Abstract: Based on the survey data of a 4hm² fixed monitoring plot in Pangquangou Nature Reserve on Guandi Mountain, we analyzed correlation between different lifeform species distribution and two habitat factors-terrain and soil in a secondary Picea forest using variance decomposition (RDA) and fitted relationships between the dominant species of trees, small trees and shrubs and principal components of habitat factors by generalized additive model (GAM). The results showed that: (1) Terrain and soil together explained the distribution of trees (53.34%), small trees (55.65%) and shrubs (45.83%), including an independent explanation by terrain (8.36%, 5.06% and 5.48%, respectively), an independent explanation by soil (31.12%, 44.22% and 32.04%, respectively), and a joint part explanation (13.86%, 6.37% and 8.31%, respectively); (2) The principal component analysis of 13 habitat indexes showed that the first four principal components represented 38.76%, 25.73%, 19.41% and 8.93%, respectively, with a cumulative contribution rate of 92.83%; (3) For different species and lifeform species, the degree of correlation between their distribution and the four principal components was different by GAM fitting. The deviations explained by the model were 0.76%-29.00% for tree layer, 0-20.30% for small tree layer and 0.10%-23.50% for shrub layer. Each species and life form were better associated with PC1, and altitude, aspect, available K, N and Mg played an important role in species distribution. The result indicated that the effects of habitat factors such as terrain and soil on the different lifeform species distribution were different, which promoted the coexistence and biodiversity maintaining of species in the secondary Picea forest on Guandi Mountain.