I agree tank size is a factor, thought of mentioning it above but had time for just a short post and didn't go into it. However, tank size is not the whole story ime. In some species, like A. ruviulatus, the wilds I've had have been a whole different animal than captive bred. As far as overwriting millenia of adaptation in a short time, it's a reasonable point that seems logical on the surface, but there are subtle ways in which it doesn't necessarily hold up scientifically.
There are many instances of rapid adaptations known in science, some within just a few generations, not to mention morphological plasticity built into some species (intestinal length of tropheus, for example) and plasticity of gene expression and regulation. There's a lot more to this than I'll take time and space to go into, but as just one example, there was a study done with salmon a few years ago (Kihslinger RL, Nevitt GA (2006) - Early rearing environment impacts cerebellar growth in juvenile salmon. J Exp Biol. 2006 Feb 1;209(Pt 3):504-9) that found an enriched environment for salmon fry, as opposed to a bare tank, affects brain structure and behavior. Effectively, bare tanks can potentially produce stupid (uh, developmentally challenged) fish, at least in some respects.
Also, go back some years and our model of DNA was as something hard wired, it was often compared to a blueprint. Now researchers are seeing it more as software. In software there's a lot of "if, then" logic.
If this happens or this or that condition is detected,
then carry out this routine,
else (otherwise) carry out this other routine. You can code a lot of triggers that respond to environmental conditions as monitored by various hardware, something as simple as a thermostat, for example. Research is finding that a lot of that sort of thing appears to happen genetically. Which leads me to the epigenome, chemistry apart from but associated with DNA that
regulates DNA. It can turn genes off and on, tell genes
when to do something, respond to environmental triggers, etc. The epigenome responds to 'signals', including from the environment, can learn (record) from its 'experiences' and pass signals (information) from certain environmental triggers on to the next generation (
here's a non-technical link), providing a mechanism for changes in response to the environment that may take place within the lifetime of an organism or may be passed on to the next or future generations, contrary to the post Lamarckian biology most of us grew up with.
So, on more than one level it's not impossible for a species' behavior to change in captivity in a few generations. To what extent no doubt depends on the species and individual.