This was, of course, discovered by examining fossilized dino shit, also known as coprolites.
This is very interesting from a biological standpoint, as the above quote indicates. I wonder if any of the grasses back then had developed THC. 
Fascinating! For what it’s worth, the last two big changes in paleoflora were thought to be the rapid diversification of angiosperms (flowering plants generally) starting at about 100,000,000 years ago, in the middle of the Cretaceous, and then the spread of grasses as the standard ground cover in the Miocene, say 15,000,000 years ago. (The late Cretaceous flora looked fairly modern, although a lot of things were still missing; it was the fauna that was significantly different.) The success of the perissodactyls, the artiodactyls, and the proboscideans (horses and rhinos, cloven-hooved animals, and elephants and allies respectively) were thought to be due to their evolving dentitions that could cope with grasses, and thus edging out the various other larger herbivores that had preceded them which didn’t adapt to grasslands.
Probably this indicates that titanosaurs were able to deal with grasses, and that grasses constituted a minor part of the flora at the time. But it’s still a truly major find.
My title’s cooler.
(I looked for other threads. I only checked the front page, however. I should have known I would still get scooped by Tuckerfan of the daily ‘Neat Shit’ thread.)
Polycarp: Interesting information. A much better response than any of my other Tuckerfanesque OPs here have gotten.