Visual inspection also suggests a rather low R^2 on that regression. It looks much more like a loose correlation than a formula to me.
Looks like the trees along the right have a half-life range of <10,000 to >1,000,000, even with nearly identical individual mass.