Actually, the paper goes on to explain that they discovered that a lot of the damage is caused by other pollutants that either just happen to exist on the same regions as NO2 or are created by its reaction with the air (they suspect ozone is a large guilty).
They measured NO2 because it's the best one to measure.
The data is coming from satellites, which are apparently able to directly measure NO2 levels (a good proxy for NOx levels), and they use near infrared reflectivity to estimate crop yield (which they say previous studies have shown to be a good way to estimate it).
Thank you for posting the actual quantitative data. All the sound and fury in the world pales in comparison to the awesome and terrifying power of Microsoft Excel!
Cars are a significant source, though in China "industry" is a bigger source per [0]. I'm not sure what the ration worldwide (and I didn't find a global number in the quick search that found the links I am posting).
Crop residue burning also comes to mind as a source that would be particularly concentrated near... crops [1].
Air polluion = Nitrogen Oxides
For summer crop 50% reduction of NOx , crop yield increase - China 15%, EU 10%, India 8%