I suspect the underlying issue is that we are all terrible at unthreading cause from effect.

Words matter. Talking about "species level" is misleading because talking about an effect (blue phenotype) distracts us from the cause (a gene for blue).

And evolution is confusingly another step removed: the red survives because the blue did not.

Most of us struggle to see the counterfactual absence of a something.

The blue "species" never exists - our brains want to find groups and patterns. Our words mislead us.

The same thing is alluded to in the paper: the deaths of people without the gene is what is driving the gene to spread (through reproductive success of those that didn't die).

Reproduction is what matters (which death or sickness may interfere with).

Even the paper is confusing because it talks about 50 year old women (measuring survival bias; not whether they are expected to pass on genes at 50).

(reedits for clarity).