Because if one can postulate a causal relationship between the two variables, then the correlation becomes evidence of said relationship. For example, if I could explain how one's foot size determines one's IQ, then the correlation between the two would be evidence of such an explanation.
However, the difference between the evolutionary example and the foot size / IQ example is the following:
1) Evolution is the most parsimonious (and therefore most probable (Not necessarily-opinion only)) explanation for the relevant correlations (nearly identicle ape genomes, for instance).
2) The hypothetical explanation for the correlation between foot size and IQ that posits a direct dependance between the two is not the most parsimonious. In reality, the 'true' explanation (common dependance on a third variable: age) is far more pasimonious, and far more demonstratable.
I realise this is long winded, but I want to cover my bases. Correlation does not imply causation, so you cannot use the existance of correlation as grounds for a causal relationship, but, if you can demonstrate the possibility of a causal relationship irrespective of the correlation, then you can use said correlation as evidence of said relationship.