Predicting the potential distribution of Atractylodes lancea in China under CMIP6 climate scenarios using an...

Atractylodes lancea is an important medicinal plant in China, and climate change may affect wild-resource conservation and cultivation planning. However, its current potential ecological suitability, future range dynamics, and relationship with traditional or reported producing regions remain insufficiently clarified under optimized species distribution modelling and CMIP6 climate scenarios. We used 220 occurrence records and environmental variables related to climate, topography, vegetation, and soil to predict the current and future potential suitable habitats of A. lancea in China with an optimized MaxEnt species distribution model. The final parameterization (feature class = LH; regularization multiplier = 3) was selected using partial ROC, the 5% training omission rate, and AICc. The optimized model retained good discrimination against background locations (mean training AUC = 0.923) while reducing the omission rate and AICc relative to the default settings. July precipitation, temperature seasonality, altitude, and vegetation subtype were identified as the main environmental drivers. Under current conditions, suitable habitats were mainly distributed in northern, central, and eastern China, covering approximately 2.72 × 10⁶ km². Under SSP1-2.6 and SSP5-8.5, projections based on the selected BCC-CSM2-MR climate model indicated a general increase in total suitable area, while geographic centroid locations fluctuated among western Shandong, southern Hebei, and northern Henan. The county/district-level overlay showed that several modern or reported producing regions in Hubei and Henan overlapped well with predicted suitable habitats, whereas some Maoshan-related traditional producing regions showed relatively weak current climatic suitability. These results suggest that historical Dao-di (authentic provenance) production regions of A. lancea are not determined solely by present-day climatic suitability, but also reflect historical distribution, cultivation traditions, harvesting and processing practices, and quality evaluation systems. Because the model used one GCM and did not incorporate dynamic land use, harvesting pressure, cultivation management, or medicinal-quality indicators, the results should be interpreted as potential ecological suitability rather than as direct evidence of actual resource distribution, cultivation yield, or geoherbal quality.