Multi-model evaluation on integrated quality regionalization of the medicinal and edible plant Smilax glabra

Under climate change, shifts in the geographic distribution of medicinal plants and the mechanisms underlying their quality responses provide the scientific basis for forward-looking conservation and utilization strategies. Smilax glabra Roxb., an important medicinal and edible plant in China, is currently threatened by both habitat degradation and rising market demand, yet systematic research on the evolution of its suitable habitat and the spatial patterns of quality formation remains limited. Using 571 spatially rarefied occurrence records, this study integrated ten species distribution algorithms-including an ENMeval-optimized MaxEnt model, generalized linear models, and several machine learning methods-with environmental variables screened by Spearman correlation (with FDR correction) and variance inflation factor analysis. An ensemble forecasting framework was constructed on the Biomod2 platform to simulate habitat suitability for S. glabra under current conditions and four shared socioeconomic pathways (SSP126, SSP245, SSP370, and SSP585) for the 2050s and 2090s. The results showed that the dominant factors limiting its distribution were precipitation of the driest month (Bio14, 56% contribution), annual precipitation (Bio12, 28.9%), and minimum temperature of the coldest month (Bio6, 7.2%), with suitable ranges of 15.36~217.80 mm, 1194.92~4222.90 mm, and -0.34~23.11 °C, respectively. Future projections indicated an overall southwestward shift of the suitable habitat centroid. Stepwise movement analysis revealed that the centroid shifted markedly as early as the 2050s under climate change, with a displacement exceeding 625 km relative to the present under SSP245, and a migration distance exceeding 450 km by the 2090s under SSP585. Area statistics further showed that highly suitable habitat did not contract under the high-emission scenario but instead expanded locally. Astilbin contents from 30 geographic populations were then used to generate a chemical suitability layer via universal co-kriging with leave-one-out cross-validation and kernel density uncertainty assessment. After rigorously excluding areas at high risk of spatial extrapolation, this layer was overlaid with the ecological suitability layer to delineate, for the first time, an Integrated quality regions for S. glabra. The reliable core Integrated quality regions was concentrated in southwestern Guizhou and northern and southeastern Guangxi. The highly quantitative, statistically bias-avoiding, multi-model two-dimensional assessment framework developed here overcomes the limitations of single-suitability evaluations and provides spatially explicit theoretical support for germplasm conservation, habitat-approximation cultivation site selection, and climate-adaptive management of S. glabra.