Unfortunately that's how the data is provided [1]. Maybe because the Great Lakes were carved by geologically recent glaciation, and the model is built off tectonic plate movement.

A writeup of the methodology is available if you're interested. [2]

[1] https://www.earthbyte.org/paleomap-paleoatlas-for-gplates/

[2] https://drive.google.com/file/d/1-q0WIa7ofISFHyBe4UxvN8DIPs8...

The formation of the great lakes was an eye-blink ago on the scale of this map. They are only about four times as old as the great pyramid (woah)

The maps are hand-painted representations of the various ages. On the earlier maps, there was a lot of care, but also unavoidably a lot of handwaving. For the 0 Ma map, I don't know if Scotese had a particular time in mind? It looks like modern coastlines, which means an interglacial period. A 100 ky average would be more glacial - the Boston coast would stick out more. I suspect he simply used the current coast. The absence of the Great Lakes, suggests it's not the current interglacial. A duration-weighted average of say the last 1 My of 10-ish interglacials might fade the Lakes to invisibility. So perhaps it's an eye blink of averaged interglacials, or a recent but previous interglacial, or a map rule like "don't include transient features", or a consistency of approach (eg, disregarding flavors of data unavailable for earlier maps, except for the coastline), or ... ?