The study didn't test for mathematical ability. It tested for numeracy which is completely different.
The study even shows that most of the variance(34%) is explained by fluid reasoning and working-memory. I'd wager that fluid reasoning is strongly correlated with mathematical ability.
In my opinion this paper disproves it's stated hypothesis but doesn't realize it.