Quality suitability regionalization analysis of Angelica sinensis in Gansu, China

The genus Angelica encompasses 80 species worldwide. Among them, only Angelica sinensis is widely used in China and Japan. To explore the quality and geographical distribution of A. sinensis, we collected 1,530 plants from Gansu Province and analyzed them for their contents of chlorogenic acid (CA), ferulic acid (FA), senkyunolide I(SI), senkyunolide A(SA), senkyunolide H (SH), coniferyl ferulate (CF), ligustilide (LI), and butenyl phthalide (BP) using UPLC. We also assessed the relationship between the ecological environment and quality of A. sinensis through maximum entropy modeling and a geographical information system. The habitat suitability distribution demonstrated that the most influential ecological factors for the growth of A. sinensis were altitude, precipitation during March, May, and December, precipitation during the wettest month, and the soil pH. The most suitable areas for cultivation are concentrated to the south of Gansu Province, including Linxia Hui Autonomous Prefecture, Dingxi City, Tianshui City, south of Wuwei City, east of Gannan Tibetan Autonomous Prefecture, north of Longnan City, and northwest of Pingliang City. The quality suitability regionalization analysis divulged that the most influential ecological factors for the index components of A. sinensis were the altitude, sunshine, rainfall, temperature, and soil pH. The highest quality A. sinensis grow in Dingxi City, Tangchang, Lixian, and Wen counties in Longnan City, Wushan County in Tianshui City, Lintan, Zhouqu, and Zhuoni counties in Gannan Tibetan Autonomous Prefecture, Kangle and Linxia counties in Linxia Hui Autonomous Prefecture. The experiment yielded highly accurate results (accuracy of 0.955), suggesting that the results were consistent with the actual distribution of A. sinensis in Gansu. The inferences of this research will naturally draw the attention of the authorities in the fields of natural resources and agriculture departments and provide a scientific basis for the rational selection of A. sinensis cultivation areas.


Introduction
The root of Angelica sinensis (Oliv.) Diels (Umbelliferae family) is used as a health supplement and drug in Asian countries and a dietary supplement in women's care in Europe [1][2][3]. A. sinensis has been cultivated in China for more than 2,000 years. The root of this herb is an important traditional Chinese medicine, primarily prescribed for tonifying the blood and treating anemia, rheumatism, and menstrual disorders. This plant is primarily cultivated in the Gansu, Yunnan, Sichuan, Shaanxi, and Hubei Provinces in China [4][5][6]. Owing to its high commercial value and large export market, wild A. sinensis has been overharvested. Currently, the cultivated varieties of herbs are preferably used for medicinal purposes. A. sinensis is a typical habitation-dominated medicinal herb. The soil pH, rainfall, temperature, altitude, and other ecological factors greatly influence the active components of A. sinensis [7][8][9]. For economic benefits, farmers have blindly expanded the cultivation areas of A. sinensis, even in places where the environment is not suitable for the growth of A. sinensis, which has eventually deteriorated the quality of A. sinensis. Therefore, the relationship between the quality and ecological environment should be established to obtain quality suitability map. In this study, Maxent and ArcGIS techniques have been used for the first time to conduct a regionalization study on the quality of A. sinensis. Using Maxent modeling, we identified the suitable areas in Gansu for cultivating high-quality A. sinensis species and promoting their cultivation in such suitable areas.

Materials and methods
In this article, 1,530 sites were selected for collecting samples. Sampling point informations are listed in S1 Appendix. The contents of CA, FA, SI, SA, SH, CF, LI, and BP were determined. Using MaxEnt, we calculated the growth suitability of A. sinensis and identified suitable cultivation areas. To correlate the quality of the medicinal plants with their growth environment, we analyzed the positive and negative effects of the habitat conditions on plant growth based on the 8 index component accumulation process and prepared a map of quality suitability of A. sinensis in Gansu.

Survey areas and species occurrence records
The survey areas are in Gansu Province, China, at the confluence of the Qinling Mountains, Loess Plateau, and Qinghai-Tibet Plateau with abundant vegetation. The altitude varies between 1,800 and 3,247 m, which lies between 33˚58 0 N to 38˚40 0 N latitude and 102˚57 0 E to 104˚30 0 E longitude. The survey areas receive an annual average of 36.6-734.9 mm precipitation at an average temperature of 0-15˚C. The sub-tropical monsoon season of the study area extends from June to August, followed by a dry season from November to March.
Based on the preliminary study and planting scale of A. sinensis in Gansu, 1,530 cultivated samples of A. sinensis were collected from Dingxi City, Gannan Tibetan Autonomous Prefecture, Longnan Prefecture, Linxia Hui Autonomous Prefecture, Wuwei City, Tianshui City, Zhangye City, and Lanzhou City between October and November 2018. During the collection process, the latitude, longitude, and habitat information were recorded using a handheld GPS device, and the information on sampling points was stored in the table. All the samples were crushed after drying, sifted, cryopreserved (2-8˚C), and later used to determine the 8 index components. According to the principle of uniform representation, the sampling points were spaced approximately 500 to 800 meters apart. Because the index composition changes with the growth duration, the samples are all 3-year-olds, so that the variables are controlled within a reasonable range. Species occurrence records were displayed in Fig 1.

Acquisition and selection of ecological factors
The data of 60 ecological factors, including meteorological data [10], soil type [11], topographical data [12], and vegetation type [13], were collected. A total of 55 continuous variables and 5 categorical variables were analyzed ( Table 1). The ecological factor database used in this study has been derived from the "Chinese Medicine Resource Spatial Information Grid Database". The map has been downloaded from the department of natural resources of Gansu Province (http://zrzy.gansu.gov.cn/) (resolution-1:650), Map review number: Gan S(2017)64. The attributes for each sampling point were calculated before analysis using correlation software. Correlation coefficients were calculated among 60 ecological factors, and the ecological factors that contained correlation coefficients that were less than 0.8 were retained. Nine ecological factors fulfilled these requirements.

Operation and accuracy of testing by MaxEnt
The sampling information of A. sinensis and 60 ecological factors were added to MaxEnt. A total of 25% of the distribution data were randomly selected as the test set and the rest as the training set [14]. The maximum number of iterations was set as 10 6 , and the convergence threshold was set at 0.00005, while the other parameters were set as the default. These calculations were repeated five times. The contribution of each ecological factor to the growth of A. sinensis was determined. The weight of each climate factor was verified by the Jackknife procedures in MaxEnt. An ROC analysis can evaluate the utility of a model. The area under the curve (AUC) is an effective threshold-independent index to discriminate the presence from absence (or background) [15]. The accuracy of the results is proportional to the AUC value. AUC�0.9 indicates an excellent performance by the model [16]. The results of the simulation output from MaxEnt software ranged from 0 to 1, the closer was the value to 1, the greater was the probability of the existence of the species.

Habitat suitability distribution
The regionalization model demonstrated the distribution of suitable habitat for A. sinensis. We collected the habitat suitability values from 1,530 distribution points. The habitat suitability was then divided into three levels following the Natural Breaks (Jenks) method by marking in three colors for inappropriate areas (0-0.143), appropriate areas (0.144-0.378), and optimum areas (0.379-0.530). The layers with ecological overlays were loaded in ArcMap. The attributes of the setting layer in the symbol system are set based on the classification of habitat suitability. The distribution map of A. sinensis in Gansu Province was extracted by spatial analysis technology in ArcGIS. , coniferyl ferulate (CAS: 63644-62-2), and butenyl phthalide (CAS: 551-08-6) were purchased from Chengdu Reffens Biotechnology Co., Ltd (Chengdu, China). The purity of the reference substances was greater than 98%. The methanol was chromatographic grade; the water was ultrapure, and the other reagents were analytically pure.

Sample preparation
A. sinensis powder (5.0 g) was accurately weighed and placed in a conical flask with a stopper. The sample was extracted with 50% methanol (50 mL) in an ultrasonic bath for 45 min. After refilling the volume, the extract was filtered through filter paper and a 0.22 mm filter membrane for analysis.

The relationship model
A stepwise regression analysis was conducted between the ecological factor data and the index components to establish a prediction model for CA, FA, SI, SA, SH, CF, LI, and BP. The contents of 8 index components were used as the dependent variables and the ecological factors as independent variables to fit a linear relationship and set up the model for quality suitability regionalization for A. sinensis.

Comprehensive quality evaluation
Based on the distribution of habitat suitability of A. sinensis in Gansu Province, the unsuitable distribution areas were removed, and the spatial distribution maps of the index components in the suitable areas were obtained using ArcGIS. According to the regulations specified in the current Chinese Pharmacopoeia (2015), organic acids and volatile oils (the main effective components in A. sinensis) were used as the index ingredients to determine the contents in the A. sinensis, CA, FA, SI, SA, SH, CF, LI, and BP are representative of these compounds. The content of ferulic acid in A. sinensis should not be less than 0.05%, and the content of essential oils should not be less than 0.4% (ml/g). We identified the ecological factors that affected the index content and investigated the effects of the ecological factors on the accumulation of the organic acids and volatile oils. Based on the ArcGIS fuzzy superposition function, the spatial distribution superposition map of all the index components was obtained, and the quality of A. sinensis was comprehensively evaluated.

The key environmental factors and modeling results
The key environmental factors were determined according to the contributions to the modeling process using the jackknife test (

Habitat suitability distribution
The habitat suitability distribution of Angelica sinensis (Fig 2) was plotted after classifying the growth suitability of A. sinensis at different levels. The optimal areas for growth of A. sinensis were located to the south of Wuwei City, Linxia Hui Autonomous Prefecture, Dingxi City, east of Gannan Tibetan Autonomous Prefecture, north of Longnan City, surrounding Tianshui City, and northwest of Pingliang City. The appropriate areas for the growth of A. sinensis include Lanzhou City and southeast of Wuwei City. The remaining areas were predicted to be inappropriate for the growth of A. sinensis. In Fig 1, the purple dots represent the actual presence of A. sinensis. The predicted suitable distribution range was determined based on the actual presence of plants. The response curves represent the relationship between environmental variables and suitability of habitats, they help us understand the ecological niche of a species. The ranges of suitability for environmental variables were identified by the threshold of normally suitable habitats. Response curves of the important ecological factors are illustrated in (S1 File), and the suitable range for each variable is shown in (Table 4). A. sinensis grew at the altitude from 2,000 to 3,000 m, with an optimal altitude of 2760 m, and the precipitation during March was recorded from 5 to 20 mm with optimal precipitation of 16 mm. In contrast, the suitable range of isothermality was from 32 to 38% with an optimal value of 35%.

Results of the determination of content
The contents of CA, FA, SI, SA, SH, CF, LI, and BP in A. sinensis samples agreed with the regulations specified in the current Chinese Pharmacopoeia (2015), and they can be used as

The relationship models of eight index components
The eight regression equations and the regression coefficient of each factor are all significant ( Table 5). The eight index components content can be predicted by the relationship models.

Comprehensive quality evaluation of A. sinensis
Based on the relationship model between the index components and ecological factors of A. sinensis, using the spatial analysis function of ArcGIS v.10.5, the distribution of 8 index components has been estimated and depicted in (Figs 3-10). These figures show that in a suitable area, the contents of FA, SI, SH decrease from south to north, the content of CA decreases from east to west, the content of LI decreases from west to east. From north to south, the content of SA decreases, the distribution regularity of BP is not strong, and the content of CF in the suitable area is more consistent. Quality suitability regionalization of Angelica sinensis are  shown in (Fig 11). Fig 11 indicates

Effects of the meteorological factors
According to the relationship model of eight index components, the content of FA positively correlated with prec12 but negatively correlated with the soil pH. The content of CA positively correlated with index-wi, bio2, and index-hi. The content of SI positively correlated with altitude, bio2, and prec3 but negatively correlated with the soil pH. The content of SH positively correlated with prec3, prec5, and altitude but negatively correlated with rz, bio6, and bio2. The content of SA positively correlated with prec12. The content of CF negatively correlated with rz. The content of LI positively correlated with tmean7. The content of BP positively correlated with prec4 and altitude but negatively correlated with rz. The influence of temperature and moisture on the content of index components of A. sinensis is highly significant [17]. A moderately low temperature aids in the accumulation of organic acids [18], and longer periods of sunshine did not increase the volatile oil composition [19]. This may be related to the fact that A. sinensis is a low temperature, short sunshine plant [20]. The total weight of precipitation in all the ecological factors is 43.93%, Gansu is located in the arid area of northwest China with an annual precipitation of 36.6-734.9 mm, which decreases from southeast to northwest. The precipitation is concentrated in the summer, comprising 50-70% of the annual precipitation. Thus, the moderate precipitation in spring and winter positively correlated with the index composition.

Effects of topographical and soil factors
The results of this study showed that the topographic and soil factors greatly affected the content of the index components. The weight of the altitude on A. sinensis distribution reached 41.88%. The soil pH was 3.8%. All the index components positively correlated with the altitude and negatively correlated with the soil pH. The altitude was within 2,000-3,000 m; the higher the altitude, the greater the accumulated content of the index components. As the altitude increases, photosynthetic products are distributed earlier into the roots, and the rate of accumulation of dry matter increases [21]. The appropriate soil pH of A. sinensis is 6.5-8. Investigation found that the soil in the Gansu production area was alkaline; thus, the content of the index components negatively correlated with the soil pH [22].

Comparison of the study results
The map of the quality suitability regionalization analysis shows that high contents of the index compounds are present in A. sinensis that grow in Minxian, Zhangxian, Weiyuan, Lintao, Dangchang, Lixian, Wenxian, Wushan, Lintan, Zhouqu, Zhuoni, Kangle, Linxia, Jishi Mountain, Tianzhu, and Gulang. A comparison between the areas, selected based on the suitability of regionalization and the traditional farming areas of A. sinensis, suggests that most of the areas with high suitability are those in the traditional growing areas [23]. Although the range of sampling was expanded to maximize the sampled specimens of A. sinensis, some wild A. sinensis specimens from unknown areas were not included in this study owing to some limitations, such as ecological conditions. These unknown areas will be included in our future studies.

Conclusion
The suitable areas for A. sinensis were predicted successfully by the MaxEnt model and Arc-GIS. A. sinensis is primarily distributed to the south of Gansu Province, and the distribution areas are concentrated. The high suitability regions are located primarily to the south of Wuwei City, Linxia Hui Autonomous Prefecture, Dingxi City, east of Gannan Tibetan Autonomous Prefecture, north of Longnan City, surrounding Tianshui City, northwest of Pingliang City, and the sub-suitable areas include Lanzhou City, southeast of Wuwei City. The distribution of A. sinensis was affected by some key environmental variables, including the altitude, precipitation during March, May, December, mean diurnal range, isothermality, annual temperature range, precipitation during the wettest month, and soil pH. The duration of annual sunshine affects the content of ferulic acid in A. sinensis. Longer periods of sunshine are unfavorable for the biosynthesis of ferulic acid. The volatile oil content of A. sinensis is greatly affected by the soil pH, altitude, precipitation in March, April, May, and December, annual sunshine duration, and temperature. Quality suitability regionalization of A. sinensis in Gansu Province shows that A. sinensis specimens from the south of Gansu, such as Dingxi City, Tangchang County, Lixian County, the Wen county in Longnan City, Wushan county in Tianshui City; the Lintan, Zhouqu, Zhuoni county in Gannan Tibetan Autonomous Prefecture; Kangle and Linxia county in Linxia Hui Autonomous Prefecture, contain the best overall quality and the highest levels of active constituents.
Supporting information S1 Appendix. Sampling point informations.