Bamboos are typical clonal plants commonly used as food for giant pandas. Studies of the clonal structure of typical bamboo populations are of both theoretical and practical importance for bamboo forest production and tending. Studies have attempted to predict its population succession for the protection of giant pandas. In this study, simple sequence repeat fingerprints were used to reveal the clonal structure and diversity of Bashania fargersii populations at three different genet ages (A 7 years, B > 30 years and C > 60 years). We described how the clonal structure of B. fargersii populations at different genet ages established and developed at the small scale. We amplified 118 microsatellite locus using 8 selective primer pairs. A total of 49 clones were identified from 256 leaf samples collected from three populations, among which 31 clones were detected in plot A, 10 clones in plot B, and 8 clones in plot C. The size of clones in the three plots increased and the number of clones decreased with population aging. The spatial distribution pattern of clones in plots A and B exhibited a clumped distribution, while plot C showed two different patterns with simultaneously clumped and discrete distributions. The results showed that the genet generally formed a clumped distribution pattern during the seeding stage. The clones may expand into either the more sympodial type of ramets or short-distance clones, indicating that clonal propagation restricts dense seedling growth. However, with increasing clone size and genet age, compound axis mixed B. fargesii may expand into either the more monopodial type of ramets or long-distance clones representing a discrete distribution pattern when the genet are pressured by other strong clones. In addition, in our study, both the Mantel test and spatial autocorrelation analysis supported the significant presence of positive spatial clonal structures in three plots at the small-scale level. Spatial autocorrelation analysis also showed that the positive spatial genetic structure distance of the three plots in the 10 m distance class were 3.1, 28, 48 m, X-intercepts were 9.051, 30.698, and 50.536, and scope of spatial autocorrelation coefficients r were 0.1-0.167, 0.008-0.703, and 0.006-0.735, respectively. Our results showed that the size and scale of clones increased with genetic aging. Additionally, the number of ramets in the same clone increased under uniform sampling conditions, indicating that more samples with the same genotype can be collected with a positive spatial genetic structure distance. In our study, the distinguishable genotypes (G/N) from populations A, B, and C were 1, 0.14, and 0.055 and the Simpson's indices of diversity (D) were 1, 0.876, and 0.744, respectively. This result revealed that the distinguishable genotypes at the seeding stage (A 7 years) was much greater than that at the adult stage (B 30 years and C 60 years). Although the genotypic diversity of clonal populations reduced with genet aging, because of initial seedling recruitment, the value remained high. The unweighted pair-group method and principal component analysis demonstrated that clones in the same plot were always classified into the same clade. However, a few clones from different plots exhibited aggregation and overlap during analysis. In conclusion, we demonstrated that gene flow and an approximate clone origin might exist in different populations.