Altitudinal Shifts as a Climate Resilience Strategy for Angelica sinensis Production in Its Primary Cultivation Region

Abstract

Angelica sinensis, a highly valued Chinese herb renowned for its medicinal and nutritional properties, occupies a distinctive position in montane agriculture. The remote sensing monitoring of grain crops and their driving factors has been extensively studied, yet research on medicinal cash crops, particularly Angelica sinensis, remains limited. This study employed Landsat imagery and a two-step supervised classification method to map Angelica sinensis cultivation areas in southern Gansu Province while also assessing and projecting climate change impacts on its spatial distribution and yield based on the MaxEnt model and CMIP6 models. The results revealed a pronounced upward altitudinal shift in Angelica sinensis cultivation between 1990 and 2020, with the proportion of cultivation areas above 2400 m increasing from 28.75% to 67.80%. Climate factors explained 59.07% of the spatial distribution of Angelica sinensis, with precipitation, temperature, and altitude identified as the key environmental factors influencing its spatial distribution, yield, and growth. Projections for 2020 to 2060 indicate that Angelica sinensis cultivation areas will continue to shift to higher altitudes, accompanied by overall declines in both suitable area and yield. Under the SSP5-8.5 scenario, nearly all suitable areas are expected to be confined to altitudes above 2400 m by 2060, with 41.46% occurring above 2800 m. By 2060, the yield is expected to decrease to 361–421 kg/mu (down 20–31% from 2020) while the suitable area is projected to shrink to 0.98–1.80 million mu (40–60% smaller than 2040) under different scenarios. This study provides new insights into the protection and sustainable management of Angelica sinensis under changing climatic conditions, offering a scientific basis for the sustainable utilization of this valuable medicinal plant.

1. Introduction

Mountains, covering approximately 25% of the global land area, typically support higher biodiversity than other landform units owing to their complex climatic conditions and distinctive topographic characteristics [1,2]. These unique environmental features enable mountain ecosystems to cultivate a wide variety of cash crops, which play a crucial role in sustaining human livelihoods [3,4]. Among these, medicinal plants are particularly significant in mountain agriculture due to their distinct pharmacological properties and economic value [5,6]. A prominent example is Angelica sinensis, a vital traditional Chinese medicinal herb that is not only widely used in the pharmaceutical industry but also holds substantial potential for development in the food and health product sectors, demonstrating notable social benefits [7,8,9]. As the global leader in Angelica sinensis production, China exports large quantities of this herb and its processed derivatives to numerous countries [10], generating substantial economic returns for production regions. This is especially evident in typical mountainous agricultural areas such as southern Gansu Province, where Angelica sinensis cultivation has become a cornerstone of local farmers’ income and regional economic development [11]. Given its significance, investigating the cultivation distribution, growth environments, and climate change responses of Angelica sinensis is crucial for understanding the adaptive mechanisms of mountain agriculture, ensuring the sustainable utilization of traditional medicinal resources, and promoting income growth.

In recent years, remote sensing technology has emerged as a powerful tool in agricultural resource monitoring, providing robust technical support for the precise extraction of crop-related information [12,13]. Traditional ground-based survey methods, particularly in mountainous regions with complex terrain, face significant challenges due to limitations in manpower, material resources, and topographical conditions, making large-scale and high-frequency monitoring difficult [14]. In contrast, remote sensing offers extensive spatial coverage, short revisit intervals, and cost-effective solutions, enabling the efficient acquisition of crop information across expansive mountainous areas and supporting dynamic monitoring [15,16]. By integrating multispectral or hyperspectral data with phenological features, texture characteristics, and other relevant parameters, accurate crop classification can be achieved [17]. Current remote sensing monitoring on crop identification primarily focuses on major grain crops such as rice, wheat, and maize [18,19], while the monitoring of cash crops, particularly high-value medicinal plants like Angelica sinensis, remains relatively limited. During its growth period, Angelica sinensis exhibits dark green leaves with spectral characteristics similar to those of surrounding vegetation [20]. Its small-scale and spatially dispersed cultivation in mountainous regions further complicate accurate identification. Consequently, developing remote sensing methods to accurately extract Angelica sinensis cultivation areas is essential for precision agricultural management.

Climate change has profoundly impacted the growth and distribution of Angelica sinensis [21], especially in mountainous agricultural ecosystems, where these effects are more complex. Specifically, mountain temperatures exhibit strong altitudinal gradients. Angelica sinensis, as a low-temperature and short-sunshine plant, is optimally adapted to high-altitude and cool regions between 1800 m and 3000 m, while cultivation below 1800 m is highly vulnerable to summer heat stress [22]. Moderately low temperatures promote the optimal accumulation of active constituents in Angelica sinensis [23]. In terms of precipitation, Angelica sinensis requires moist environments; insufficient precipitation can trigger premature bolting, reducing yield, while excessive precipitation may cause waterlogged soil and root rot [24]. Extreme weather events, such as droughts, floods, hail, and snowstorms, may reduce or even destroy yields, posing substantial risks to Angelica sinensis cultivation [25]. Furthermore, climate change has altered the suitable habitats and ecological niches of Angelica sinensis [25]. Currently, studies on Angelica sinensis and other medicinal plants have primarily focused on pharmacological properties and chemical composition [26], with limited investigation into their spatial distribution and growth responses to climate change. Global warming poses significant challenges for Angelica sinensis cultivation, highlighting the urgent need for further research to clarify climate change impacts and predict future distribution patterns under various climate scenarios. Such efforts are critical for advancing the conservation and sustainable stewardship of medicinal plant resources.

Southern Gansu Province is the primary production area for Angelica sinensis in China and the largest production base globally. Focusing on southern Gansu Province as the study area, this study systematically analyzes the spatiotemporal dynamics in Angelica sinensis cultivation and its responses to climate change. This study addresses three key objectives: (1) detection of the changes in the spatial distribution of Angelica sinensis cultivation areas using remote sensing technology; (2) elucidation of the influence of climate change on the distribution and yield of Angelica sinensis; and (3) prediction of the changes in the spatial distribution and yield of Angelica sinensis cultivation under future climate scenarios. This study aims to investigate the critical factors influencing the spatial distribution and yield of Angelica sinensis and provide a scientific basis and practical guidance for the sustainable management of mountainous medicinal plants under future climate change.

2. Materials and Methods

2.1. Study Area

The study area is located in southern Gansu Province (Figure 1), within the transitional ecotone where the Qinghai–Tibet Plateau, Loess Plateau, and West Qinling Mountains converge. The terrain is characterized by predominantly alpine mountains with pronounced elevation gradients (2000–3000 m). The climate displays composite characteristics of plateau, continental, and monsoon types, with mean annual temperatures ranging from 4 °C to 15 °C and annual precipitation between 400 mm and 900 mm. Such heterogeneous environmental conditions support diverse agroecosystems dominated by sloping croplands and terraced fields, which create particularly favorable growing conditions for Angelica sinensis cultivation while maintaining high regional biodiversity.

As a mountainous agricultural zone, this region specializes in the cultivation of high-value cash crops as its dominant industry. It serves as the world’s largest production base for Angelica sinensis, contributing over 50% of global demand and more than 80% of China’s total output [27]. The core production area, Minxian County in Dingxi City, has cultivated Angelica sinensis for over 1500 years and currently supplies more than 50% of China’s domestic production and over 90% of the nation’s total exports of this medicinal herb. In 2024, the cultivation area of Angelica sinensis in Minxian County expanded to 300,000 mu (1 mu = 667 m²), driving the total output value of medicinal crops beyond CNY 7 billion. Notably, the medicinal industry constitutes over 60% of local farmers’ disposable income, with this proportion surpassing 90% in core production zones. Angelica sinensis cultivation has emerged as a regional cornerstone industry, playing a crucial role in promoting local socioeconomic development.

2.2. Data Preparation

This study employed multi-source datasets, comprising Landsat remote sensing images, climatic data, Angelica sinensis yield and growth parameters, and soil and terrain data. Specifically, Landsat images (30 m spatial resolution) for the years 1990, 2002, 2006, 2010, 2015, and 2020 were obtained from Geospatial Data Cloud (https://www.gscloud.cn/). The majority of images were under clear-sky conditions, with a limited number of scenes exhibiting minimal and scattered cloud contamination (<2% cloud cover, Table S1). These contaminated scenes were processed using Fmask to remove cloud-affected areas, followed by Landsat 5 gap-filling. Detailed image information is provided in Table S1, including acquisition year, sensor type, scene ID, spatial resolution, and primary spectral bands utilized. All Landsat data preprocessing was conducted using ENVI 5.3 software. The imagery was geometrically corrected through precise registration using 123 uniformly distributed ground control points and projected into the WGS-1984 geographic coordinate system. Radiometric calibration was then performed to convert digital numbers to radiance values, followed by Dark Object Subtraction to derive surface reflectance. Finally, the processed imagery was clipped to the study area boundary.

Climatic data (precipitation, temperature, relative humidity, etc.) were derived from the National Meteorological Information Center (https://data.cma.cn/). Future climate data were provided by the Coupled Model Intercomparison Project Phase 6 (CMIP6). We analyzed three scenarios (SSP1-2.6, SSP2-4.5, and SSP5-8.5) and performed downscaling to 250 m spatial resolution using the Delta method. The yield and growth parameters of Angelica sinensis were acquired from Gansu Minxian Angelica sinensis Research Institute, including yield, trilobite growth height, trilobite density, harvestable growth height, and harvestable density. Soil data were obtained from the Soil Science Database (http://vdb3.soil.csdb.cn/), encompassing five key parameters: soil organic carbon, soil pH, soil texture, soil water content, and total nitrogen. Terrain characteristics were extracted from the Shuttle Radar Topography Mission (SRTM) Digital Elevation Model (DEM) data, acquired from Geospatial Data Cloud.

2.3. Method

This study’s methodology comprises three key components: (1) spatial distribution mapping of Angelica sinensis (1990–2020) using Landsat satellite imagery; (2) analysis of environmental drivers governing Angelica sinensis distribution and growth; and (3) projections of the spatial distribution and yield performance under future climate change.

2.3.1. Spatial Distribution Mapping of Angelica sinensis Cultivation Areas

Based on multi-temporal Landsat images from 1990 to 2020, the spatial distribution of Angelica sinensis was mapped using a two-step supervised classification approach. Supervised classification, a core method in remote sensing image analysis, leverages prior knowledge to extract feature parameters from training samples and construct discriminant functions and decision rules for classification [28]. This method has been widely applied in crop remote sensing monitoring due to its reliability and effectiveness [29]. In this study, the Maximum Likelihood Classification algorithm, a supervised classification approach, was adopted.

In the two-step supervised classification, the first step aimed to extract cultivated land. Based on the spectral characteristics of cultivated land, training samples were selected through visual interpretation, and supervised classification was applied to achieve precise extraction. The second step focused on identifying Angelica sinensis cultivation areas within the extracted cultivated land. By integrating the phenological characteristics and spectral features, a second round of supervised classification was conducted.

Given the limited research on the remote sensing-based identification of Angelica sinensis, this study developed an extraction method through field observations, surveys, and literature reviews. Specifically, Angelica sinensis is a perennial herbaceous plant with a height of 0.40–1.00 m. Its leaves are green or purple, while its stems appear green with a purple tint [7,30]. From April to May, seedlings are typically planted, later than other crops in this region. At this stage, the leaves are small, and the stems and leaves exhibit a purple tint [31], making Angelica sinensis cultivation areas easily identifiable in remote sensing and unmanned aerial vehicle imagery. The main growth period of Angelica sinensis occurs from July to September, during which its leaves turn dark green. However, its spectral characteristics are similar to those of other crops, posing challenges for extraction. During the harvest season from October to November, the leaves of Angelica sinensis turn yellow later than those of other crops in this region, facilitating its remote sensing extraction. Based on these observations, the spatial distribution of Angelica sinensis was extracted for the years 1990, 2002, 2006, 2010, 2015, and 2020.

To validate the extraction accuracy of Angelica sinensis, this study conducted field investigations at 115 sample points across two regions in Minxian County, Gansu Province: Zhongzhai Town (region A) and Puma Town (region B) (Figure 2). The results indicated an extraction accuracy of 86.31% within the field investigation area. Furthermore, the extracted area data were compared with statistical data from 14 townships in Minxian County, the primary production area of Angelica sinensis, to assess the reliability of the extraction method. The results showed that the average extraction accuracy across the 14 townships was 89.19%, highlighting the high accuracy and applicability of this method in the study area. The mapping accuracy of Angelica sinensis spatial distribution for each township is provided in Table S2. Additionally, examples of the original and classified images are displayed in Figure S1.

2.3.2. Influence of Environmental Variables on Angelica sinensis

The Maximum Entropy (MaxEnt) model, a machine learning algorithm, infers species’ ecological requirements by analyzing species distribution data and associated environmental variables [32,33]. Its advantages include multivariate processing capability, adaptability to small sample sizes, high flexibility, and robust performance, and it has been widely applied in species distribution research [34,35,36]. In this study, we applied the MaxEnt model to quantify the contribution rates of each environmental variable to the spatial distribution of Angelica sinensis. A total of 137 spatial occurrence points for Angelica sinensis were obtained from the Global Biodiversity Information Facility (http://www.gbif.org/) and Chinese Virtual Herbarium (https://www.cvh.ac.cn/). Based on previous studies [21,37,38] and expert consultations with the Gansu Minxian Angelica sinensis Research Institute, we evaluated 20 environmental variables spanning four categories: climate, terrain, soil, and human factors (Figure S2). The key parameters of the MaxEnt model used in our study were provided in Table S3. Model performance was evaluated using the area under the curve (AUC) of the receiver operating characteristic (ROC) curve [39]. The 10-fold cross-validation was used to ensure model stability, and the model demonstrated consistent and robust predictive performance (AUC = 0.91 ± 0.02). Additional core outputs, including the current probability distribution map (Figure S3), response curves of key climatic variables (Figure S4), variable importance plot from the jackknife test (Figure S5a), and model evaluation ROC curve (Figure S5b), are provided in the Supplementary Materials.

Based on the top 10 contributing factors identified by the MaxEnt model, we further investigated their effects on the growth of Angelica sinensis. Five key parameters of Angelica sinensis were selected: yield, trilobite growth height, trilobite density, harvestable growth height, and harvestable density. The trifoliate stage represents the early growth phase of Angelica sinensis, characterized by the emergence and unfolding of the third leaf. In contrast, the harvestable stage marks the later growth phase, which is closely associated with the yield and quality of Angelica sinensis. Using a dataset of 134 samples, partial correlation analysis was employed to elucidate the influence of the top 10 factors on Angelica sinensis growth (Equation (1)). Statistical significance was assessed using t-tests. To mitigate inflated interpretations of significance, p-values were adjusted via False Discovery Rate (FDR) correction (Benjamini–Hochberg method) [40,41]. Additionally, Fisher’s z-transformation was employed to derive 95% confidence intervals (CIs).

$$r_{ab\left(c\right)}={\displaystyle \frac{r_{ab}-r_{ac}{r}_{bc}}{\sqrt{\left(1-r_{ac}^2\right)\left(1-r_{bc}^2\right)}}}$$

(1)

where r_ab(c) is the partial correlation coefficient between variables a and b after excluding the influence of variable c. r_ab, r_ac, and r_bc are the correlation coefficients between two variables.

2.3.3. Projections of Spatial Distribution and Yield of Angelica sinensis

Climate and soil conditions, as the pivotal environmental factors influencing the spatial distribution and yield formation of Angelica sinensis, serve as the foundation for predicting its future suitable habitats and yield potential in this study. Regarding climate factors, the four most influential factors identified by the MaxEnt model were considered, namely rainy season precipitation, annual precipitation, warmest month temperature, and mean annual temperature. Future climate projections were derived from 22 CMIP6 Earth System Models (ESMs; Table S4) across three Shared Socioeconomic Pathways (SSPs), SSP1-2.6, SSP2-4.5, and SSP5-8.5, to analyze the impact of low-emission to high-emission scenarios. Specifically, SSP1-2.6 represents a sustainable development pathway exhibiting low socioeconomic vulnerability, minimal mitigation challenges, and low emission levels [42]. SSP2-4.5 defines an intermediate-emission scenario featuring moderate societal vulnerability and intermediate radiative forcing. SSP5-8.5 refers to a high-emission scenario characterized by intensive carbon emissions, rapid population growth, and limited adaptive capacity [43]. To enhance the accuracy of future climate change signals, we applied the Delta-change method to bias-correct CMIP6 model outputs. Precipitation was adjusted using multiplicative correction (Equation (2)), while temperature employed additive correction (Equation (3)) [44]. Soil parameters (soil organic carbon, soil pH, soil texture, total nitrogen, and soil water content) were assumed to be time-invariant in projections given their long-term stability.

$$P_{delta}=P_{obs}\times {\displaystyle \frac{P_{sim\_fut}}{P_{sim\_his}}}$$

(2)

$$T_{delta}=T_{obs}+(T_{sim\_fut}-T_{sim\_his})$$

(3)

where P_delta and T_delta represent the bias-corrected precipitation and temperature, respectively. P_obs and T_obs denote the observed precipitation and temperature. P_{sim_fut}/T_{sim_fut} and P_{sim_his}/T_{sim_his} are CMIP6-simulated precipitation and temperature for future and historical periods, respectively.

To predict spatial distribution, we identified the suitable climatic and soil thresholds for Angelica sinensis cultivation (Table S5) through the integrated analysis of historical growing conditions (1990–2020), literature synthesis, field surveys, and expert consultations with the Gansu Minxian Angelica sinensis Research Institute. Suitable habitats for Angelica sinensis under future climate scenarios were mapped through the spatial overlay of climate projections and soil parameters.

For yield prediction, we established a multivariate regression model for Angelica sinensis yield incorporating climatic and soil factors (Equation (4)). The model was validated using annual-scale historical yield and climate data from 1990 to 2020 (excluding the extreme climate event in 2012), demonstrating significant predictive capability (R² = 0.71, p < 0.01) (Figure S6). Subsequently, the validated model was applied to simulate yield responses under projected climate change scenarios.

$$Yield={\beta }_0+\sum _{i=1}^4{\beta }_i{C}_i+\sum _{j=1}^4{\gamma }_j{N}_j+FE+ϵ$$

(4)

where C_i and N_j are the linear and nonlinear terms of the i-th and j-th climatic factors, respectively. β_i and γ_j are regression coefficients. Β₀ is the intercept. FE represents the fixed effects from soil factors. ϵ is random error.

3. Results

3.1. Spatiotemporal Dynamics of Angelica sinensis Cultivation

Over the past 30 years, both the total cultivation area and yield of Angelica sinensis have exhibited an increasing trend, with a notable upward shift in cultivation regions to higher altitudes. In 1990, the cultivated area of Angelica sinensis was 205,123.46 mu, with a yield of about 223 kg/mu (Figure 3a). The cultivation region was mainly concentrated in the river valley area of western Minxian County and the mountainous area of northern Dangchang County (Figure 4a). Notably, the cultivation was primarily distributed at relatively lower altitudes, with 39.04% at 2200–2400 m, 32.21% below 2200 m, and 24.00% at 2400–2600 m, while areas above 2600 m accounted for only 4.75% (Figure 3b).

In 1990–2002, the cultivated area and yield increased by 31,363.78 mu and 131 kg/mu, respectively (Figure 3a). The cultivation of Angelica sinensis in the eastern mountainous areas of Minxian County exhibited expansion markedly (Figure 4b), with the proportion of land at altitudes between 2400 and 2600 m and above 2600 m rising to 26.87% and 8.64%, respectively. Between 2002 and 2010, both the cultivated area and yield of Angelica sinensis continued to increase. New cultivation zones have been reclaimed in the eastern mountainous areas of Minxian County and Lintan County (Figure 4d–k), resulting in the proportion of areas above 2400 m reaching 52.84%.

From 2010 to 2020, the cultivation area of Angelica sinensis expanded to 452,942.18 mu, with the yield reaching 524 kg/mu (Figure 3a). The cultivation area declined in the western valley area of Minxian County (the primary production area in 1990) (Figure 4g–i), while the primary cultivation zones shifted to the eastern mountainous area of Minxian County (Figure 4j–l), the southeastern part of Lintan County, part of Zhuni County, and the northern mountainous area of Dangchang County (Figure 4f). By 2020, the proportion of cultivation areas above 2400 m reached 67.80%, 2.36 times higher than in 1990. Notably, 10.56% of Angelica sinensis cultivation occurred above 2800 m, reflecting a clear upward migration along the altitude gradient over the three decades.

3.2. Impacts of Environmental Factors on the Spatial Distribution and Growth of Angelica sinensis

The spatial distribution of Angelica sinensis was determined by the combined effects of climate, soil, terrain, and human factors. The MaxEnt model indicated that these four environmental variable groups explained 97.80% of its spatial distribution. Specifically, climate factors exhibited dominant control (59.07%), with rainy season precipitation being the principal determinant (21.62%), followed by annual precipitation (9.81%), warmest month temperature (9.23%), and mean annual temperature (8.61%) (Figure 5a). Soil factors formed the secondary contribution (24.44%), in which soil water content (7.21%), soil organic carbon (5.56%), and soil pH (4.54%) were the key drivers (Figure 5c). Terrain factors contributed to 12.91% of the spatial distribution, predominantly influenced by altitude (8.63%), while slope and aspect exhibited minimal effects (<2.50%) (Figure 5b). Human factors showed a limited contribution rate (1.38%), with market distance (0.69%) marginally exceeding both population density and road density (each < 0.50%) (Figure 5d). Overall, climate factors (especially precipitation and temperature) dominated Angelica sinensis distribution, with secondary effects from soil and terrain factors through microenvironmental modulation, while human factors remained negligible.

Partial correlation analysis revealed differential influences among the top 10 environmental factors on Angelica sinensis yield (Figure 6). Yield was most strongly correlated with soil water content (r = 0.87, 95% CI [0.81, 0.90], FDR-adjusted p < 0.01), highlighting water availability as the primary yield constraint. This was followed by altitude (r = 0.68, 95% CI [0.57, 0.76], FDR-adjusted p < 0.01), suggesting that the favorable light-temperature conditions at higher altitudes may enhance yield. In terms of precipitation factors, both rainy season precipitation (r = 0.65, 95% CI [0.54, 0.75], FDR-adjusted p < 0.01) and annual precipitation (r = 0.63, 95% CI [0.51, 0.73], FDR-adjusted p < 0.05) showed significant positive correlations, further confirming the crucial role of water supply in yield determination. In contrast, soil pH exhibited a weak negative correlation (r = −0.13, 95% CI [−0.30, 0.05], FDR-adjusted p > 0.05). Although statistically insignificant, this relationship may suggest potential adverse effects under elevated pH conditions in this region.

At the trilobate stage (Figure 6), altitude emerged as the predominant environmental factor, demonstrating significant positive correlations with both growth height (r = 0.87, 95% CI [0.81, 0.90], FDR-adjusted p < 0.01) and density (r = 0.72, 95% CI [0.62, 0.79], FDR-adjusted p < 0.01). Meanwhile, trilobite growth height showed higher positive correlations (r > 0.5) with several environmental factors, exhibiting the strongest correlations with water-related factors (soil water content: r = 0.66, 95% CI [0.54, 0.75], FDR-adjusted p < 0.01; rainy season precipitation: r = 0.56, 95% CI [0.43, 0.67], FDR-adjusted p < 0.01), followed by temperature conditions (warmest month temperature: r = 0.54, 95% CI [0.40, 0.65], FDR-adjusted p < 0.01) and soil nutrients (total nitrogen: r = 0.53, 95% CI [0.39, 0.64], FDR-adjusted p < 0.01; soil organic carbon: r = 0.52, 95% CI [0.38, 0.64], FDR-adjusted p < 0.01). Similarly to yield, a negative correlation was observed with soil pH (r = −0.25, 95% CI [−0.41, −0.08], FDR-adjusted p > 0.05). With respect to trilobite density, it was primarily regulated by a combination of precipitation and soil nutrients, with rainy season precipitation (r = 0.74, 95% CI [0.65, 0.81], FDR-adjusted p < 0.01) showing the strongest correlation, followed by total nitrogen (r = 0.68, p < 0.001) and annual precipitation (r = 0.54, 95% CI [0.40, 0.66], FDR-adjusted p < 0.01).

During the harvestable period (Figure 6), although the influence of altitude on growth indices weakened, it still maintained significant positive correlations with both growth height (r = 0.66, 95% CI [0.54, 0.75], FDR-adjusted p < 0.01) and density (r = 0.52, 95% CI [0.37, 0.63], FDR-adjusted p < 0.05). Compared to the trilobate stage, temperature showed enhanced influence during this stage, with mean annual temperature exhibiting stronger correlations with both growth height (r = 0.53, 95% CI [0.39, 0.65], FDR-adjusted p < 0.05) and density (r = 0.62, 95% CI [0.49, 0.71], FDR-adjusted p < 0.05). Furthermore, harvestable growth height remained positively correlated with soil water content (r = 0.57, 95% CI [0.43, 0.67], FDR-adjusted p < 0.05) and rainy season precipitation (r = 0.54, 95% CI [0.40, 0.65], FDR-adjusted p < 0.05). Similarly, harvestable density showed positive correlations with soil water content (r = 0.57, 95% CI [0.43, 0.67], FDR-adjusted p < 0.05) and soil organic carbon (r = 0.54, 95% CI [0.40, 0.66], FDR-adjusted p < 0.05) but exhibited a negative correlation with soil pH (r = −0.36, 95% CI [−0.50, −0.19], FDR-adjusted p > 0.05). Comprehensive analysis revealed stage-dependent responses in Angelica sinensis to environmental factors: altitude and water availability were dominant factors during early growth stages, while temperature exerted increasing influence during the harvestable period.

3.3. Projections of Suitable Cultivation Areas and Crop Yields Under Future Climate Scenarios

By 2040, Angelica sinensis suitable cultivation areas showed marked upward altitudinal shifts along the SSP1-2.6 to SSP5-8.5 emission gradient (Figure 7a–c). Specifically, under the SSP1-2.6 scenario, the suitable cultivation area was 3,030,866.96 mu, characterized by a core–periphery distribution pattern. The core areas were concentrated in the eastern mountainous regions and western valley areas of Minxian County, while fragmented distributions occurred in western Lintan County, southern Zhangxian, and northern Dangchang County. Altitudinal analysis revealed that 2600–2800 m (34.95%) and 2400–2600 m (28.83%) constituted the suitable altitude zones, with only 13.05% suitability above 2800 m (Figure 7g). The SSP2-4.5 scenario maintained similar spatial patterns (2,579,551.92 mu suitable area), predominantly within 2400–2800 m altitudes. Notably, compared to SSP1-2.6, suitable zones below 2400 m decreased by 8.14 percentage points, while those above 2800 m increased by 4.35 percentage points, indicating an upward migration. Under the SSP5-8.5 scenario, suitable areas sharply contracted to 2,436,716.01 mu, becoming concentrated in eastern Minxian County and western Lintan County, with only sporadic distributions in adjacent Zhangxian and Dangchang Counties. More critically, the altitudinal distribution underwent fundamental changes: the 2600–2800 m zone increased to 37.23%, while areas above 2800 m surged to 29.24%.

By 2060, climate warming will further shift the suitable cultivation zones of Angelica sinensis to higher altitudes, accompanied by a substantial reduction in suitable area compared to 2040 (Figure 7d–f). Under the SSP1-2.6, SSP2-4.5, and SSP5-8.5 scenarios, the suitable cultivation areas are projected to be 1,802,347.18 mu, 1,276,558.54 mu, and 985,405.00 mu, respectively, primarily concentrated in Minxian and Lintan Counties. Compared to 2040, the suitable altitude ranges under all scenarios exhibit a pronounced upward shift. In the SSP1-2.6 and SSP2-4.5 scenarios, the 2600–2800 m altitude emerges as the most suitable zone, accounting for 48.62% and 41.64% of the total area, respectively, while the proportion above 2800 m increases to 26.31% and 34.39% (Figure 7g). Notably, under the high-emission SSP5-5.8 scenario, the suitable cultivation area demonstrates a more marked shift toward higher altitudes, with regions above 2800 m representing the dominant altitudinal zone (41.46%), surpassing all other altitude ranges.

During the 2020–2060 period, mean yields of Angelica sinensis under SSP1-2.6, SSP2-4.5, and SSP5-8.5 scenarios are projected at 437 kg/mu, 428 kg/mu, and 408 kg/mu, respectively, all surpassing the 1990–2020 baseline average (338 kg/mu). Under the SSP1-2.6 scenario, the yield shows an overall declining trend (−0.46 kg mu⁻¹ yr⁻¹) while maintaining yields above 400 kg/mu (Figure 8). More precisely, the yield exhibits a decreasing trend during 2020–2039, followed by a recovery phase from 2039 to 2054, before resuming its downward trajectory post-2054. The SSP2-4.5 scenario exhibits a more substantial yield reduction (−2.00 kg mu⁻¹ yr⁻¹), peaking in 2030 (435 kg/mu). Similarly to the SSP1-2.6 scenario, a yield recovery is observed during 2039–2054. Most severely, the SSP5-8.5 scenario projects maximum yield declines (−2.83 kg mu⁻¹ yr⁻¹), decreasing to 361 kg/mu by 2060. These trajectories demonstrate that climate warming imposes progressively greater production pressures, with higher-emission scenarios exacerbating yield losses despite transient recovery periods.

4. Discussion

4.1. Spatial Distribution Mapping of Angelica sinensis

Remote sensing technology has been widely applied in grain crop monitoring [45,46], but its application to high-value medicinal crops, particularly Angelica sinensis, remains limited. In this study, we developed a novel extraction method specifically for southern Gansu Province, the primary region of Angelica sinensis production in China. Previous approaches relying on “dark green, striated” spectral features often lead to misclassification between Angelica sinensis and forestland [20]. Our study effectively resolved this classification challenge through a two-step supervised classification approach: (1) initial extraction of cultivated land, followed by (2) precise identification of Angelica sinensis by integrating its distinctive phenological and spectral traits. It should be noted that Angelica sinensis cultivation extends to other provinces, including Sichuan, Yunnan, and Hubei Provinces [31,47], but the proposed methodology has only been validated in southern Gansu Province. Given regional variations in climatic conditions and cultivation practices, its broader applicability to other production areas requires further exploration.

4.2. Impacts of Climate Change on Angelica sinensis

From 1990 to 2020, Angelica sinensis cultivation exhibited a pronounced upward altitudinal shift, with high-altitude cultivation (>2400 m) increasing from 28.75% to 67.80% of total production areas. Spatial analysis revealed heterogeneous patterns in Minxian County: marked contraction in western valleys (Figure 4g–i) contrasted with expansion in eastern highlands (Figure 4j–l). These distributional changes showed strong covariation with regional climate warming–wetting trends (Figure 9). Over the past century, China has experienced a warming trend of approximately 0.1 °C/10 a, which has accelerated to at least 0.25 °C/10 a in the last 60 years [48]. From 1961 to 2014, the northwestern region of China (including our study area) recorded a temperature increase of ~0.9 °C and a precipitation rise of 28.2 mm. Notably, southern Gansu exhibited a more pronounced warming trend, with temperatures increasing by 10–30% during 1991–2014 compared to the 1961–1990 baseline [49]. This warming trend exceeds the national average [50]. Our observations further indicate that southern Gansu has experienced a temperature rise of 0.1–0.4 °C/10 a in both mean and warmest month temperatures, alongside an annual and rainy season precipitation increase of 10–30 mm/10 a (Figure 9). Angelica sinensis displays marked temperature sensitivity in both seedling and mature growth stages [22,25]. High temperatures trigger premature bolting and flowering, reducing medicinal quality, with the warmest month temperature serving as the critical determinant for successful summer survival [51]. By 2020, rising temperatures had rendered western Minxian valleys—the primary production area in 1990—climatically unsuitable for cultivation. Concurrently, Angelica sinensis cultivation expanded in the eastern mountainous area of Minxian and adjacent Lintan County. Compared to Minxian’s traditional cultivation zones, these areas’ slightly higher altitude provided more favorable growing conditions under climate change, making it an emerging production hotspot for local farmers.

Climatic factors explained 59.07% of the spatial distribution in Angelica sinensis (Figure 5). The rainy season precipitation showed the highest contribution rate (21.62%), as sufficient precipitation is critical for ensuring growth. Temperature, precipitation, and altitude were identified as the dominant determinants, consistent with previous studies [21,52]. The yield and growth indices of Angelica sinensis showed positive correlations with these three factors. Yield improvements occurred when increases in temperature and precipitation remained within their respective optimal ranges, which was confirmed by a previous study [53]. Elevated temperatures reduce frost periods, enabling earlier seedling transplantation and extending the growth cycle of Angelica sinensis, while increased precipitation enhances soil moisture and mitigates drought stress [54]. Meanwhile, we observed a negative correlation between Angelica sinensis yield/growth and soil pH, consistent with the findings of Xu et al. [21]. Furthermore, extreme weather events impacted Angelica sinensis productivity, as exemplified by the 2012 yield reduction (126 kg/mu) caused by hailstorms.

Our analysis also revealed distinct stage-specific effects of precipitation and temperature on Angelica sinensis growth. Precipitation showed a stronger correlation with growth at the trilobate stage than at the harvestable stage (Figure 6). Seedling growth requires sufficient water, especially during the 3–5 leaf stages, to maintain optimal growth. Water deficit during this phase impairs shoot growth and root development [54]. During medicinal maturity, however, plants exhibit enhanced drought tolerance and reduced water requirements [54]. Regarding temperature, temperature correlated more strongly with growth at the harvestable stage than at the trilobate stage (Figure 6). Optimal temperature and light during later stages enhance photosynthesis and promote root tuber expansion.

From 2020 to 2060, projections indicate persistent climate warming–wetting trends, which has been extensively documented in previous studies [49,55,56]. Compared to the baseline period (1986–2005), China’s mean temperature is projected to rise by 1.31–1.45 °C (2021–2040) and 1.75–2.66 °C (2041–2060) under different SSP scenarios, with greater warming under higher-emission pathways [57]. In southern Gansu Province—the focus of this study—the warming–wetting trend is particularly evident, aligning with broader regional climate patterns [49,58]. Since Angelica sinensis thrives in low-temperature environments, climate warming will induce premature bolting, which reduces its medicinal quality and poses significant challenges to sustainable resource utilization. Our findings demonstrate that in response to climate change, the suitable cultivation areas for Angelica sinensis are projected to shift toward higher altitudes, consistent with previous findings [25]. Due to the progressively intensified warming magnitudes under SSP1-2.6, SSP2-4.5, and SSP5-8.5 scenarios [58], the proportion of suitable high-altitude (>2800 m) cultivation areas shows corresponding gradual increases across these three emission scenarios (Figure 7g). In southern Gansu Province, the limited availability of high-altitude areas, combined with climate warming, is projected to reduce suitable cultivation areas, with the magnitude of reduction increasing with emission intensity (SSP5-8.5 > SSP2-4.5 > SSP1-2.6).

Globally, elevated atmospheric CO₂ concentration and global warming have promoted vegetation greening [41,59,60]. Our high-resolution analysis reveals that regional-scale greening may arise from climate adaptation-driven ecological transformations. This is exemplified by the climate-induced upward shift in Angelica sinensis cultivation. The high-altitude regions within the study area, particularly in eastern Minxian, Lintan, and Zhuoni Counties, are predominantly grassland ecosystems where traditional agriculture focuses on animal husbandry [61]. To accommodate Angelica sinensis cultivation, pastoral grasslands have been converted to cultivated areas, resulting in regional increases in vegetation cover that manifest as localized greening. Notably, the environmental carrying capacity of high-altitude regions in southern Gansu remains limited. Although such land-use conversion may generate short-term economic benefits and vegetation greening, it risks triggering long-term environmental degradation, including soil erosion and land deterioration, ultimately compromising regional ecological security and biodiversity conservation [62]. Our findings underscore the necessity for a multiscale analysis of greening patterns, elucidating interactions between macroscale climatic forcing and microscale adaptive land-use transformations.

4.3. Suggestions for Angelica sinensis Cultivation and Management

To mitigate climate change impacts on Angelica sinensis and ensure sustainable production, the following measures are recommended. First, a previous study found that premature bolting in medium-sized seedlings was reduced by 5% and 5.5% under 75% and 50% shade nets, respectively [63]. Given its preference for cool, moist environments, shading methods (e.g., shade nets) should be employed under warming conditions to reduce field temperatures and minimize water loss. Second, with traditional growing areas becoming unsuitable for Angelica sinensis due to warming, farmers can transition to more heat-tolerant crops like Codonopsis pilosula and Astragalus membranaceus. Third, as extreme weather events are projected to increase in frequency, authorities should strengthen meteorological monitoring and forecasting systems to establish more comprehensive early warning mechanisms for Angelica sinensis cultivation. Disaster impacts can be mitigated by artificial rain dispersal, hail suppression, and precipitation enhancement techniques.

4.4. Limitations and Prospects

This study systematically evaluates climate change impacts on Angelica sinensis, yet the following limitations remain. First, the fragmented distribution of Angelica sinensis cultivation in complex mountainous terrain may lead to detection omissions with current remote sensing monitoring. Future studies will integrate high-resolution satellite imagery to improve extraction accuracy [64]. Second, the spatial distribution and yield prediction methods assume constant soil properties. However, these factors may vary dynamically due to soil improvement techniques, new cultivar development, and farming innovations. Such static assumptions could introduce prediction biases, which should be addressed in future research by incorporating these dynamic variables. Third, while this study has identified key environmental factors affecting the spatial distribution of Angelica sinensis, additional variables such as pest/disease incidence and agricultural management practices were not considered due to data availability constraints. Future studies will incorporate more possible factors to improve the model’s comprehensiveness. Finally, climate projections from CMIP6 also introduce uncertainties. Our study employs an equally weighted ensemble mean, and subsequent studies will examine individual model performance across the study area while developing weighted ensemble approaches for more accurate projections.

5. Conclusions

This study investigated the spatial distribution of Angelica sinensis from 1990 to 2020, evaluated climate change impacts, and projected future spatial distribution and yield under multiple climate scenarios. Our results revealed a clear altitudinal migration of cultivation zones, with the proportion of high-altitude (>2400 m) production areas increasing from 28.75% to 67.80% between 1990 and 2020. This shift was strongly correlated with regional warming–wetting trends. Climate factors accounted for the majority of influence on spatial distribution (59.07%), followed by soil (24.44%) and terrain (12.91%) factors. The key determining factors included rainy season precipitation, annual precipitation, warmest month temperature, mean annual temperature, and altitude, which all exhibited significant positive correlations with yield and growth indices. Projected climate change is expected to push suitable cultivation zones to progressively higher altitudes. Under SSP1-2.6, SSP2-4.5, and SSP5-8.5 scenarios, the proportion of suitable cultivation areas above 2800 m is predicted to reach 26.31%, 34.39%, and 41.46%, respectively, by 2060. Agricultural management practices such as shade nets should be implemented. These measures can regulate microclimates, reduce thermal stress, and stabilize phytochemicals, thereby supporting sustainable resource management.