Ecological niche modeling of Astragalusmembranaceus var. mongholicus medicinal plants in Inner Mongolia, China

Radix Astragali is commonly used in traditional Chinese medicine, and its quality is closely related to ecological factors, such as climate and soil, in the production area. To provide high-quality Radix Astragali to Chinese and foreign markets, we used maximum entropy model and statistical analysis method, combined with data on ecological factors, Astragalus membranaceus var. mongholicus geographical distribution, and index component content to predict the ecological suitability distribution of A. membranaceus var. mongholicus and establish the relationship between astragaloside IV and calycosin-7-glucoside in this species and ecological factors. Subsequently, we could determine the suitability regionalization of high-quality A. membranaceus var. mongholicus in Inner Mongolia, China. The results showed that the standard deviation of seasonal changes in temperature (40.6%), precipitation in October (15.7%), vegetation type (14.3%), soil type (9.2%), and mean sunshine duration in the growing season (9.1%) were the top five most influential factors out of the 17 main ecological factors affecting the distribution of A. membranaceus var. mongholicus. The standard deviation of seasonal changes in temperature, precipitation in October, precipitation in April, soil pH, and mean sunshine duration in the growing season were found to be the key ecological factors affecting the accumulation of astragaloside IV and calycosin-7-glucoside in A. membranaceus var. mongholicus. The regions with the highest-quality A. membranaceus var. mongholicus were distributed in Baotou (Guyang County), Hohhot (Wuchuan County), and central Wulanchabu (Chahar Right Middle Banner, Chahar Right Back Banner, and Shangdu County) and its surroundings in Inner Mongolia. Baotou, Hohhot, and their surrounding areas were the main traditional production areas of A. membranaceus var. mongholicus, and central Wulanchabu was a potentially suitable distribution area of this species. The main production areas were consistent with the actual production base of A. membranaceus var. mongholicus. This study therefore provides a scientific basis to guide the cultivation of A. membranaceus var. mongholicus.

, and 21 ecological factors with correlation coefficients of ≤ 0.8 were retained after preliminary screening (Fig. 3), namely SoilTexture, pH, aspect, mean precipitation in January, February, April, and October (Prec1, 2, 4, and 10), SoilSand, SoilType, mean annual temperature (TempAnnu), vegetation type (VegType), SoilCarbon, Soil-Water, slope, standard deviation of the seasonal changes in temperature (TempSeasonality), Smean4-10, Sun-shineAnnu, altitude, Tmean4-10, range of mean annual temperature (TempRange), mean temperature in March (Tmean3). Accuracy of Maxent model. Table 1 provides the estimates of the relative contributions of the environmental variables to the Maxent model. After 10 reiterations of the model was complete, the mean AUC of the A. membranaceus var. mongholicus test samples was estimated to be 0.989, and the standard deviation was 0.007, thereby indicating that the model was efficient in terms of its prediction capability, and therefore, the accuracy of its prediction results can be guaranteed (Fig. 4).
ecological suitability regionalization of A. membranaceus var. mongholicus in inner Mongolia. As Table 2.
In Inner Mongolia, Guyang County in Baotou City, Wuchuan County in Hohhot City, and their surrounding areas are traditional A. membranaceus var. mongholicus production areas, and large wild populations of the species were once distributed in these areas. This species is being cultivated in these areas since the 1970s, and with the increased demand, the production areas have continuously expanded 27 . At present, Guyang County, Wuchuan County, Harqin Banner in Chifeng City and Urad Front Banner in Bayannaoer City encompass the main cultivation areas of A. membranaceus var. mongholicus in Inner Mongolia, and Xingshunxi Town and Xiashihao Town    According to the response curves of each ecological factor in Fig. 8, TempSeasonality ranging from 11,300 to 13,750 was suitable for A. membranaceus var. mongholicus growth (Fig. 8A). The precipitation in October, April, January, and February was 20, 13, 3, and 3 mm, respectively, which was the most beneficial to A. membranaceus var. mongholicus growth (Fig. 8B,G,N,P); excessive rainfall did not benefit this herb. VegType conducive to the growth of A. membranaceus var. mongholicus was swamp meadows with temperate grasses, Carex, and weeds (Fig. 8C). SoilType suitable for the cultivation of A. membranaceus var. mongholicus was common chestnut calcareous and leached chestnut calcareous soils (Fig. 8D). Smean4_10 and SunshineAnnu ranging from 1653 to 1938 h and 2,750 to 2000 h were beneficial to A. membranaceus var. mongholicus growth (Fig. 8E,J). An altitude and slope of 250-2,400 m and 0°-5°, respectively, and a southwestern aspect of 202.5°-247.5° were conducive to the growth of A. membranaceus var. mongholicus (Fig. 8F,I,O,H). TempAnnu of 44-50 °C and Tmean3 of -3.8 °C were most favorable for A. membranaceus var. mongholicus growth (Fig. 8L,M). Grade-three SoilWater (value of 100 mm/m), loamy sandy soil, and soil with a pH of 7.2 were conducive to the growth of A. membranaceus var. mongholicus (Fig. 8O,K,Q).  (Table 3); three ecological factors (Smean4-10, Tmean3 and pH) have a significant positive correlation with the content of calycosin-7-glucoside (Table 4). Further study on the correlation equations of astragaloside IV and calycosin-7-glucoside contents with the ecological factors were respectively established as y 1 = 0.431 + 0.007 x 1 −0.008 x 2 -0.000026 x 3 ( R 2 = 0.542, P ≤ 0.05; y 1 = Astragaloside IV, x 1 = Prec4, x 2 = Prec10, x 3 = TempSeasonality) and y 2 = − 0.758 + 0.049 x 1 + 0.000025 x 2 ( R 2 = 0.345, P ≤ 0.05; y 2 = calycosin-7-glucoside, x 1 = pH, x 2 = Smean4-10). According to the analysis of the correlation matrix and correlation equations, Prec4, Prec10, and TempSeasonality are the most important ecological factors for the accumulation of astragaloside IV in A. membranaceus var. mongholicus. The Prec4 was beneficial to the accumulation of astragaloside IV in A. membranaceus var. mongholicus within a certain range, whereas TempSeasonality and Prec10 limited its accumulation. Smean4-10 and pH are the most important ecological factors for the accumulation of calycosin-7-glucoside in A. membranaceus var. mongholicus, which was conducive to the accumulation of this compound with a suitable soil environment and light conditions. www.nature.com/scientificreports/ Suitability regionalization of high-quality A. membranaceus var. mongholicus in inner Mongolia. According to the spatial distribution suitability of astragaloside IV and calycosin-7-glucoside in A.

Astragaloside IV and calycosin-7-glucoside contents in
membranaceus var. mongholicus depicted in Fig. 9A Table 5 presents the level of suitability for astragaloside IV and calycosin-7-glucoside distribution at various administrative levels (e.g., banner, district, county, and city). We estimated the positive and negative effects of various ecological factors in Inner Mongolia on the astragaloside IV and calycosin-7-glucoside content to analyze the spatial distribution of the two index components. We consequently found that Guyang County in Baotou City; Wuchuan County in Hohhot City; and central Wulanchabu City (Chahar Right Middle Banners, Chahar Right Back Banners, and Shangdu County) were not only suitable for the growth of A. membranaceus var. mongholicus, but also conducive to the accumulation of the abovementioned index components (Fig. 9C). The quality of A. membranaceus var. mongholicus produced in the aforementioned areas was the highest; this finding can be used as a reference for the cultivation of high-quality A. membranaceus var. mongholicus throughout Inner Mongolia. In addition to traditional production areas, such as Guyang County in Baotou City and Wuchuan County in Hohhot City, Chahar Right Middle Banner, Chahar Right Back Banner, and Shangdu County in Wulanchabu City were potentially the most suitable areas for producing high-quality A. membranaceus var. mongholicus.

Discussion
Considerable effort was made to expand the collection and increase the number of A. membranaceus var. mongholicus samples, which reached > 90% of the total known growth area in this study. Owing to limiting factors, such as climatic and environmental conditions, topography, and time constraints during sampling, some locations where A. membranaceus var. mongholicus grows may have been overlooked and not sampled. This shortcoming is believed to have slightly affected our results; hence, future studies should more comprehensively identify areas where this species grows and incorporate such samples into analyses.
In this study of ecological suitability regionalization for A. membranaceus var. mongholicus, the response curves of the main ecological factors reflect the most suitable environment for the growth of this species. According to the biological understanding of A. membranaceus var. mongholicus, the climatic factors such as temperature (Tempseasonality, TempRange, and Tmean3), precipitation (Prec10, Prec4, Prec1, and Prec2) and sunshine duration (Smean4-10, SunshineAnnu); soil (soil pH, SoilWater, SoilTexture, and SoilType) and topographic (slope, aspect, and altitude) factors; and the biological factor (VegType) analyzed in this study, explain the niche of this species based on the accurate prediction results of the Maxent model. However, for medicinal plants, it is worth noting that the environment suitable for growth does not necessarily accumulate more active components. Many experiments of environmental stress related to medicinal plants have shown that harsh environments may induce them to accumulate more active components [28][29] . This phenomenon explains why the response curves of the main ecological factors involved in this study are favorable to A. membranaceus var. mongholicus growth limited to a certain range.
Correlations between the index components in A. membranaceus var. mongholicus and ecological factors showed that the accumulation of astragaloside IV was primarily affected by climatic factors, including temperature and precipitation (TempSeasonality, Prec4, and Prec10), whereas the accumulation of calycosin-7-glucoside was mainly affected by soil factors and sunshine duration (soil pH and Smean4-10). Compared with the accumulated calycosin-7-glucoside, that of astragaloside IV in A. membranaceus var. mongholicus was more susceptible to the influence of the ecological environment, and different ecological factors had positive and negative effects on its growth. Therefore, as the main bioactive saponin in A. membranaceus var. mongholicus and the saponin extract commonly used in the market 30,31 , astragaloside IV can be obtained in large quantities by artificially controlling the ecological environment of the production area to cultivate high-quality herbs rich in this extract.
In this study of suitability regionalization of high-quality A. membranaceus var. mongholicus, the positive and negative effects of various ecological factors on the accumulation of astragaloside IV and calycosin-7-glucoside were comprehensively analyzed. The most suitable areas for the growth of high-quality A. membranaceus var. mongholicus were found to be Guyang County in Baotou City; Wuchuan County in Hohhot City; and central Wulanchabu City (Chahar Right Middle Banner, Chahar Right Back Banner, and Shangdu County). These regions located in the Yinshan Mountains have an important geographical location (the geographic boundary of east-west mountains in northern China) and unique ecological environment (ecological barriers in central Inner Mongolia). Therefore, the areas are influenced by topography and climate and have distinct seasons with little precipitation but significant differences in temperature, which are conducive to the accumulation of secondary metabolites in A. membranaceus var. mongholicus. The requirements of this species in terms of soil are not specific, but soil texture and the thickness of the soil layer affect the yield and quality of this herb 32 . The prediction results of the study further show that the most suitable distribution areas were mostly located in well-drained, high terrain areas.
At present, an increasing number of Chinese patent medicines, food, health products, and cosmetics are produced using A. membranaceus var. mongholicus as a raw material 33,34 . For example, Shenqi granules from Sun Flower, Huangqi tablets, Huangqi Danggui Ajiao from Tongrentang Chinese Medicine, and Chunjuan Astragalus Cream. Although several high-quality products are produced by reputable companies, healthcare professionals, patients, and consumers are concerned about the questionable botanical ingredients present in various consumer   www.nature.com/scientificreports/ products 35 . By understanding the internal and external links between production, processing, and trade networks, the quality of medicinal plant products may be more clearly discerned in different markets 36 . The value addition of medicinal plants can be introduced at different stages of their production. As the first step of its value chain, the cultivation technology of A. membranaceus var. mongholicus is essential for increasing its value 37 . In our study on the ecological and quality suitability regionalization of A. membranaceus var. mongholicus, the quality of this medicinal plant is closely related to site selection for cultivation. Therefore, the realization of value in the process of transforming raw materials from medicinal plants into high-value products not only depends on actual production, but scientific guidance is also an indispensable step. Notably, this study involved the application of theoretical knowledge from various disciplines, including meteorology, pharmacology, and computer science. Interdisciplinary research to advance the traditional Chinese medicine industry is expected to improve the current understanding on scientific research, through which a variety of scientific research methods may emerge 38 . Combining different methods is likely to be conducive to the industrialization of traditional Chinese medicine. However, this field is still in its infancy, and limited research technology and understanding of the influence of different, seemingly unrelated, factors substantially limit the progress of scientific research 39 . Therefore, although new theories and technologies explored during interdisciplinary research may face challenges, multidisciplinary research can reveal meaningful scientific truths about traditional Chinese medicine.
This study predicted the ecological suitability and distribution areas for cultivating high-quality A. membranaceus var. mongholicus based on environmental factors in Inner Mongolia. However, human disturbances also have important effects on the quality of A. membranaceus var. mongholicus during cultivation 40 . In fact, in terms of both environmental and anthropogenic factors, no single factor is expected to completely control the quality of raw medicinal plant materials. Rather, combinations of these factors likely offer the best control of medicinal material quality 41 . Therefore, in the cultivation and production of A. membranaceus var. mongholicus, in addition to considering environmental factors (e.g., climate, soil, and topography), anthropogenic factors (e.g., soil fertility and irrigation) also need to be considered to achieve high-yield and -quality crops.

conclusions
The suitability regionalization of high-quality A. membranaceus var. mongholicus in Inner Mongolia showed that Guyang County in Baotou City; Wuchuan County in Hohhot City; and central Wulanchabu City (Chahar Right Middle Banner, Chahar Right Back Banner, and Shangdu County) and its surroundings were favorable for the cultivation of A. membranaceus var. mongholicus and accumulation of its active compounds. The quality of A. membranaceus var. mongholicus produced in these areas was the best, and Chahar Right Middle Banner, Chahar Right Back Banner, and Shangdu County in Wulanchabu City were potentially the most suitable areas for producing high-quality crops. This study can be used to inform site selection for cultivating this species to   Table 3. Correlation analysis between the main ecological factors and astragaloside IV. **Significantly correlated at 0.01 level (both sides); *Significantly correlated at the 0.05 level (both sides).

Astragaloside IV (%)
VegType  Acquisition of ecological factor data. In this study, we considered the influence of 74 ecological factors, including climate, soil, topography, vegetation type, and meteorological factors, on the distribution of A. membranaceus var. mongholicus. Climatic data included 59 ecological factors, e.g., monthly mean precipitation (mm), temperature (°C × 10), and sunshine duration (h × 10) from January to December (mm), mean precipitation (mm), temperature (Tmean4-10, °C × 10), and sunshine duration (Smean4-10, h × 10) in the growing season, mean annual sunshine duration (SunshineAnnu, h × 10), and 19 comprehensive climatic factors. These data were based on spatial interpolation of meteorological observation data from 752 surface and automatic meteorological stations in China, collected from 1951 to 2000, with a resolution of 1 km.  www.nature.com/scientificreports/ The soil data comprised eight ecological factors, which were determined according to a 1:100,000 soil map of the People's Republic of China (compiled in 1995) provided by the Second National Land Survey. These factors were soil pH, cation exchange capacity (cmol kg −1 ), sand content (SoilSand, %), clay content (%), soil type (SoilType from FAO-90), soil available water content level (SoilWater), soil texture (SoilTexture, USDA), and organic carbon content (SoilCarbon, %).
Topographic data included three ecological factors, namely altitude (m), slope (°), and aspect with a resolution of 1 km. Vegetation type data included an ecological factor based on vegetation subtype data from a vegetation map of the People's Republic of China (1:100,000) published by the Institute of Botany, Chinese Academy of Sciences. The comprehensive meteorological data comprised three ecological factors, namely the warmth and coldness indexes (°C) derived from Kira's thermal index and the humidity index (mm•°C −1 ) derived from Xu's modified version of Kira's humidity index 42,43 .
The abovementioned data of ecological factors for studying the ecological suitability and quality regionalization of A. membranaceus var. mongholicus were taken from the "Traditional Chinese Medicine Resources Spatial Information Grid Database" (https ://www.tcm-resou rces.com/) provided by the National Resource Center for Chinese Materia Medica of the China Academy of Chinese Medical Sciences (Beijing, China). The relevant information for each ecological factor, including the category, name and type, is presented in Appendix 1 in Supplementary Information 1.   collection of A. membranaceus var. 44 . Based on a full understanding of the regional characteristics of Inner Mongolia, we adopted traditional route and quadrat surveys, taking the village or gacha (level with administrative village) as the smallest sampling unit, to conduct a field survey of A. membranaceus var. mongholicus in the eastern, central, and western regions of Inner Mongolia in 2016. To ensure the uniformity and representativeness of the sample data, we sampled different production areas with significant differences in ecological factors, such as terrain, soil, and vegetation type, according to these two routes. Three to five sampling points were set in each sampling area with a distance of 1 km, and a total of 63 samples of A. membranaceus var. mongholicus were collected in Inner Mongolia. The geographic coordinates of the sampling points were recorded using the handheld GPS. To diminish the influence of different harvesting periods or growth years on the contents of active components in A. membranaceus var. mongholicus, the samples were biennial medicinal materials and were collected in October 2016. Figure 10 shows the survey route and location of the sampling points. The geographical coordinates (latitude and longitude data) of each sampling point are listed in Appendix 4 in Supplementary Information 1. preliminary screening of ecological factors. To diminish the influence of high correlations between ecological factors at the time of MaxEnt modeling, we used Biosim2 to calculate the correlation coefficients between all the ecological factor values extracted from the longitude and latitude of 63 sampling points. According to the correlation coefficient tree diagram calculated using Biosim2, we excluded any ecological factor that had a low correlation with A. membranaceus var. mongholicus growth, based on their potential biological relevance to this species, for each set of highly correlated ecological factors with correlation coefficients > 0.8 45 . We loaded the retained ecological factors into Biosim2 again and repeated the above-mentioned operation until the correlation coefficient between all the retained ecological factors was ≤ 0.8. The ecological factors finally screened are depicted in Fig. 3.
calculation and accuracy testing of Maxent model. The point locality data of A. membranaceus var.
mongholicus and the retained ecological factor data were imported into the MaxEnt model for the calculation.
The model parameters were set as follows: The model was run 10 times, the maximum number of iterations was 1,000,000, the convergence threshold was 0.0005, the random test percentage was set to 10, namely, 90% of the point locality data were randomly selected as training data, and the remaining 10% of data points were the test data. Cross validation (the data set was divided into ten parts, of which 9 were used as training data and 1 as test data in turn for the experiment) was used as the replicated run type, and the max number of background points and the remaining parameters were set as default.
In this study, the receiver operating characteristic curve analysis of the distribution of A. membranaceus var. mongholicus in the model was used to evaluate the accuracy of MaxEnt. The area under the receiver operating characteristic (AUC) was not affected by the threshold, and its value ranged from 0 to 1. The larger the value, the higher was the accuracy of the model. When the AUC was in the range 0.5-0.8, the accuracy of the prediction made by the model was inferior; however, the prediction accuracy was reasonable when the AUC was in the range 0.8-0.9. Finally, when the AUC was greater than 0.9, the model produced reliable and accurate prediction results, and the potential distribution of the species could be accurately predicted 46,47 . The results of 10 training and test sample data repeatedly calculated using MaxEnt are presented in Table 1, and the mean AUC and standard deviation value of the test samples are depicted in Fig. 4. The average growth suitability image of A. membranaceus var. mongholicus obtained through the model was used as the probability layer file of potential A. membranaceus var. mongholicus distribution for studying the ecological suitability regionalization of A. membranaceus var. mongholicus. ecological suitability regionalization of A. membranaceus var. mongholicus in inner Mongolia. The point locality data of A. membranaceus var. mongholicus and its average growth suitability image (distribution probability layer of this species) were simultaneously loaded into ArcGIS. The distribution data of the A. membranaceus var. mongholicus sampling points were used to extract the ecological suitability values in the distribution probability layer, which was rasterized according to the maximum and minimum values of the suitability value. This was done to remove the data outside the range of the sampling points and obtain the region suitable for the cultivation of A. membranaceus var. mongholicus. The natural breaks method in ArcGIS was used to divide the ecological suitability distribution area of the species into four levels: unsuitable (0.00-0.02), secondarily suitable (0.02-0.18), suitable (0.18-0.42), and optimum (0.42-0.90); we used an appropriate color ramp in ArcGIS to indicate the aforementioned levels. Finally, a legend, north arrow, and scale bar were added to complete the map of the ecological suitability of A. membranaceus var. mongholicus at the city level in Inner Mongolia (Fig. 5). To accurately obtain the potential distribution area of A. membranaceus var. mongholicus, based on the ecological suitability of this species in Inner Mongolia, we added county-level administrative data to ArcGIS. Next, we extracted the distribution probability layer of this species by mask to obtain a map of the ecological suitability of A. membranaceus var. mongholicus at the county level. Subsequently, the areas of suitable habitat in the Leagues or Cities in Inner Mongolia were statistically analyzed ( Table 2). www.nature.com/scientificreports/ training presence and background data were randomly permuted. Finally, the model was reevaluated on the permuted data and the contribution of each factor was obtained (Fig. 6). Ecological factors with a contribution rate of > 0% were selected as the main factors to analyze the response curves ( Fig. 8A-Q). Those contributing to the growth of the species were used as the main ecological factors for studying the suitability regionalization of high-quality A. membranaceus var. mongholicus in Inner Mongolia.

content of index components and relationships with main ecological factors. Saponins and
flavonoids are the primary active components of Radix Astragali, and they are valuable indicators for evaluating the quality of Radix Astragali in the Chinese, British, and European pharmacopoeia 3,48,49 . The contents of the saponin astragaloside IV and flavonoid calycosin-7-glucoside in 63 Radix Astragali samples were determined via high performance liquid chromatography according to the Chinese Pharmacopoeia (2015 Edition) (Appendix 3 in Supplementary Information 1). In addition, we used SPSS17.0 statistical analysis software to analyze differences in astragaloside IV and calycosin-7-glucoside content in A. membranaceus var. mongholicus from different production areas in Inner Mongolia. The relationships between astragaloside IV, calycosin-7-glucoside, and the main ecological factors were determined using the correlation matrix (Tables 3, 4). The relationship equations between these index components and the main ecological factors were obtained by stepwise linear regression analysis.
Suitability regionalization of high-quality A. membranaceus var. mongholicus in inner Mongolia. The relationship equations were respectively inputted into ArcGIS's grid calculator to obtain the quantitative distribution layers of astragaloside IV and calycosin-7-glucoside in A. membranaceus var. mongholicus.
Using the spatial calculation function of ArcGIS, the two abovementioned layers were overlain on the ecological suitability distribution layer of A. membranaceus var. mongholicus, and the spatial suitability distribution regions of astragaloside IV and calycosin-7-glucoside in A. membranaceus var. mongholicus in Inner Mongolia were finally obtained. According to the content limits of these index components in 63 A. membranaceus var. mongholicus samples, the spatial suitability distribution regions of the index components were divided into five grades in ArcGIS, represented by a color ramp from blue to red. A map of the spatial distribution of astragaloside IV and calycosin-7-glucoside in A. membranaceus var. mongholicus in the study area was plotted in ArcGIS (Fig. 9A,B). To determine the regions in Inner Mongolia that were suitable for cultivating high-quality A. membranaceus var. mongholicus, we overlaid the spatial distribution layer of the two active ingredients and the administrative distribution data at the county level to find out where these contents are both maximized (Fig. 9C). The administrative areas under various suitability levels of astragaloside IV and calycosin-7-glucoside distribution are presented in the Table 5.