Yeah, the sample set will be highly biased towards young animals. There will old animals as well -- the ultimate fate of the breeding stock is the same -- but I'm not familiar enough with the study I'm referencing to know if any problems in that subsample would be obvious.
It's not obvious to me that you would expect there to be problems that only show up after years but not in young animals with exposure from birth -- young animals are fast-growing, and fast-growing things are very susceptible to poisons of all flavors. (It's the basis of chemotherapy: you give the patient a poison, and the fast-growing cancer dies faster than the rest of the person.) Which is to say, a study of animals fed a substance for only the first tenth of their life is way, way more useful than a study that only looks at the second tenth, because of the rapid development during that time. Whether that makes the full lifespan data "useful incremental data", "vitally important data" or "uselessly redundant" I don't know enough to have an intuition on.
