When constructing programs, it is a usual practice to compose algorithms that solve simpler problems to solve a more complex one. This principle adapts so well to software development because it provides a structure to understand, design, reuse and test programs. In functional languages, algorithms are usually connected through the use of intermediate data structures, which carry the data from one algorithm to another one. The data structures impose a load on the algorithms to allocate, traverse and deallocate them. To alleviate this ine ciency, automatic program transformations have been studied, which produce equivalent programs that make less use of intermediate data structures. We present a set of automatic program transformation techniques based on algebraic laws known as Acid Rain. These techniques allow to remove intermediate data structures in programs containing primitive recursive functions, mutually recursive functions and functions with multiple recursive arguments. We also provide an experimental implementation of our techniques which allows their application on user supplied programs.