[Objective] This study aims to decompose and spatially map the uncertainties in projected biodiversity changes contributed by three key factors Species Distribution Models（SDMs）, General Circulation Models（GCMs）, and Shared Socioeconomic Pathways（SSPs） to better understand the sources of uncertainty in biodiversity change and make conservation and management decisions. [Methods] Using three well-established SDMs and eight critical climate variables, the study developed ecological niche models for ten plant species. Model performance was assessed via the area under the receiver operating characteristic curve（AUC）. The rigorously validated niche models were projected onto future climate scenarios combining five GCMs and four SSPs, generating 60 distinct biodiversity distribution maps for the Loess Plateau（3SDMs×5 GCMs×4 SSPs,2060-2080）. Uncertainty associated with different components was precisely quantified through three-way analysis of variance and spatially mapped using ArcGIS. [Results] All three SDMs demonstrated strong predictive performance, with mean AUC values exceeding 0.8 across all ten species, indicating a high level of accuracy. However, significant differences were observed among projection results from different SDMs, GCMs and SSPs combinations, with an average variation of 34%. Uncertainty decomposition revealed that SDMs and GCMs together contributed approximately 60% of the total uncertainty, far outweighing the uncertainty associated with SSPs in biodiversity simulation. [Conclusion] Comprehensive consideration of uncertainty differences among SDMs, GCMs and SSPs is crucial for climate change adaptation and biodiversity conservation. The quantitative uncertainty mapping methodology introduced in this study can effectively address the limitations of existing research and substantially enhance the scientific rigor and effectiveness of policy formulation.