It's too bad indeed that the labels are rounded to one digit, but at least the plot bars are from unrounded data (as the second figure with 6%, 7% and 8% bars makes it clear), so I think readers still likely get the right impression of the effect size.
One interesting point is that a women-only or woman-solo paper gets more credit (9%) than a mixed-gender paper (8%).
I'm wondering why they rounded single-digit number, it's 15% imprecision (8.4% and 8.5% aren't significantly different, whereas 7.5% and 9.4% would be a great gap).