansu Provinces, the moderately suitable areas were mainly in Hunan, Hubei and Sichuan Provinces, with local distribution in Anhui, Shandong, Hebei, Guangxi, Xizang Autonomous Region and Yunnan Provinces, and the slightly suitable areas were mainly in Jiangxi, Zhejiang, Fujian and Jiangsu Provinces, and sporadically in Guangdong and Guangxi Provinces. Under 2021-2100, the highly suitable areas are mainly in Sichuan, Chongqing, Guizhou, Hubei, Henan and Shaanxi provinces, with a few in Anhui and Jiangsu provinces, the moderately suitable areas are mainly in Hunan, Jiangxi, Zhejiang, Anhui, Guangxi and Shandong provinces, and the slightly suitable areas are mainly in Guangxi, Guangdong, Hunan and Yunnan provinces, with a few in Liaoning, A few are found in Liaoning, Hebei and Xizang Autonomous Regions (Figure 2A).
Validation of the model
According to the evaluation criteria in 2.5, under the environmental conditions from 1970-2100, the range of the PROC (E = 0.05) of the simulation results is between 0.8-0.9, and the average value is 0.852249, which performs well. Moreover, the values of ORmtp and OR10 are close to 0 and 0.1, the omission rate of the test data is close to the actual predicted value, and the deviation value is small, which proves that the model has good performance, high accuracy, and strong reliability. (Figure 2B and 1C and Table 2).
Delineating the area of different classes of suitable areas
The average area of the unsuitable areas was the largest and occupied the highest proportion (7560514.05km2 and 78.76%), followed by the slightly suitable areas (840151.93km2 and 8.75%), followed by the moderately suitable areas (675999.91km2 and 7.04%) and finally highly suitable areas (523334.11km2 and 5.45%), with a significant increase in the area of the highly suitable areas from 1970-2000 to 2021-2100 under climatic conditions. Under environmental conditions from 2021-2100, the area of highly suitable areas varied less and reached a maximum at 2081-2100-SSPs245. Fluctuating changes in area occur in the unsuitable and slightly suitable areas (Table 3).
Calculation of mass deflections
The plasmas were in Hubei Province during the 1970-2000 environmental conditions, then the plasmas were offset overall to the southeast, from higher to lower latitudes, and were all in Hunan Province during the 2021-2100 environmental conditions, with the offset being furthest from 1970-2000 to 2021-2040-SSPs245 and 2041-2060-SSPs126 to 2061- 2080-SSPs126 have the largest offset angles, reaching the lowest point in latitude for the 2061-2080-SSPs126 climatic condition. (Figure 2D and Table 4).
DISCUSSION
The 25 environmental factors (22 climatic factors and 3 topographic factors) of the 4 CO2 emission scenarios (SSPs126-585) in the past (1970-2000) and future (2021-2100) were classified by three methods, and 8 factors were identified as key factors affecting the distribution of O. songi. Slope was the key topographic factor influencing the distribution of O. songi, who concluded that the number and density of insects received a greater influence from slope in standing conditions. Bio02 (Mean Diurnal Range), Bio03 (Isothermality) and Bio04 (Temperature Seasonality), Tmin01, Tmax01, Tmin10, Tmax10 and Tmax06 were the key temperature factors that most significantly influenced the survival and reproduction of O. songi. Prec02, Prec03 and Prec09 were 2-67mm (18.8mm), 17-157mm (53.5mm) and 68-231mm (131.6mm) respectively, which were the most suitable precipitation levels for the moth, and too much precipitation in September could easily lead to failure of pupation of older mature larvae. Excessive precipitation in February and March is not conducive to the growth and development of the younger larvae.
In this study, the predicted results of the MaxEnt model were extracted, and it was found that the past distribution pattern and potential future distribution of O. songi did not show any significant changes, but remained mainly concentrated in the southeastern central region of China, with scattered distribution in the western and northeastern regions, but under future environmental conditions the area of highly suitable areas increased significantly and the distribution area became more extensive. Southeastern China has a humid subtropical climate with warm, humid winters, average temperatures above 1 ºC, hot summers, annual precipitation above 1000 mm, and a topography of plains and low hills, while global temperatures are expected to continue to increase in the coming decades according to a number of different CO2 emission scenarios from the Coupled Climate Modeling Project (CMIP6): until 2100, a projected Global average temperatures are predicted to increase by 1.8 to 4.0 ºC, making the region even more suitable for the growth and reproduction of O. songi.
The mass point of the total habitat area of O. songi shifted from Hubei to Hunan under past (1970-2000) to future (2021-2100) climatic conditions. This is inextricably linked to the distribution of host plants. The potential suitable range of Eucommia in the future climate is mainly distributed in the low valleys and hilly areas of southeastern China, and the suitable area is expanding (LIU et al., 2020), so the shifting distribution of O. songi is consistent with the migration of most species to areas with a dense distribution of host plants under a warming climate.
The average PROC value of the MaxEnt model in the paper reached 0.85, which indicates the highest accuracy of the results and is in the good range according to the rubric of AUC values. It has the advantage of performing better than other models when assessing potential species distributions using small sample sizes and presence-only datasets. Also in this study, the MaxEnt model was evaluated by setting the parameter sets for the element types and regularization coefficients, and by taking the PROC, ORmtp, OR10 and AUC ratios to avoid high AUC values and to bring them to a normal range, ensuring that the distribution of O. songi in China under past and future climatic conditions is correctly and effectively reflected.This study also provided an important reference for the future control of O. songi. Firstly, it will be used to guide the delineation of key control areas, with a focus on medium to high fitness areas in the future, and to strengthen monitoring and early warning for low fitness areas to prevent further damage. The second is to strengthen the attention to key meteorological factors, when the values reach the suitable range of O. songi, it is necessary to strengthen the monitoring of the forest to prevent the outbreak of insect damage.
CONCLUSION
This paper combined past and future global climatic data with the distribution points of O. songi and used the MaxEnt model to simulate the potential distribution areas in China. It was found that temperature, precipitation and topography factors play an important role in the distribution of O. songi under past and future environmental conditions, and are inextricably linked to its life history. The potential distribution area of the Eucommia is mainly in the low mountain and hilly areas of southeastern China, with an increase in the area of high habitat. The mass of the total habitat of the moth is shifted to the south-east by the distribution of Eucommia.
However, in this article, the model’s parameter settings are not complete enough to improve the simulation capability, the criteria for evaluating the model are too single. Secondly, in the species distribution points, fewer samples are taken, and only Chinese distribution data are used. It reflects that Orthosia songi has limitations in its response to global warming climate change. Thirdly, there are disadvantages in the selection of environmental variables. Only 2 variables, bioclimatic variables and topographic factors, are selected. Environmental variables such as sunshine, solar radiation, forest land factors, host plants, human factors, and soil fertility are not considered.
In the future research, it is necessary to expand and filter the distribution data of O. songi, reduce the spatial autocorrelation, and set the internal parameters of different models to achieve the best fitting results, and then choose more different environmental variables can reflect the distribution of pests and diseases from different levels, making the model more complete. In the context of the selection of global big data, it is necessary to add local real-time monitoring data at the same time to better reflect the spread of O. songi at different scales, which will also become one of the future research trends.
REFERENCES
* ABOU-SHAARA, H. et al. Modeling the invasion of the large hive beetle, Oplostomusfuligineus, into north africa and south europe under a changing climate. Insects, v.12, n.4, p.275, 2021. Available from: <Available from: https://www.mdpi.com/2075-4450/12/4/275 >. Accessed: Mar. 19, 2023. doi:10.3390/insects12040275.
» https://doi.org/10.3390/insects12040275.» https://www.mdpi.com/2075-4450/12/4/275
* DU, H. Y.; DU, Q. X. Foundation, problems and countermeasures of Eucommia ulmoides industry high-quality development in China. Nonwood For Research, v.38, n.1, p.1-10, 2020. Available from: <Available from: https://jjlyj.csuft.edu.cn/#/digest?ArticleID=1201 >. Accessed: Sept. 10, 2023. doi:10.14067/j.cnki.1003-8981.2020.01.001.
» https://doi.org/10.14067/j.cnki.1003-8981.2020.01.001.» https://jjlyj.csuft.edu.cn/#/digest?ArticleID=1201
* LIU, P. F. et al. Estimation of potential suitable distribution area and the ecological characteristics of Eucommia ulmoides Oliv. in China. Acta Ecologica Sinica, v.40, n.16, p.5674-5684, 2020. Available from: <Available from: https://www.ecologica.cn/stxb/article/abstract/stxb201907091450?st >. Accessed: Oct. 05, 2023. doi: 10.5846/stxb201907091450.
» https://doi.org/10.5846/stxb201907091450.» https://www.ecologica.cn/stxb/article/abstract/stxb201907091450?st
* LOBO, J. M. et al. AUC: A misleading measure of the performance of predictive distribution models. Global Ecology And Biogeography, v.17, n.2, p.145-151, 2008. Available from: <Available from: https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1466-8238.2007.00358.x >. Accessed: Apr. 03, 2022. doi: 10.1111/J.1466-8238.2007.00358.X.
» https://doi.org/10.1111/J.1466-8238.2007.00358.X.» https://onlinelibrary.wiley.com/doi/abs/10.1111/j.1466-8238.2007.00358.x
* NATIONAL FORESTRY AND GRASSLAND ADMINISTRATION FOREST AND GRASSLAND PEST CONTROL STATION. Overview of China’s forestry pests: 2014-2017 national forestry pest survey results. p.1291, 2020, China Forestry Publishing House.
* NGAREGA, B. K. et al. Forecasting the effects of bioclimatic characteristics and climate change on the potential distribution of Colophospermum mopane in southern Africa using Maximum Entropy (Maxent). Ecological Informatics, v.5, 101419, 2021. Available from: <Available from: https://www.sciencedirect.com/science/article/pii/S1574954121002107 >. Accessed: Apr. 18, 2022. doi: 10.1016/j.ecoinf.2021.101419.
» https://doi.org/10.1016/j.ecoinf.2021.101419.» https://www.sciencedirect.com/science/article/pii/S1574954121002107
* PETERSON, A.T. et al. Rethinking receiver operating characteristic analysis applications in ecological niche modeling. Ecological Modelling, v.213, n.1, p.63-72, 2008. Available from: <Available from: https://www.sciencedirect.com/science/article/pii/S0304380007006163 >. Accessed: Dec. 05, 2022. doi: 10.1016/j.ecolmodel.2007.11.008.
» https://doi.org/10.1016/j.ecolmodel.2007.11.008.» https://www.sciencedirect.com/science/article/pii/S0304380007006163
* QIAO, H. et al. No silver bullets in correlative ecological niche modelling: insights from testing among many potential algorithms for niche estimation. Methods in Ecology and Evolution, v.6, n.10, p.1126-1136, 2015. Available from: <Available from: https://besjournals.onlinelibrary.wiley.com/doi/full/10.1111/2041-210X.12397 >. Accessed: Jan. 01, 2024. doi: 10.1111/2041-210X.12397.
» https://doi.org/10.1111/2041-210X.12397.» https://besjournals.onlinelibrary.wiley.com/doi/full/10.1111/2041-210X.12397
* XIN, X. G. et al. Introduction of BCC models and its participation in CMIP6. Advances in Climate Change Research, v.15, n.533, p.533-5392019. Available from: <Available from: http://www.climatechange.cn/EN/abstract/abstract1147.shtml >. Accessed: Feb. 12, 2024. doi: 10.12006/j.issn.1673-1719.2019.039.
» https://doi.org/10.12006/j.issn.1673-1719.2019.039.» http://www.climatechange.cn/EN/abstract/abstract1147.shtml
* ZHAO, Y. et al. Toxicity and field efficacy of three biopesticides to Orthosia songi Plant Protection, v.41, n.3, p.210-214, 2015. Available from: <Available from: http://www.plantprotection.ac.cn/ch/reader/view_abstract.aspx?flag=1&file_no=20150341&journal_id=zwbh >. Accessed: Feb. 06, 2022. doi: 10.3969/j.issn.0529-1542.2015.03.042.
» https://doi.org/10.3969/j.issn.0529-1542.2015.03.042.» http://www.plantprotection.ac.cn/ch/reader/view_abstract.aspx?flag=1&file_no=20150341&journal_id=zwbh
* ZHU, G. P. et al. Effects of model parameters in MaxEnt modeling of ecological niche and geographic distribution: case study of the brown marmorated stink bug, Halyomorpha haly Journal of Biosafety, v.27, n.2, p.118-123, 2018. Available from: <Available from: http://www.jbscn.net/swuaqxbcn/article/abstract/20180207 >. Accessed: Jun. 08, 2022. doi: 10.3969/j.issn.2095-1787.2018.02.007.
» https://doi.org/10.3969/j.issn.2095-1787.2018.02.007.» http://www.jbscn.net/swuaqxbcn/article/abstract/20180207
Wu, Sijun
Sichuan Agricultural University
Qiao, Tianmin
Sichuan Agricultural University
Li, Shujiang
Sichuan Agricultural University
Hu, Binhong
National Forestry and Grassland Administration Key Laboratory of Forest Resources Conservation and Ecological Safety on the Upper Reaches of the Yangtze River&Forestry Ecological Engineering in the Upper Reaches of the Yangtze River Key Laboratory of Sichuan Province
Zhu, Tianhui
Sichuan Agricultural University
You have requested "on-the-fly" machine translation of selected content from our databases. This functionality is provided solely for your convenience and is in no way intended to replace human translation. Show full disclaimer
Neither ProQuest nor its licensors make any representations or warranties with respect to the translations. The translations are automatically generated "AS IS" and "AS AVAILABLE" and are not retained in our systems. PROQUEST AND ITS LICENSORS SPECIFICALLY DISCLAIM ANY AND ALL EXPRESS OR IMPLIED WARRANTIES, INCLUDING WITHOUT LIMITATION, ANY WARRANTIES FOR AVAILABILITY, ACCURACY, TIMELINESS, COMPLETENESS, NON-INFRINGMENT, MERCHANTABILITY OR FITNESS FOR A PARTICULAR PURPOSE. Your use of the translations is subject to all use restrictions contained in your Electronic Products License Agreement and by using the translation functionality you agree to forgo any and all claims against ProQuest or its licensors for your use of the translation functionality and any output derived there from. Hide full disclaimer
© 2025. This work is published under https://creativecommons.org/licenses/by-nc/4.0/ (the “License”). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.
Abstract
The Orthosia songi Chen et Zhang is one of the important leaf-eating pests of eucommia. In this paper, combined global climatic data with the distribution points of Orthosia songi and identified the key environmental factors affecting its distribution by knife cutting, PCA principal component analysis and Spearman correlation analysis. Used MaxEnt to fit the potential distribution areas in the past and future, and marked the offset mass points to analyse the migration trend of Orthosia songi distribution. The results showed that in 1970-2000, the 10 key variables were Tmin01, Prec02 ,Bio03, Bio04, Tmax01, Slope, Aspect , Tmax04, Bio02 and Tmax06 ; in 2021-2100, the factors were Bio02, Tmin10, Bio04, Tmax06,Prec09, Bio03,Prec03, Tmax03, Slope and Tmax10. Potential suitable areas for Orthosia songi in China were divided into 4 classes, with the unsuitable areas having the highest proportion (78.76%) and the high-suitable areas were lowest (5.45%). The fitness zone mass point of Orthosia songi was in Hubei Province in 1970-2000, and then shifted overall to the southeast, and shifted to Hunan Province in 2021-2100.
