Because originally Lisp was going to be written in m-expression syntax which the parser would turn into s-expressions, which are a very simple kind of AST. I think. But before they got the parser written, people had already gotten used to writing code in s-exprs, and preferred them to m-exprs.