Yep, or almost more likely by other genes. But "self"-regulation - say by temperature in ectotherms - is real. Well, here you have to use your common sense. If the mutation confers the loss of a character that's arguably useless in the new environment (assuming the switch or gradient is very strong) then it's probably "adaptive". If it confers a new function useful in that environment, then it's probably adaptive. Single toes in horses was probably an adaptation to long-distance trotting or running on the plains developing at the time their feet were evolving from five towards one toe; yet, three-toed horses do occur. They're generally described as atavisms, but would probably be slightly maladaptive over time in a natural setting. You can also have mutations that are maladaptive - cancer being one. Of what? How do you mean this?