@Chembiodad, you just sent me down a (quite enjoyable) rabbit hole.
The Robertson et al. paper you linked to, first of all, has a lot more hedging in the body (just from a quick scan) than in the abstract. This isn’t surprising, but the extent to which math testing at age 13 (which is what they use) predicts future outcomes turns out to be quite incompletely predictive—in fact, given the way the paper sets everything out, I’d suggest that the overall purpose of their study was to show that earlier results that had pointed to math testing results being predictive were flawed by ignoring the degree to which other factors act as confounds.
Another very, very important issue: Robertson et al. deal only with those identified early on as having high quantitative ability; no conclusions can be drawn from this about those with high ability in non-quantitative fields.
There’s also a potential confound in that those who are identified as such early on would be more likely to come from families with the academic/cultural capital to recognize the possibility and get testing done and so on.
One fascinating sentence in a footnote, for which, sadly, a source isn’t given (emphasis my own): “Indeed, modern talent searches miss more than half of young adolescents in the top 1% on spatial ability, because they the cut on both quantitative and verbal selection measures.”
Anyway, Robertson et al. is reasonably well-cited. I haven’t done a serious review of the literature that follows, of course, but there are some intriguing articles that point to rather more basic factors (e.g., working memory capacity) being what’s really driving the differences found in studies like those of Robertson et al.'s. There’s also a pretty decent amount of research showing that implicit bias plays a role, in that those who are identified as top students are given (not earn, but rather are given) opportunities that may increase differences in possible outcomes, which can create the illusion of larger differences in outcomes than would otherwise occur.
TL;DR: For quantitative abilities, innate ability matters (at the very least for the top 1% vs. the rest of the top 5% and 10%; no claims within the top 1%). However, other factors have at least as big an effect as innate ability, and quite likely a larger one.