In this paper, we address the debates on functional ascriptions/explanations in the philosophy of biology, with special emphasis on Larry Wright's selectionist etiological approach and Robert Cummins' functional analysis. We discuss some implications of these approaches to the scientific practice in biology, ecology, and Earth system science. We argue that, in a selectionist etiological approach, we cannot appeal to function in order to explain the origin of biological traits, but only to explain their spread in a population, after the appearance of a functional novelty. This limits the range of the explanandum of the etiological approach, but still preserves a domain in which we can legitimately apply it. As a consequence, this is a counter-argument to Cummins' attempt to deny the legitimacy of any etiological appeal to function in other scientific areas. There are, however, important limits to the domain of application of the etiological approach in biology. For instance, the etiological perspective on function is not legitimate to deal with appeals to functional language in ecology or Earth system science, because natural selection cannot act at the hierarchical level of the systems addressed by these sciences. However, the use of a functional language in these fields can be consistently formulated in terms of Cummins' functional analysis. In support of this argument, we develop a treatment of functional language in ecology and in the new field of Earth system science, by taking as a starting point a functional analysis (a la Cummins) of a particular biogeochemical system.