Where is it most likely that an hexapod creature branched from the evolutionary tree

Question

Suppose we have an alternate earth where there in addition to the real species on earth are large hexapod creatures similar to mammals and reptiles.

Where on the evolutionary tree is it most likely that they branched of from and why.

I came up with the following ideas but feel free to propose others.

1. The early ancestor to lobe finned fish before the number of limbs was fixed in evolution.
2. Flying fish transforming their wings and other fins to legs in convergent evolution.
3. Early vertebrate fish evolved another fin layout more amenable to evolving into six legs
4. Convergent evolution from some ancestor to insects evolving an internal skeleton
5. Early reptile evolving toes into legs and then re-evolving toes/fingers.
6. ...

Edit: My intent by similar to mammals and reptiles was that they should be a group of creatures with internal skeletons ranging in sizes from tiny like a mouse to huge like at least a rhino and competing for the same ecological niches.

Answer 1

IMHO the only realistic option is the first.

Tetrapods have a very stable conformation and it didn't change during all evolution.

Likely ancestor should be of Sarcopterigii clad.

Reason why I do not think various types of convergent evolution are highly unlikely is that, contrary to popular belief, it is not possible to code "whatever" in a gene set.

The biggest divergence in ontogenesis between Arthropod and Chordata is the "mangling" of homeobox genes which means that:

• in Arthropod homeobox are, more or less sequential and control a single aspect of the body; and thus is relatively easy to "recompose" insects with a different set of legs and/or wings.
• in Chordata homeoboxes are still, more or less sequential (encoding body characteristics from head to tail), but no longer control a single aspect, but many; this made possible to control many more aspects of the body using just a little more genes. We have, in modern chordata many more controlled characteristics than controlling genes.

This means changing a single gene will change many characteristics, not a single one.

For this reason evolution tends to diverge, even if some global characteristics seem to be the same. Cetacean fins do not resemble fish's and neither are similar to penguin wings, beyond the (very) general shape dictated by function.

Full explanation would be a bit too long to be done here, but my understanding is once two (vertebrate) branches diverge there won't be an easy way to make them really converge again but for some specific traits which may be mandatory in a certain habitat.

This does not hold true for Arthropods, but they are confined to a somewhat lower complexity due to the "simpler" matching between genes and bodily characteristics.

Answer 2

I'd consider #4 reasonable given the role of segmentation in body plans. It is reasonable to imagine that just as millipedes and insects learned to duplicate functions by duplicating segments, that some mammal - like or reptile - like animals would as well. Embryologically we do have segments, and I would guess that tinkering with homeobox genes could get you where you want to go. Here's one link.

I am not certain what the evolutionary advantage would be. Possibly extra legs allow an animal to remain mobile if they lose one. This means they either learn to get by with 5 or 7 (etc) legs, or they regenerate.

Speaking of regeneration, pollutants can cause extra legs in amphibia. You might consider amphibia.