Biological versus statistical hypothesis

IT allows the simultaneous assessment of multiple competing biological hypotheses that are translated to statistical models. The user thus inherently assumes that one statistical model or hypothesis corresponds to one biological hypothesis, and these are interchangeable phenomena. But (I) y~a+b and (II) y~a are really different biological hypotheses?

They certainly can and should be, whether they are is of course a different issue. That's really on the individual researchers and whether or not they are making concrete choices about what statistical models to include and how they correspond to biological hypotheses. It is a lot easier to construct a statistical hypothesis from a biological hypothesis but I wonder how often we all create post-hoc biological hypotheses for well supported statistical hypotheses.

It also is an issue that holds for any statistical approach.