It's a bit surprising that a Nature Neuro paper didn't have more control groups. But on the other hand, it's trickier to get ahold of first time adoptive mothers etc. And on the third hand, there's always somebody requesting increasingly specific control groups and one just has to stop and publish at some point.
Similar to open sourcing "ugly" code. At some point the benefit of releasing is worth putting out non-perfect work. If there's interest, someone can build on what you've done and fork your progress.
Incomplete control group. They should have included first time adoptive mothers and women who gave their baby up for adoption.
Also extremely small sample size.
EDIT: It would have also been very relevant to include a sample of lesbian parents of the same child where one parent gave birth and one didn't.