Because that's a definite-weight state of SU(3) flavor. And by definite-weight state I mean an eigenstate of specific SU(3) generators (specifically what are called Casimir operators, the analog of \vec{J}^2 for SU(2)). It's analogous to working out the product states of two spin 1/2 particles, but a little bit more complicated because the algebra of SU(3) has more generators than SU(2).