Let $G$ be a transitive permutation group acting on a set $\Omega$. A structure tree $T$ for $(G,\Omega)$ is defined as follows: if $G$ is primitive, then it consists of a root node connected by edges to each element of $\Omega$, now the set of leaves; if $G$ is not primitive, we choose a maximal block system $B_1, \dotsc, B_k$ (that is, a block system where the blocks are minimal), define a structure tree for $G$ as a group of permutations of $B_1,\dotsc,B_k$, and then draw edges from the node corresponding to each $B_i$ to the elements of $B_i$. It is clear that $G$ has a natural action on $T$.
My understanding is that the more general definition of a structure forest goes back to (Luks and Mackenzie, FOCS, 1985). Where did the concept of a structure tree first appear? (Does it really originate in the (relatively recent) study of algorithms for permutation groups, or is it older?)