“…The Tokunaga model (2) completely specifies a combinatorial tree shape (up to a permutation of side branch attachment within a given branch) with only two parameters (a, c), hence suggesting a conventional modeling paradigm. The empirical validity of the Tokunaga constraints (2) has been strongly confirmed for a variety of river networks at different geographic locations [10,[35][36][37][38], as well as in other types of data represented by trees, including botanical trees [39], the veins of botanical leaves [26,27], clusters of dynamically limited aggregation [39,40], percolation and forest-fire model clusters [41,42], earthquake aftershock sequences [14,15,19], tree representation of symmetric random walks [43], and hierarchical clustering [44]. The conditions (2), however, lack a theoretical justification.…”