Conclusion
Reducing the uncertainty of SDMs and improving the accuracy of model precision and predicted suitability map is among the unavoidable issues of SDMs applied to biogeography research. The extensive ecological adaptability of orchids may lead to more uncertainty in the correspondence between orchids’ locality and environmental information when the absence of preprocessing occurrence data in orchid SDMs. This has been confirmed in our study. Multiple modeling approaches and validation demonstrate that classification modeling based on physiological characteristics enhances the accuracy of orchid SDMs. Such a feature is also reflected in the suitability maps of the model predictions. Mountainous areas with heterogeneous environments in the Hengduan Mountains hold the hotspots of orchids. But different living forms of orchids are influenced by distinct environmental variables, thus presenting diverse critical regions geographically. In addition, we propose a method to quantify human activity that allows it could be a significant non-natural predictor of input models. It does improve model accuracy in our study, but not a critical variable affecting the geographic distribution pattern of orchids. Importantly we provide a method that can be borrowed and referenced by biogeographic studies of other regions or species, which is flexible and replacement. Our research underscores that, in addition to conventional model optimization methods, classification modeling based on physiological characteristics and incorporation of human activity also make valuable contributions to orchid SDMs. Ultimately, we expect that our study will inspire the thinking of related researchers in modeling SDMs to promote further biogeographic and conservation studies of orchids or other species.