If something is not conserved by selection, then the default is for it to be overwritten. So more like "need it or lose it". But I feel like that's pretty close to the article's phrasing.
Same way mammals keep losing vitamin C synthesis when they settle on a diet that's rich in it.
This isn't "use it or lose it" this is just "lose it and don't die". Having the synth path is better, dying is worse.
This is a terrible example; all mammals eat a diet rich in vitamin C, but loss of the ability to synthesize it is quite rare, being restricted to some primates, some rodents, and most bats.
https://en.wikipedia.org/wiki/Vitamin_C#Biosynthesis
It was a breakthrough when scurvy was observed in guinea pigs in 1907. Before that time, it had never been observed outside of humans, though it is known in humans for most of recorded history.
A human can eat nothing but fresh meat without suffering from scurvy. And in fact, a diet of fresh meat is sufficient to cure scurvy in a human already suffering from it. You could argue that humans conserve vitamin C better than lions do, but you can't argue that lions are eating a diet that is not rich enough in vitamin C to support an animal that can't synthesize its own. It's not an issue of the amount in the diet.
After all, one of the things that's so hard for us to grok about evolution is that it's not a goal-oriented process in any way. It's not engineering, it's the process of random chance and what can persist.
For example, a lot of people think that all our junk DNA actually does something profound and important, and while some of it probably does have a purpose we don't yet understand, I think it's almost certain that any process that makes a bunch of random decisions that are mindlessly culled by environmental forces will inevitably generate some flotsam and noise in its code. It's kind of like training AI; it definitely won't give you the most minimalist method of carrying out that task, but it will eventually find some way to carry it out. Almost as a brutal, automatic tropism.
Traits that are maladaptive are culled. Ones that have no effect are preserved, increased or diluted depending entirely on what other genes they appear with. Not the gene themselves.
If pale eyes were not beneficial in northern latitudes, we would still have blue eyes because the norsemen had them, and they were very successful. That gene will stick around until something forces it out of the pool because there is otherwise nothing to stop it.
In the absence of selection pressure the override genes will mutate randomly. The end state of that process are broken genes that don't do much of anything (loss of function), and so the color goes back to the default.
For human eyes, the default is blue eyes and there are then some genes that overrides this by producing melanin, which turns the eyes brown.
In such a scenario, the only surviving, reproducing individuals would have mutated in such a way to suppress expression of the gene. (Again, there is no evidence of this, it's just conjecture)
Ha—the plants were hosting an API for the bugs.
However, in general the concept of drift causing loss is solid; guinea pigs and humans have both independently lost the ability to create their own vitamin c. It's possible that some rather successful member of the species for some other reason (stronger, smarter, or just sexier) happened to have the vitamin c gene loss, and the need for vitamin c just wasn't pressing enough to overcome the other appeal.
If there is no selection happening on the gene, then the relative distribution of that gene in the population effectively drifts at random.
So yes, a gene can definitely be dormant, but it can also be lost if it's not actively being selected for.
There's no guarantee that the gene will be lost if it's not used, but it's definitely a possibility that increases with time.
Why would this be true? Can't a gene be dormant? Or perhaps over-ridden but another gene or genes?
Or is, over the long haul (read: __many__) years any and all genes will naturally mutate (that is, be lost). Then if the "replacement" is more advantageous, the replacement will persist.
Fair enough. But that's not use it / lose it per se. It's still in use, but less so (and is over time replaced).