Potential distribution study of Caragana microphylla in China based on MaxEnt modeling Authors: Hongcheng Zhao, Xiaohuang Liu, Peng Li, Xinping Luo, Jiufen Liu, Chaozhu Li, Run Liu, Junnan Li, Xiaotong Liu, Feiran Wang Published: 2026 ABSTRACT Climate, soil factors, and topography play a decisive role in the spatial distribution of species. Predicting the potential distribution of Caragana microphylla Lam. over time is of great significance for maintaining ecological balance, soil and water conservation, and evaluating ecological services in northwestern China. In this study, we used the Global Biodiversity Information Service (GBIF) and SDM Toolbox v2.5 to remove spatial autocorrelation from 103 occurrence records of C. microphylla. We then employed the Maximum Entropy model and ArcGIS software with correlation analysis to select 6 climatic factors, 23 soil factors, and 3 topographic factors for analyzing the potential suitable habitats of C. microphylla, and predicted its distribution changes under three SSP scenarios (2030, 2050, 2070).The results show: (1) Current suitable habitat covers 3.329 million km2 (34.7% of China), mainly in northern regions; (2) Annual precipitation (bio-12, 350–400 mm), slope (89–90°), Precipitation of Coldest Quarter (bio-19, 0–10 mm), Topsoil Calcium Carbonate (t_CaCO₃, 3–5%), and temperature seasonality (bio-4, 1060–1080) were dominant factors (cumulative contribution: 67.6%); (3) Higher CO₂ scenarios (SSP37.0) increased suitable areas compared to SSP12.6; (4) Future distribution centers will migrate westward, with maximal displacement (448.51 km) under 2050-SSP24.5. It is recommended that priority be given to the protection of existing populations in Inner Mongolia and Xinjiang. while assisted migration programs could target western expansion zones under high-emission scenarios. In vulnerable areas like the Qinling Mountains, soil amendments should be applied to mitigate waterlogging risks from extreme precipitation events (> 50 mm). Furthermore, a long-term monitoring network should be established across the species’ distribution range to dynamically assess population changes of C. microphylla and inform adaptive management strategies. Introduction Plants are fundamental to material cycles and energy flows in ecosystems, making them a crucial component of terrestrial ecosystems1,2. Global climate change and soil factors significantly influence species distribution by affecting their habitats, including ecosystem structure and function3,4. These changes have a profound impact on plant geographical distribution. As human activities intensify, global warming and surface disturbances are increasing5. Consequently, changes in climate and soil factors have become key drivers of species distribution shifts, affecting ecological habits, geographical distribution, ecosystem structure, and function, leading to biodiversity loss and ecological degradation6,7. Understanding these factors is vital for studying trends in species habitat distribution patterns across different periods, which is essential for developing effective conservation strategies, preserving biodiversity, and maintaining environmental stability. Ecological niche models (ENMs) quantify the relationship between species occurrence and various environmental variables to describe the ecological niche or habitat suitability of a species8. They are effective tools for explaining changes in species distribution due to environmental factors such as climate and soil9. There are many types of ecological niche models, including the Genetic Algorithm for Rule Set Production (GARP)10, the Bioclimatic Analysis and Prediction System (BIOCLIM), Bioclimatic Model (BM)11, Random Forest Models (RFMs)12, and Maximum Entropy (MaxEnt) model13. The MaxEnt model is particularly popular among researchers due to its intuitive modeling, high prediction accuracy, ease of use, and strong explanatory power14,15. It can accurately and reliably simulate potential habitats for species using only distribution points and environmental data16. The MaxEnt model provides a framework that is both flexible and powerful for learning complex probability distributions from limited observational data. It is one of the indispensable tools in the field of statistical inference and modern machine learning while maintaining the simplicity of the model and maximizing the predictive power of unknown data17,18. With the enhancement of computing power and the continuous optimization of algorithms, MaxEnt models and their variants will continue to play an important role in many fields. In recent years, Deng et al. used the MaxEnt model to analyze the primary ecological factors affecting the distribution of Dendrocalamus latifloru, accurately predicting its habitat distribution19. Liao et al. constructed potential distribution patterns for Miscanthus nudipes over three different periods and quantitatively assessed changes in its suitability under various ecological and climatic scenarios. The results indicate that the distribution center of Miscanthus nudipes is likely to shift toward its glacial period distribution center in the future20. Wang et al. employed the MaxEnt model to analyze the spatial distribution and trends of alpine oak habitats, finding that temperature is a critical factor influencing these plants. They observed that global warming might negatively impact the long-term growth of alpine oak, with its center of mass likely shifting to lower altitudes and latitudes under future climate conditions21. This widespread use underscores the model’s effectiveness in ecological research. Caragana microphylla Lam is a perennial deciduous shrub belonging to the genus Caragana Fabr in the Fabaceae family22. It is a significant plant in the north temperate region of China, contributing to grassland scrub, and is widely distributed across Inner Mongolian grasslands23,24. It plays a crucial role in windbreaks, sand fixation, and soil and water conservation. Its shoots and leaves are also high-quality shrub fodder for desert and desert grassland zones, rich in proteins and other nutrients, making them suitable for pasture. Research indicates that using C. microphylla as fodder enhances livestock nutritional intake and offers notable ecological and economic value25. Additionally, its extracts have demonstrated significant anti-inflammatory, antibacterial, and antioxidant effects26. To explore the spatial distribution pattern of C. microphylla over different periods, this study investigated the potential distribution of C. microphylla in China using contemporary and future environmental data for the periods 2021–2040 (2030), 2041–2060 (2050), and 2061–2080 (2070). These projections were based on the BCC-CSM2-MR global climate model from WorldClim, considering three Representative Concentration Pathways (SSP12.6, SSP24.5, and SSP37.0) that reflect varying greenhouse gas emission trajectories: SSP12.6 represents low emissions with radiative forcing reaching 2.6 W/m2 by 2100, SSP24.5 intermediate emissions (4.5 W/m2), and SSP37.0 high emissions (7.0 W/m2) with CO₂ concentrations exceeding 850 ppm.The study aimed to reveal the dominant environmental factors influencing the distribution of C. microphylla and the changes in spatial patterns of potential habitats and the migration of the species under three future greenhouse gas emission scenarios (SSP12.6, SSP24.5, and SSP37.0). This research provides a basis for understanding the future spatial distribution and migration patterns of C. microphylla, offering a reference for the conservation, resource management, and ecological restoration of the species. Data and methods Data on the geographic distribution of the C. microphylla The geographic distribution data of C. microphylla were obtained from the GBIF public data platform, yielding 331 samples from China. These samples were imported into Excel for screening and filtering, and to ensure the accuracy of the distribution data, points with duplicated geographic coordinates, points with inaccurate geographic coordinates, and points significantly deviating from the distribution range of the species were excluded. To ensure the accuracy of the distribution points, the duplicated, inaccurate, and significantly deviated from the distribution range of the species were excluded27. To ensure the accuracy of the distribution data, duplicate geographic coordinates, inaccurate points, and points significantly deviating from the species’ known distribution range were excluded. The study area covers the natural distribution range of C. microphylla across China, though the filtered occurrence records are predominantly clustered in northern regions (including Inner Mongolia, Xinjiang, Gansu and northeast China), reflecting the species’ typical arid/semi-arid habitat preferences. Southern records were excluded during filtering as they represented either marginal occurrences or potential misidentifications.The map, downloaded from the National Centre for Basic Geographic Information (http://www.ngcc.cn), was imported into ArcGIS 10.2, a geographic information system (GIS) tool. Along with the distribution point data for the C. microphylla, all points were displayed on the Map. A threshold of 10 km was applied to reduce spatial self-correlation28,29. The data points were then saved in.csv format, resulting in 103 valid data points. The coordinate system was standardized, and the data was classified and sorted. The results are shown in Table 1, and the locations of the distribution points are illustrated in Fig. 1.Table 1Geographical distribution sample statistics of Caragana microphylla Lam.Serial numberProvinceQuantitiesRecording intervalOccurrence rate/percentLongitude range/°Latitude range/°1Heilongjiang Province3121.1831 ~ 135.0885 E43.4230 ~ 53.5609 N2.912Inner Mongolia Autonomous Region3697.1717 ~ 126.0656 E37.4066 ~ 53.3338 N34.953Xinjiang Uighur Autonomous Region673.5011 ~ 96.3848 E34.3361 ~ 49.1831 N5.834Jilin Province in northeast China2121.6390 ~ 131.3099 E40.8642 ~ 46.3022 N1.945Liaoning Province4118.8397 ~ 125.7856 E38.7215 ~ 43.4885 N3.886Gansu Province892.3378 ~ 108.7090 E32.5963 ~ 42.7945 N7.777Hebei Province12113.4549 ~ 119.8488 E36.0487 ~ 42.6155 N11.658Capital of People’s Republic of China3115.4168 ~ 117.5083 E39.4421 ~ 41.0590 N2.919Shanxi Province12110.2302 ~ 114.5629 E34.5838 ~ 40.7450 N11.6510Shaanxi Province10105.4883 ~ 111.2419 E31.7069 ~ 39.5825 N9.7111Ningxia Hui Autonomous Region4104.2843 ~ 107.6617 E35.2385 ~ 39.3879 N3.8812Shandong Province2114.8101 ~ 122.7056 E34.3774 ~ 38.3999 N1.9413Henan Province1114.84 E34.06 N0.97Fig. 1Distribution point map of Caragana microphylla Lam.. Map was created using ArcGIS 10.7 (ESRI, Redlands, CA, USA). Table 1Geographical distribution sample statistics of Caragana microphylla Lam.Serial numberProvinceQuantitiesRecording intervalOccurrence rate/percentLongitude range/°Latitude range/°1Heilongjiang Province3121.1831 ~ 135.0885 E43.4230 ~ 53.5609 N2.912Inner Mongolia Autonomous Region3697.1717 ~ 126.0656 E37.4066 ~ 53.3338 N34.953Xinjiang Uighur Autonomous Region673.5011 ~ 96.3848 E34.3361 ~ 49.1831 N5.834Jilin Province in northeast China2121.6390 ~ 131.3099 E40.8642 ~ 46.3022 N1.945Liaoning Province4118.8397 ~ 125.7856 E38.7215 ~ 43.4885 N3.886Gansu Province892.3378 ~ 108.7090 E32.5963 ~ 42.7945 N7.777Hebei Province12113.4549 ~ 119.8488 E36.0487 ~ 42.6155 N11.658Capital of People’s Republic of China3115.4168 ~ 117.5083 E39.4421 ~ 41.0590 N2.919Shanxi Province12110.2302 ~ 114.5629 E34.5838 ~ 40.7450 N11.6510Shaanxi Province10105.4883 ~ 111.2419 E31.7069 ~ 39.5825 N9.7111Ningxia Hui Autonomous Region4104.2843 ~ 107.6617 E35.2385 ~ 39.3879 N3.8812Shandong Province2114.8101 ~ 122.7056 E34.3774 ~ 38.3999 N1.9413Henan Province1114.84 E34.06 N0.97 Geographical distribution sample statistics of Caragana microphylla Lam. Distribution point map of Caragana microphylla Lam.. Map was created using ArcGIS 10.7 (ESRI, Redlands, CA, USA). Environmental variable data In this study, we used 19 bioclimatic variables provided by the WorldClim International Climate Website (http://www.worldclim.org/) for historical climate data. These variables reflect temperature and precipitation characteristics and their seasonal variations over time, with a spatial resolution of 5 km. For future climate data, we utilized the BCC-CSM2-MR model from WorldClim and selected two or three greenhouse gas (GHG) emission scenarios expressed in terms of Shared Socioeconomic Pathways (SSP): a low concentration scenario (SSP12.6), a medium concentration scenario (SSP24.5), and a high concentration scenario (SSP37.0). The time periods analyzed were 2021–2040 (2030), 2041–2060 (2050), and 2061–2080 (2070), with a resolution of 2.5 min. Topographic data were obtained from the Geospatial Data Cloud (https://www.gscloud.cn), with slope and aspect derived from elevation data resampled by ArcGIS 10.2. Soil data, including soil properties, type, and land use/cover data, were downloaded from the World Soil Database HWSD (www.fao.org), with 35 soil factors extracted using ArcGIS. Using a vector map of China’s administrative divisions (1:4 million) as the base map, each environmental variable was cropped to the Chinese region and unified to the World Geodetic System (WGS1984). All environmental factors were processed by ArcGIS with a resolution of 30″ (1 km × 1 km). To avoid data redundancy and overfitting due to high correlations among variables, Pearson correlation analysis was performed using SPSS 2.0 software. Variables with strong correlations (|r|≥ 0.85) were excluded, and 32 environmental variables were retained30,31 for simulation training (Table 2).Table 2Environmental variables.Serial numberTypologyVariantDescriptions1Climacticbio-2Mean Diurnal Range (Mean of monthly (max temp—min temp))2bio-4Temperature Seasonality (standard deviation × 100)3bio-10Mean Temperature of Warmest Quarter4bio-12Annual Precipitation5bio-15Precipitation Seasonality (Coefficient of Variation)6bio-19Precipitation of Coldest Quarter7Groundt_CaCO3Topsoil Calcium Carbonate8t_bsTopsoil Base Saturation9t_bulk_densityTopsoil Bulk Density10t_slitTopsoil Silt Fraction11t_gravelTopsoil Gravel Content12t_ph_H2OTopsoi1 pH (H2O)13t_ocTopsoil Organic Carbon14t_clayTopsoil Clay Fraction15t_CaSO4Topsoil Gypsum16s_CaCO3Subsoil Calcium Carbonate17s_cec_claySubsoil CEC (clay)18s_CaSO4Subsoil Gypsum19s_tebSubsoi1 TEB20s_bsSubsoil Base Saturation21s_gravelSubsoil Gravel Content22s_espSubsoil Sodicity (ESP)23s_bulk_densitySubsoil Bulk Density24s_claySubsoil Clay Fraction25s_cec_soilSubsoil CEC (soil)26s_eceSubsoil Salinity (ECe)27s_ocSubsoil Organic Carbon28awc_classAWC Range29add_propOther properties (gelic, vertic, and petric)30Topographicalslope31altElevation (dem)32aspet Table 2Environmental variables.Serial numberTypologyVariantDescriptions1Climacticbio-2Mean Diurnal Range (Mean of monthly (max temp—min temp))2bio-4Temperature Seasonality (standard deviation × 100)3bio-10Mean Temperature of Warmest Quarter4bio-12Annual Precipitation5bio-15Precipitation Seasonality (Coefficient of Variation)6bio-19Precipitation of Coldest Quarter7Groundt_CaCO3Topsoil Calcium Carbonate8t_bsTopsoil Base Saturation9t_bulk_densityTopsoil Bulk Density10t_slitTopsoil Silt Fraction11t_gravelTopsoil Gravel Content12t_ph_H2OTopsoi1 pH (H2O)13t_ocTopsoil Organic Carbon14t_clayTopsoil Clay Fraction15t_CaSO4Topsoil Gypsum16s_CaCO3Subsoil Calcium Carbonate17s_cec_claySubsoil CEC (clay)18s_CaSO4Subsoil Gypsum19s_tebSubsoi1 TEB20s_bsSubsoil Base Saturation21s_gravelSubsoil Gravel Content22s_espSubsoil Sodicity (ESP)23s_bulk_densitySubsoil Bulk Density24s_claySubsoil Clay Fraction25s_cec_soilSubsoil CEC (soil)26s_eceSubsoil Salinity (ECe)27s_ocSubsoil Organic Carbon28awc_classAWC Range29add_propOther properties (gelic, vertic, and petric)30Topographicalslope31altElevation (dem)32aspet Environmental variables. Modelling The 103 valid data points and 32 environmental variables for C. microphylla were imported into the MaxEnt model. Bootstrap was selected as the test method, with 25% of the distribution data randomly selected as the test dataset. The Maximum number of iterations was set to 1× 106, with 10 repetitions conducted in total. The environmental factor response curves and variable importance were output using the knife-cut method, and results were saved in ASCII Logistic format, with other settings at default32,33. The MaxEnt software’s built-in function evaluated the model’s simulation ability by outputting the Receiver Operating Characteristic Curve (ROC Curve) and calculating the area under the ROC Curve (AUC). The accuracy test primarily evaluates the MaxEnt model by analyzing the area under the Receiver Operating Characteristic (ROC) curve, quantified as the AUC value. Data on distribution and environmental variables are input into the MaxEnt model. The training set is simulated using the model’s internal function, iterated 10 times to generate the ROC curve. Classification criteria for AUC values are as follows: 0.50–0.60 (fail), 0.60–0.70 (poor), 0.70–0.80 (average), 0.80–0.90 (good), and 0.90–1.00 (very good)34,35. Classification of suitable classes and area calculation Using ArcGIS 10.2 software, we visualized the simulation results and classified habitat suitability for Chinese C. microphylla. Based on the simulated suitability and collected geographic distribution data, we classified the habitat into four levels using the Jenks’ natural breaks method: non-suitable, low suitable, medium suitable, and high suitable36. The Reclassify tool in ArcGIS 10.2 was then used to calculate and summarize the area of each class. Results and analyses Model accuracy The resulting AUC value is 0.868, indicating high accuracy and reliability of the model’s calculations (Fig. 2). The model’s strong performance (AUC = 0.868) effectively characterizes C. microphylla’s ecological niche requirements, particularly its adaptation to environments with minimal winter precipitation (0–10 mm) and moderately calcareous soils (3–5% carbonate content). This is demonstrated through the model’s internal consistency across 10 replicate runs (SD ± 0.015) and enhanced predictive accuracy in core distributional regions like Inner Mongolia (AUC = 0.89). These results significantly exceed the typical accuracy range reported for similar arid-zone shrub species (AUC 0.75–0.85), underscoring the value of our integrated modeling approach that simultaneously considers both climatic and edaphic drivers. The model’s robust performance supports its application for projecting potential climate refugia under future emission scenarios.Fig. 2ROC and AUC of the Caragana microphylla Lam. suitable habitat region model. ROC and AUC of the Caragana microphylla Lam. suitable habitat region model. Dominant environment variables Analysis of the contribution rates of different environmental factors revealed that annual precipitation had the highest contribution rate of 18.4% among the 32 variables. Other significant factors with contribution rates exceeding 10% include slope, precipitation in the coldest season (bio-19), carbonate or lime content in the top layer (t_CaCO3), and the coefficient of seasonal variation of air temperature (bio-4). Together, these factors account for 67.6% of the model’s variability, annual precipitation (bio-12) exhibited both the highest contribution rate (18.4%) and most pronounced AUC reduction when excluded (‘without variable’ in Fig. 3), confirming its primary role, whereas other high-contribution variables (e.g., t_CaCO₃) showed attenuated AUC dependence, potentially reflecting covariance with other soil characteristics(Table 3, Fig. 3). The factor with the lowest contribution was bottom exchangeable salinity (s_oc) at 0.1%. The response curves for three environmental variables were analyzed, revealing that the optimal growing conditions for C. microphylla include annual precipitation of 350–400 mm, a slope of 89–90, precipitation of 0–10 mm in the coldest season, carbonate or Lime content in the top layer of 3%–5%, and a seasonal temperature variation coefficient of 1060–1080. These findings suggest that C. microphylla is tolerant to drought and salinity, with higher annual precipitation correlating with a lower survival probability for C. microphylla. The closer the slope is to 90°, the higher the probability of survival of C. microphylla, which may be due to the fact that there is almost no risk of soil water accumulation under the condition of vertical slope, while C. microphylla is drought tolerant and waterlogged, and the steep slope can avoid root rot, and it is also related to the shading effect of the steep slope, which can reduce direct sunlight and water evaporation.This also verifies the conclusion that the higher the annual precipitation, the lower the survival probability of C. microphylla. (Fig. 4).Fig. 3Results of using the knife-cut method to test the importance of ecological factors on the distribution of Caragana microphylla Lam..Table 3Contribution of the dominant environmental variables.Environment variableContribution (%)bio-1218.4slope13.8bio-1912.3t_CaCO312.2bio-410.9t_bs5bio-104.7alt4.1bio-23.2bio-151.9s_CaCO31.8t_bulk_density1.7aspect1.5t_silt1.3s_cec_clay1.2s_CaSO40.9s_teb0.8s_bs0.5s_gravel0.5s_esp0.4t_gravel0.4awc_class0.4t_ph_H2O0.3add_prop0.3t_oc0.2s_bulk_density0.2s_clay0.2s_cec_soil0.2t_clay0.2s_ece0.2t_CaSO40.1s_oc0.1Fig. 4Response curve of key environmental variables. Results of using the knife-cut method to test the importance of ecological factors on the distribution of Caragana microphylla Lam.. Table 3Contribution of the dominant environmental variables.Environment variableContribution (%)bio-1218.4slope13.8bio-1912.3t_CaCO312.2bio-410.9t_bs5bio-104.7alt4.1bio-23.2bio-151.9s_CaCO31.8t_bulk_density1.7aspect1.5t_silt1.3s_cec_clay1.2s_CaSO40.9s_teb0.8s_bs0.5s_gravel0.5s_esp0.4t_gravel0.4awc_class0.4t_ph_H2O0.3add_prop0.3t_oc0.2s_bulk_density0.2s_clay0.2s_cec_soil0.2t_clay0.2s_ece0.2t_CaSO40.1s_oc0.1 Contribution of the dominant environmental variables. Response curve of key environmental variables. Potential habitat zones of C. microphylla under contemporary climate and soil The habitat modeling and classification of the C. microphylla in China indicate that it is predominantly distributed across 13 provinces. Its primary habitat is concentrated in the south-central part of the Inner Mongolia Autonomous Region, northern Shaanxi, central Shanxi, central Hebei, central Gansu, and eastern Ningxia Hui Autonomous Region. There are also scattered distributions in the southeastern Inner Mongolia Autonomous Region, eastern Liaoning, and northwestern Xinjiang. In the middle fertile area, scattered distributions are found in northeastern Inner Mongolia Autonomous Region, northern Hebei, southwestern Shanxi, northern Shaanxi, central Inner Mongolia, southeastern Gansu, northern Liaoning, and northern Hebei. In the low fertile area, the distribution is scattered across central Shaanxi, southwestern Shanxi, southwestern Inner Mongolia, southeastern Gansu, northern Ningxia, northern Liaoning, and southwestern Xinjiang. Generally, suitable areas for the C. microphylla are primarily located in the Loess Plateau, the Inner Mongolian Plateau, the North China Plain, the Northeastern Plain, and the Junggar Basin between the Tien Shan and Altai Shan mountains, all situated in northern China above 30 N latitude (Fig. 5).Fig. 5Distribution of the suitable areas for Caragana microphylla Lam.. Map was created using ArcGIS 10.7 (ESRI, Redlands, CA, USA). Distribution of the suitable areas for Caragana microphylla Lam.. Map was created using ArcGIS 10.7 (ESRI, Redlands, CA, USA). Modelling of current and future potential habitat areas for the C. microphylla In this study, four periods were selected to predict the potential distribution of C. microphylla. As shown in Fig. 6 and Table 4, the current distribution of C. microphylla covers most of northeast and north China, as well as a small part of northwest China. The total suitable area for C. microphylla spans 3.329 million km2, primarily in the northern part of China.Fig. 6Area of the potential habitat of Caragana microphylla Lam. plants at different time periods. Map was created using ArcGIS 10.7 (ESRI, Redlands, CA, USA).Table 4Area of potential habitat for Caragana microphylla Lam. plants at different periods of time.Climate scenariosHighly (× 104 km2)Medium (× 104 km2)Low (× 104 km2)Total (× 104 km2)Current64.3105.8162.8332.92030-SSP12.640.580.0152.3272.82030-SSP24.50.18.258.266.52030-SSP37.0429.1163.5122.5715.12050-SSP12.64.342.5149.5196.32050-SSP24.5338.7170.5125.8635.02050-SSP37.06.834.7111.9153.42070-SSP12.6302.7160.9127.3590.92070-SSP24.5358.1166.5115.4640.02070-SSP37.052.8101.1184.5338.4 Area of the potential habitat of Caragana microphylla Lam. plants at different time periods. Map was created using ArcGIS 10.7 (ESRI, Redlands, CA, USA). Table 4Area of potential habitat for Caragana microphylla Lam. plants at different periods of time.Climate scenariosHighly (× 104 km2)Medium (× 104 km2)Low (× 104 km2)Total (× 104 km2)Current64.3105.8162.8332.92030-SSP12.640.580.0152.3272.82030-SSP24.50.18.258.266.52030-SSP37.0429.1163.5122.5715.12050-SSP12.64.342.5149.5196.32050-SSP24.5338.7170.5125.8635.02050-SSP37.06.834.7111.9153.42070-SSP12.6302.7160.9127.3590.92070-SSP24.5358.1166.5115.4640.02070-SSP37.052.8101.1184.5338.4 Area of potential habitat for Caragana microphylla Lam. plants at different periods of time. According to Fig. 6 and Table 4, from the present to the future, under the 2030-SSP37.0 high-concentration GHG emission scenario, the total suitable area increases to over 7 million km2. Both high- and medium-sized suitable areas show an increasing trend, with high-suitability areas growing by 3,648,000 km2 and medium-suitability areas by 577,000 km2. The high-suitability area increases the most, while the low-suitability area decreases slightly. Under the SSP24.5 medium GHG emission scenario, all three habitat types decrease in area compared to the present, with the total suitable area reducing by 2.664 million km2. By 2050, under the SSP37.0 high GHG emission scenario, all three habitat types decrease, with medium-suitability areas showing the greatest reduction of 711,000 km2. The total suitable area shrinks to 1,534,000 km2. Under the SSP24.5 medium GHG emission scenario, high-suitability areas grow, while low-suitability areas decline. However, the total suitable area still increases under all three GHG emission scenarios for 2070, with the SSP24.5 medium GHG emission scenario showing the largest increase, doubling the current total suitable area. Overall, the SSP24.5 scenario has a significant impact on suitable habitat across different future periods, while medium and high-concentration emission pathways generally lead to an increase in potentially suitable habitat compared with low-concentration pathways. Changes in the spatial pattern of potential habitat for C. microphylla From Fig. 7 and Table 5, it can be seen that compared to the contemporary period, the 2030-SSP37.0, 2050-SSP24.5, and 2070-SSP24.5 scenarios show the largest area gains: 2,853,000 km2, 2,212,000 km2, and 2,424,000 km2, respectively. These gains are primarily in northwestern Inner Mongolia, northwestern Xinjiang, Qinghai, Gansu, Northeast China, and southern Shaanxi, Henan, and Hebei. The smallest area of loss is only 0.1 million km2, 15,000 km2, and 10,000 km2, respectively. The 2070-SSP12.6 scenario has the second-largest gain of 1.86 million km2, with a loss of 12,000 km2. The 2030-SSP24.5, 2050-SSP12.6, and 2050-SSP37.5 scenarios also show a gain of 1.86 million km2, with a loss of 12,000 km2. The 2030-SSP24.5, 2050-SSP12.6, and 2050-SSP37.0 scenarios experience the largest losses of 1.927 million km2, 1.036 million km2, and 1.316 million km2, respectively, concentrated in the Qinling Mountains of Shaanxi Province, northern Shanxi Province, southeastern Gansu Province, eastern Qinghai Province, northwestern Xinjiang, Henan, Hebei, northwestern Inner Mongolia, and parts of the three northeastern provinces. The 2070-SSP37.0 scenario shows the least change in the total suitable area, with a reduction of 39,000 km2, indicating minimal impact on the C. microphylla habitat. Conversely, the 2030-SSP37.0 scenario exhibits the largest change, with a reduction of 2,852,000 km2, suggesting that high-concentration GHG emissions have the most significant impact on the C. microphylla habitat.Fig. 7Changes in the area of the potentially suitable zone for Caragana microphylla Lam. plants at different time periods. Map was created using ArcGIS 10.7 (ESRI, Redlands, CA, USA).Table 5Spatial variation in the habitable zones of Caragana microphylla Lam. in different climate scenarios.Climate scenariosIncrease (× 104 km2)Stable (× 104 km2)Shrink (× 104 km2)Change (× 104 km2)2030-SSP12.68.8185.952.944.12030-SSP24.50.446.1192.7192.32030-SSP37.0285.3238.70.1285.22050-SSP12.67.2135.2103.696.42050-SSP24.5221.2237.31.5219.72050-SSP37.02.2107.3131.6129.42070-SSP12.6186.0237.61.2184.82070-SSP24.5224.9237.81.0223.92070-SSP37.045.2197.541.33.9 Changes in the area of the potentially suitable zone for Caragana microphylla Lam. plants at different time periods. Map was created using ArcGIS 10.7 (ESRI, Redlands, CA, USA). Table 5Spatial variation in the habitable zones of Caragana microphylla Lam. in different climate scenarios.Climate scenariosIncrease (× 104 km2)Stable (× 104 km2)Shrink (× 104 km2)Change (× 104 km2)2030-SSP12.68.8185.952.944.12030-SSP24.50.446.1192.7192.32030-SSP37.0285.3238.70.1285.22050-SSP12.67.2135.2103.696.42050-SSP24.5221.2237.31.5219.72050-SSP37.02.2107.3131.6129.42070-SSP12.6186.0237.61.2184.82070-SSP24.5224.9237.81.0223.92070-SSP37.045.2197.541.33.9 Spatial variation in the habitable zones of Caragana microphylla Lam. in different climate scenarios. Future climate scenarios predict a decrease in the concentration of suitable distribution for C. microphylla, particularlyunder high-emission pathways, although some regions may show area expansion. The primary loss of distributional areas is concentrated in the Qinling region of Shaanxi, northwestern Xinjiang, northwestern Inner Mongolia, southeastern Gansu, and parts of northeastern provinces. Conversely, areas of increased suitability are mainly in most of Xinjiang, western and northeastern Inner Mongolia, northern Heilongjiang, southern Jilin, and the Liaoning coastal areas. These areas are bounded by the Qinling Mountains but also extend into the southern regions of Yunnan, Guizhou, Sichuan, Hubei, Anhui, and Jiangsu. Transfer of suitable habitat center of gravity The geometric center, or center of mass, is illustrated in Fig. 8 and Table 6. The contemporary center of Mass is located west of Jining City, Inner Mongolia. For the 2030-SSP37.0 scenario, the center of Mass shifts southwest compared to its current position, with a migration distance of 425.24 km—the farthest among different climate scenarios for the same period. This indicates a significant climate change from the present to 2030 under the high GHG emission scenario of SSP37.0, with a notable impact on the habitat area of the C. microphylla. Conversely, under the SSP24.5 and SSP12.6 scenarios, the center of mass shifts northwesterly, with the closest migration distance observed under SSP12.6 (57.24 km). In the 2050-SSP12.6 and 2050-SSP37.0 scenarios, the center of Mass migrates southwest and northwest, respectively, with migration distances of 88.23 km and 91.10 km, indicating minimal impact on the C. microphylla suitable area under these high and low GHG emission scenarios for 2050. The 2050-SSP24.5 scenario results in the furthest migration (448.51 km), suggesting this scenario is most conducive to the growth of the C. microphylla. The plant’s strong vitality and adaptability further support its potential range expansion during this period. Under the three climate scenarios for 2070, migration distances are approximately similar—351.03 km, 318.96 km, and 344.39 km—primarily westward. The migration direction remains consistent under the SSP12.6 and SSP37.0 scenarios. Overall, the migration paths of the center of mass for C. microphylla. show a westward bias under both current and future climate conditions, with the center of mass remaining in northern China despite future changes.Fig. 8Location of center-of-mass migration in the aptitude zones of Caragana microphylla Lam. in different climatic scenarios at different time periods. Map was created using ArcGIS 10.7 (ESRI, Redlands, CA, USA).Table 6Latitude, longitude, and migration distances of the distribution of Caragana microphylla Lam. plastids in different climate scenarios at different time periods.PeriodClimate scenariosLongitude (°)Latitude (°)Distance (km)Current112.100741.32472030SSP12.6112.650641.547757.24SSP24.5113.219843.0355195.27SSP37.0108.037839.6361425.242050SSP12.6111.508440.630888.23SSP24.5107.491640.7229448.51SSP37.0111.229741.690791.102070SSP12.6108.462641.3348351.03SSP24.5108.873940.6097318.96SSP37.0108.531441.3184344.39 Location of center-of-mass migration in the aptitude zones of Caragana microphylla Lam. in different climatic scenarios at different time periods. Map was created using ArcGIS 10.7 (ESRI, Redlands, CA, USA). Table 6Latitude, longitude, and migration distances of the distribution of Caragana microphylla Lam. plastids in different climate scenarios at different time periods.PeriodClimate scenariosLongitude (°)Latitude (°)Distance (km)Current112.100741.32472030SSP12.6112.650641.547757.24SSP24.5113.219843.0355195.27SSP37.0108.037839.6361425.242050SSP12.6111.508440.630888.23SSP24.5107.491640.7229448.51SSP37.0111.229741.690791.102070SSP12.6108.462641.3348351.03SSP24.5108.873940.6097318.96SSP37.0108.531441.3184344.39 Latitude, longitude, and migration distances of the distribution of Caragana microphylla Lam. plastids in different climate scenarios at different time periods. Discussion Dominant environmental factors affecting the distribution of C. microphylla The MaxEnt model is advantageous for predicting species distribution even with a small sample size37. It has become an important tool in conservation biology, ecology, and evolutionary ecology38. In this study, the MaxEnt model was used to predict the distribution of C. microphylla. The model revealed that the dominant environmental factors affecting the distribution of C. microphylla were annual precipitation (bio-12), slope, coldest quarter precipitation (bio-19), top carbonate or lime content (t_CaCO3), and the coefficient of seasonal variation in temperature (bio-4). Of these, bio-12 demonstrates consistent dominance across both metrics (highest contribution rate and most significant AUC reduction when excluded), indicating that precipitation is the most critical environmental factor, followed by slope and coldest quarter precipitation (bio-19). Discrepancies between contribution metrics and Jackknife results highlight ecological complexity, where structurally important variables (e.g., soil carbonates) May demonstrate predictive replaceability through covarying environmental factors. The closer the slope is to 90°, the higher the survival rate of C. microphylla, which may be due to the fact that there is almost no risk of soil water accumulation in straight slope conditions, while C. microphylla is drought tolerant and waterlogged, and steep slopes can avoid root rot, and it is also related to the shading effect of steep slopes, which can reduce direct sunlight and water evaporation. Notably, while C. microphylla exhibits remarkable drought tolerance, its distribution is paradoxically sensitive to bio-19. This apparent contradiction can be explained by the species’ unique water-use strategy: minimal winter precipitation (0–10 mm) provides critical moisture for early root development prior to summer droughts, whereas excessive winter rainfall (> 50 mm) may lead to soil waterlogging, particularly in carbonate-rich soils (t_CaCO3 = 3–5%). This aligns with findings for other xeric shrubs in arid regions, where winter snowmelt similarly enhances seedling establishment. The geographical distribution of plants at a regional scale is primarily governed by climate, with hydrothermal conditions playing a dominant role39. The distribution is mainly influenced by temperature and precipitation. Yang et al. demonstrated that the geographic distribution of C. microphylla in the northern temperate zone is affected by daily and seasonal variations in temperature and precipitation, as well as temperature and humidity during the growing season. The plant is less sensitive to winter temperature and humidity changes but more responsive to heat extremes, precipitation extremes, and seasonal precipitation variations40. Wang and Ni identified humidity as a key factor affecting the distribution of C. microphylla, particularly during the growing season, with precipitation and average annual precipitation having significant impacts. Additionally, factors such as human activities, interspecific interactions, soil, topography, environmental heterogeneity, and the lagged effects of climate change must be considered. Mechanistic models and analyses of dispersal mechanisms and time lag effects are essential for future research41. Narantsetseg et al. (2014) found that regional differences in C. microphylla density were influenced by climatic drought, grazing regimes, fire disturbances, and interspecific interactions, with precipitation having a particularly significant effect. This underscores that the density of C. microphylla is notably impacted by precipitation42, aligning with the findings of the present study. Changes in spatial distribution patterns and center-of-mass migration analyses of potential habitat areas for C. microphylla plants Understanding changes in potential species distribution patterns in the context of climate change is crucial for assessing its impacts and developing effective conservation strategies to maintain ecological balance43. Climate change is driving a global redistribution of plant and animal species, affecting ecosystems and human societies. Species are migrating to polar regions and higher altitudes, creating new ecosystems and altering existing ones44. Such changes can reduce intraspecific diversity and affect the global evolutionary and adaptive capacity of species, particularly those with low dispersal capabilities45. In this study, we used the MaxEnt model and ArcGIS software to simulate the potential geographic distribution of the C. microphylla plant under various greenhouse gas emission scenarios over different time periods, from the present to the future. Generally, it appears that emission scenarios in the next three periods significantly affect the area of suitable habitat for the C. microphylla. Specifically, the medium- to high-consistency pathways tend to show an increase in the potential suitable habitat area compared to the low-consistency pathways. Throughout all periods, the medium- and high-concentration emission pathways have resulted in a larger potential suitable area than the low-concentration pathways, indicating that higher atmospheric carbon dioxide levels benefit the growth of the bush mallow, consistent with the findings of Knoop and Walker (1985), Polley et al. (2006), and Bond et al. (2011)46–48. Compared to the contemporary period, the scenarios for 2030-SSP37.0, 2050-SSP24.5, and 2070-SSP24.5 show the largest gains and smallest losses in suitable habitat areas. This suggests that climate change during these periods is significant and that the expansion of C. microphylla plants is more favorable under medium to high-level emission scenarios, which have the greatest impact on the plant’s fitness zone. In the 2070-SSP37.0 scenario, the total area of suitable habitat changed the least, with both expansion and contraction areas remaining relatively stable, indicating that the impact on the C. microphylla’s suitable area is minimal in this period. Overall, while the plant May adapt better to future climate conditions, the 2050-SSP37.0 high-concentration emission scenario shows a minimal loss of gained area, whereas the 2070-SSP12.6 low-concentration emission scenario shows a relatively small gain. This might be due to high-concentration emission scenarios leading to increased habitat fragmentation from factors such as drastic and irregular precipitation changes and extreme events49,50. The results suggest that the plant may face survival risks as global warming continues. Under future climate scenarios, the potential distribution of suitable areas for the C. microphylla decreases, and habitat fragmentation increases. The loss of suitable habitat is primarily concentrated in the Qinling region of Shaanxi, northwestern Xinjiang, northwestern Inner Mongolia, southeastern Gansu, and parts of the northeastern provinces. Conversely, suitable areas are expected to increase mainly in Xinjiang, western and northeastern Inner Mongolia, northern Heilongjiang, southern Jilin, and the Liaoning coastal areas, extending beyond the Qinling Mountains into southern Yunnan, Guizhou, Sichuan, Hubei, Anhui, and Jiangsu. Conclusions In this study, we utilized the MaxEnt model and ArcGIS software to identify the dominant environmental factors influencing the distribution of C. microphylla. We analyzed the potential suitable areas for C. microphylla using climate and soil factors to determine its contemporary distribution. We also projected the potential suitable areas and center-of-mass migration of C. microphylla under various greenhouse gas emission scenarios for the next three periods. The model further predicted potential habitat areas, changes in spatial distribution patterns, and center-of-mass migration under different emission scenarios.(1) Under current climatic conditions, the potential habitat for C. microphylla encompasses most of northeastern and northern China, with a smaller portion in northwestern China. This habitat covers approximately 3,329,000 km2 or about 34.7% of China’s land area.(2) The dominant environmental variables affecting C. microphylla distribution are annual precipitation (bio-12), slope, coldest quarterly precipitation (bio-19), carbonate or lime content in the top layer (t_CaCO3), and the seasonal variation coefficient of temperature (bio-4), with a cumulative contribution rate of 67.6%. Optimal growing conditions include annual precipitation of 350–400 mm, a slope of 89°–90°, coldest season precipitation of 0–10 mm, carbonate or Lime content of 3%–5% in the top layer, and a seasonal temperature variation coefficient of 1060–1080. Annual precipitation is the most significant factor affecting the plant.(3) The potentially suitable areas for C. microphylla are currently concentrated in northern China. Under different greenhouse gas emission scenarios for the next three periods, the high-concentration emission pathway shows an increase in potentially suitable areas compared to the low-concentration pathway, suggesting that higher atmospheric carbon dioxide levels benefit shrub and plant growth. Generally, C. microphylla is expected to adapt well to future climatic conditions. However, there is a significant area loss under high emission scenarios, likely due to uncertain factors such as extreme precipitation changes and events affecting the plant. It is hypothesized that variability in precipitation, anthropogenic interventions, and extreme events will impact C. microphylla, leading to reduced suitable habitat area and increased fragmentation. The loss is primarily in the Qinling region of Shaanxi, northwestern Xinjiang, northwestern Inner Mongolia, southeastern Gansu, and parts of northeastern provinces. Conversely, the increase is notable in most of Xinjiang, western and northeastern Inner Mongolia, northern Heilongjiang, southern Jilin, Liaoning’s coastal areas, and regions south of Yunnan, Guizhou, Sichuan, Hubei, Anhui, and Jiangsu.(4) According to the climate model, the center of distribution for the C. microphylla is expected to migrate westward due to climate change. By 2070, the center of distribution will still be located in the northern inland of China, despite future climate changes. The 2050-SSP24.5 scenario shows the largest shift from the current center, with a migration distance of 448.51 km, suggesting it is the most favorable scenario for the C. microphylla’s growth. This period is optimal for the expansion of the C. microphylla’s growth range, as the species exhibits strong vitality and adaptability43 amidst global climate change and increasing grassland scrubland.(5) In order to cope with environmental factors such as climate change, soil and topography, the following management measures should be taken: Based on the special adaptability of C. microphylla to bio-19(0–10 mm optimal, > 50 mm harmful), soil carbonate content (3–5%) and high survival rate under vertical slope conditions, it is recommended to give priority to the protection of existing populations in Inner Mongolia and Xinjiang. Using UAV aerial positioning of winter snow accumulation areas, micro-terrain transformation, combined with satellite remote sensing and ground sensors, a real-time early warning system of ‘precipitation-soil-growth’ was constructed. At the same time, under the SSP37.0 high emission scenario, auxiliary migration should be carried out to potential suitable areas in western China, such as Xinjiang, soil improvement measures such as gravel cover should be taken to alleviate the damage caused by bio-19 excess, and a long-term monitoring network should be established along the westward migration route to dynamically assess population changes and adjust management strategies in time. This integrated management approach takes into account the ecological adaptation thresholds of species and effectively addresses the challenges of distribution change caused by climate, soil and topographic changes. (1) Under current climatic conditions, the potential habitat for C. microphylla encompasses most of northeastern and northern China, with a smaller portion in northwestern China. This habitat covers approximately 3,329,000 km2 or about 34.7% of China’s land area. (1) Under current climatic conditions, the potential habitat for C. microphylla encompasses most of northeastern and northern China, with a smaller portion in northwestern China. This habitat covers approximately 3,329,000 km2 or about 34.7% of China’s land area. (2) The dominant environmental variables affecting C. microphylla distribution are annual precipitation (bio-12), slope, coldest quarterly precipitation (bio-19), carbonate or lime content in the top layer (t_CaCO3), and the seasonal variation coefficient of temperature (bio-4), with a cumulative contribution rate of 67.6%. Optimal growing conditions include annual precipitation of 350–400 mm, a slope of 89°–90°, coldest season precipitation of 0–10 mm, carbonate or Lime content of 3%–5% in the top layer, and a seasonal temperature variation coefficient of 1060–1080. Annual precipitation is the most significant factor affecting the plant. (2) The dominant environmental variables affecting C. microphylla distribution are annual precipitation (bio-12), slope, coldest quarterly precipitation (bio-19), carbonate or lime content in the top layer (t_CaCO3), and the seasonal variation coefficient of temperature (bio-4), with a cumulative contribution rate of 67.6%. Optimal growing conditions include annual precipitation of 350–400 mm, a slope of 89°–90°, coldest season precipitation of 0–10 mm, carbonate or Lime content of 3%–5% in the top layer, and a seasonal temperature variation coefficient of 1060–1080. Annual precipitation is the most significant factor affecting the plant. (3) The potentially suitable areas for C. microphylla are currently concentrated in northern China. Under different greenhouse gas emission scenarios for the next three periods, the high-concentration emission pathway shows an increase in potentially suitable areas compared to the low-concentration pathway, suggesting that higher atmospheric carbon dioxide levels benefit shrub and plant growth. Generally, C. microphylla is expected to adapt well to future climatic conditions. However, there is a significant area loss under high emission scenarios, likely due to uncertain factors such as extreme precipitation changes and events affecting the plant. It is hypothesized that variability in precipitation, anthropogenic interventions, and extreme events will impact C. microphylla, leading to reduced suitable habitat area and increased fragmentation. The loss is primarily in the Qinling region of Shaanxi, northwestern Xinjiang, northwestern Inner Mongolia, southeastern Gansu, and parts of northeastern provinces. Conversely, the increase is notable in most of Xinjiang, western and northeastern Inner Mongolia, northern Heilongjiang, southern Jilin, Liaoning’s coastal areas, and regions south of Yunnan, Guizhou, Sichuan, Hubei, Anhui, and Jiangsu. (3) The potentially suitable areas for C. microphylla are currently concentrated in northern China. Under different greenhouse gas emission scenarios for the next three periods, the high-concentration emission pathway shows an increase in potentially suitable areas compared to the low-concentration pathway, suggesting that higher atmospheric carbon dioxide levels benefit shrub and plant growth. Generally, C. microphylla is expected to adapt well to future climatic conditions. However, there is a significant area loss under high emission scenarios, likely due to uncertain factors such as extreme precipitation changes and events affecting the plant. It is hypothesized that variability in precipitation, anthropogenic interventions, and extreme events will impact C. microphylla, leading to reduced suitable habitat area and increased fragmentation. The loss is primarily in the Qinling region of Shaanxi, northwestern Xinjiang, northwestern Inner Mongolia, southeastern Gansu, and parts of northeastern provinces. Conversely, the increase is notable in most of Xinjiang, western and northeastern Inner Mongolia, northern Heilongjiang, southern Jilin, Liaoning’s coastal areas, and regions south of Yunnan, Guizhou, Sichuan, Hubei, Anhui, and Jiangsu. (4) According to the climate model, the center of distribution for the C. microphylla is expected to migrate westward due to climate change. By 2070, the center of distribution will still be located in the northern inland of China, despite future climate changes. The 2050-SSP24.5 scenario shows the largest shift from the current center, with a migration distance of 448.51 km, suggesting it is the most favorable scenario for the C. microphylla’s growth. This period is optimal for the expansion of the C. microphylla’s growth range, as the species exhibits strong vitality and adaptability43 amidst global climate change and increasing grassland scrubland. (4) According to the climate model, the center of distribution for the C. microphylla is expected to migrate westward due to climate change. By 2070, the center of distribution will still be located in the northern inland of China, despite future climate changes. The 2050-SSP24.5 scenario shows the largest shift from the current center, with a migration distance of 448.51 km, suggesting it is the most favorable scenario for the C. microphylla’s growth. This period is optimal for the expansion of the C. microphylla’s growth range, as the species exhibits strong vitality and adaptability43 amidst global climate change and increasing grassland scrubland. (5) In order to cope with environmental factors such as climate change, soil and topography, the following management measures should be taken: Based on the special adaptability of C. microphylla to bio-19(0–10 mm optimal, > 50 mm harmful), soil carbonate content (3–5%) and high survival rate under vertical slope conditions, it is recommended to give priority to the protection of existing populations in Inner Mongolia and Xinjiang. Using UAV aerial positioning of winter snow accumulation areas, micro-terrain transformation, combined with satellite remote sensing and ground sensors, a real-time early warning system of ‘precipitation-soil-growth’ was constructed. At the same time, under the SSP37.0 high emission scenario, auxiliary migration should be carried out to potential suitable areas in western China, such as Xinjiang, soil improvement measures such as gravel cover should be taken to alleviate the damage caused by bio-19 excess, and a long-term monitoring network should be established along the westward migration route to dynamically assess population changes and adjust management strategies in time. This integrated management approach takes into account the ecological adaptation thresholds of species and effectively addresses the challenges of distribution change caused by climate, soil and topographic changes. (5) In order to cope with environmental factors such as climate change, soil and topography, the following management measures should be taken: Based on the special adaptability of C. microphylla to bio-19(0–10 mm optimal, > 50 mm harmful), soil carbonate content (3–5%) and high survival rate under vertical slope conditions, it is recommended to give priority to the protection of existing populations in Inner Mongolia and Xinjiang. Using UAV aerial positioning of winter snow accumulation areas, micro-terrain transformation, combined with satellite remote sensing and ground sensors, a real-time early warning system of ‘precipitation-soil-growth’ was constructed. At the same time, under the SSP37.0 high emission scenario, auxiliary migration should be carried out to potential suitable areas in western China, such as Xinjiang, soil improvement measures such as gravel cover should be taken to alleviate the damage caused by bio-19 excess, and a long-term monitoring network should be established along the westward migration route to dynamically assess population changes and adjust management strategies in time. This integrated management approach takes into account the ecological adaptation thresholds of species and effectively addresses the challenges of distribution change caused by climate, soil and topographic changes. The C. microphylla is an endemic species on the Inner Mongolia Plateau in China. Utilizing the MaxEnt model and ArcGIS for adaptive analysis and quantitative simulation of its spatial distribution under current and future climates is crucial for maintaining ecological balance, soil and water conservation, biodiversity, and ecological service evaluation in northwest China.