I agree with your point. This is why I’m in awe of the people who manage to do genetics in wheat. I just have two caveats related to the different between recent tetraploids (plants like wheat, potato and octoploid strawberries) and ancient ones (which, if you look far enough back, includes every plant genome sequenced so far).

1. An ancient tetraploid doesn’t necessarily have a large genome (relative to grape, arabidopsis has gone through two whole genome duplications, yet the grape genomes is ~3 times as big) and it certainly isn’t highly redundant. Very few genes get retained by chance over tens of millions of years in plants, generally if there are still two copies it is because there is a fitness cost of losing either one, either because the two genes have specialized in somewhat different functions and losing either one will make the plant less fit, or as a result of dosage (if the plant loses one copy of gene X or gene Y but still has two copies of the other gene the balance of X and Y is off causing problems and, presumably, reduced fitness).

2. Reading back over what I wrote, I see that it wasn’t clear that I wasn’t suggest that arabidopsis or maize became popular model systems because people knew the species were ancient tetraploids. For one thing I don’t know that anyone suspected arabidopsis was an ancient tetraploid back when it was first becoming popular. I meant to convey was that if ancient tetraploids have a higher incidence of interesting non-lethal mutant phenotypes when mutagenesis knocked out members of gene pairs retained for the reasons I mention above (pure conjecture), the result might be to attract interest in working with the system from more plant researchers (not knowing that the plant was an ancient tetraploid something that would make life complicated for genomics researchers when the genome was sequenced decades later), contributing to a species becoming a popular model system.

Surely the importance of Arabidopsis can’t rest solely on having a sequenced genome. I’d at least include the floral dip method of plant transformation, being sexually self compatible and an incredibly short generation time to the list, although you’re right it probably didn’t become THE plant model system until it became the first plant to have its genome sequenced.

As far as gene redundancy, I was picturing cases of partial subfunctionalization (the arabidopsis tetraploidy is tens of millions of years old) where there’s still enough over lap in function to moderate the effect of mutant phenotypes. But since no examples of this happening are coming to mind, I can’t make much of an argument for the point.

While I’m speaking about arabidopsis, whoever suggested to you that arabidopsis became a model due to lots of redundancy in the genome is wrong. Arabidopsis has a TINY genome — which was a reason for it being a model — easy to sequence. Corn became a model just because it is an important crop, and probably because the bajillion transposons make mutation common. Redundancy in the genome actually makes genes much harder to study — because more mutants have no effect whatsoever.

Dozens or hundreds or thousands of weird, but non-lethal, mutant phenotypes piled on top of each other seems like a reasonable explanation for how a couple of species have been bred to create such a wide range of forms.