I do it beginning with the second observation but usually only if it is something I intend to make further Observations of. I have a new plant in my garden I will likely use SDS for the next time I observe it, for example, and in that case I will link it because I think it may be multiple Observations or some small (unknown to me) growth detail for it to be identifiable. (Also, it’s just a cool little plant.)
In the case of the two plants referenced in the original post, the tuliptree and the redbud, there conceivably are different seasons or growth stages.
I click on Similar Data Set anytime I come across it on an Observation set to see what else is included in that particular one, and to see a less familiar or unfamiliar plant at various stages (even if just fall, winter, spring, summer) would be very interesting to me.