I want to read the actual paper, but it's paywalled. Suppose I could probably find it, but I'm a bit lazy.

I think the thing that stands out to me is that they did their analysis in terms of "sites". For urban ag, that could mean community gardens, backyard gardens, balcony gardens, and indoor plants including hydroponics/aquaponics. I'd be willing to bet that they only looked at community gardens, though, because it's not like they can get a good sample of the other categories. The carbon footprint of someone replacing a tiny rectangle of non-native turfgrass that requires watering, fertilizing, and mowing with veggies is not going to be like converting a vacant lot into a community garden.

Community gardens have to fill multiple roles, as they aren't just spaces for agriculture, they are essentially parks. The fact that they are typically segmented into individual plots also contributes to unnecessary cost for things like raised beds and paths. I don't think it's necessarily fair to compare a given urban land plot to a rural land plot since the alternative uses of those spaces are different. An vacant urban plot could be developed or used for parking instead of agriculture, while a rural plot could just be left wild instead. It's like the economic principal of an opportunity cost.

I'm also sometimes uneasy by the endorsement people have of using recycled materials for gardening. Obviously the idea is nice, but I've seen plenty of people using things like car tires for raised beds, and I don't think that would be good for growing food. People always forget that the first leg of "reduce, reuse, recycle" is reduce. Instead of reusing materials to build a raised bed, just grow plants right in the ground.