Abstract: Acacia confusa Merr. is the dominant forest type on Fuying Island in Xiapu County, China. This study investigated the niche characteristics and interspecific associations of the main species in the A. confusa community shrub layer using data from 92 shrub plots. Niche analysis was conducted using Levins niche breadth (B_L), Shannon niche breadth (B_S), Schoener niche similarity (C_ik), and Pianka niche overlap (O_ik). Interspecific associations were examined using the variance ratio (VR), χ² test, and association (A_C), Ochiai (O_I), Dice (D_I), and Jaccard (J_A) coefficients. Results showed that: (1) Among the 20 species analyzed, Callicarpa pedunculata R. Br. exhibited the largest niche breadth and held an absolute competitive advantage in the community. (2) Niche similarity and overlap among species in the shrub layer were low, suggesting efficient resource utilization by each species. (3) Greater niche breadth showed a general correspondence to higher niche similarity and overlap, although this relationship did not show an absolute positive correlation. (4) The W test revealed that the overall interspecific associations were insignificantly negative, reflecting a dynamic stage of succession with instability. (5) The ratio of positive to negative associations among species was 0.98, and the low significance rates across multiple association tests indicated weak interspecific associations and limited competition. (6) The A_C, J_A, O_I, and D_I coefficients showed significantly positive correlations with niche similarity and overlap, indicating that stronger positive associations corresponded to greater niche similarity and overlap among species. These findings suggest that the main shrub species within the A. confusa community efficiently utilize environmental resources, with relatively weak interspecies competition. However, young A. taiwanensis trees failed to establish a dominant position in the community. The overall stability of the community is low, with potential for reverse succession. Therefore, targeted artificial management strategies should be adopted.