While a tree grows over many years, somatic mutations accumulate and form genetic variation among branches within an individual. Trees can transmit such mutations to subsequent generations, potentially enhancing the genetic diversity of the population. We study a mathematical model to understand the relationship between within-individual genetic variation and branching architecture. We generate branching architecture by repeatedly adding two new branches (main and lateral daughter branches) to each terminal branch (mother branch). The architecture is characterized by two key parameters: main-lateral ratio (ML) and daughter-mother ratio (DM). During branch elongation, somatic mutations accumulate in the stem cells of a shoot apical meristem (SAM) at the tip of each branch. In branching, all the stem cells are passed on from the mother to the main daughter branch, but only one stem cell is chosen for the lateral daughter branch. We evaluate genetic variation by Z¯, the mean genetic differences between all pairs of branches of a tree, and examine how Z¯ varies with DM and ML while keeping the total branch length constant. As a result, (1) Z¯ increases monotonically with ML
(2) Z¯ attains the maximum for an intermediate DM, when stem cells in a SAM are genetically homogeneous
(3) Z¯ decreases monotonically with DM when stem cells are heterogeneous. The effect of branching architecture varies significantly depending on the genetic heterogeneity within a SAM, which results from the behavior of stem cells during growth. Our study sheds light on the overlooked role of branching architecture in storing genetic diversity.