I recall Richard Dawkins writing that a species could be defined as a group of genes that travel together through a river of time (I’m probably messing it up a bit). This sounds reasonable to me. The implication here is that it becomes harder to define this fly species as a separate life form from the cactus – the fly genes will always be tied to the cactus genes. I see a parallel in the transmissible cancers that occur in some animals like Tasmanian devils.

Reply to Charlie Jones: asking whether such evolutionary losses of metabolic activity confer any advantage is indeed an important question if one wants to better understand evolution. In this study, because we identified the mutations responsible for the loss of metabolic activity in Drosophila pachea, we could tackle the problem. We found that the exact same mutations that made D. pachea dependent on its cactus also confer an advantage in presence of the cactus lathosterol. More larvae make it to the adult stage when they have the D. pachea neverland gene than when they have a reverted neverland sequence with ancestral amino acids. This benefit is found only when larvae are raised on lathosterol. So these mutations have probably spread rapidly due to their fitness advantage, and coincidentally they made D. pachea dependent on the cactus.
Regarding the evolutionary loss of vitamin C synthesis, this change happened so long ago that it is hard to reconstruct the individual mutational steps. So I think that we don’t know if there was any potential advantage to this loss of function, or if it evolved just neutrally.

Virginie (author of the study)

On a related matter, do we know how many mutations it took for the common ancestor of primates to loose its ability to make vitamin C? Was there any potential advantage to this loss of function, or was it just neutral because our ancestors at enough fruit to not need the gene? Again, I wonder why this broken gene spread unless it was in a small founder population.