[alanbernstein]

So there are 0) aquatic animals, 1) land animals (evolved from 0), and 2) aquatic mammals (evolved from 1).

You want to know if there are e.g. 3) land animals that evolved from aquatic mammals?

Not that it isn't an interesting question, but what exactly would this tell us?

[asfarley]

It might suggest the existence of loops in the evolutionary tree, making it more of a general graph.

[BbzzbB]

The evolutionary tree, maybe more accurately the phylogenetic tree, wouldn't loop as it is derived from genetics in modern times. Our ancestors' attempts at making evolutionary trees from morphology (can't fault them for not sequencing DNA) have got a lot of things wrong because seemingly relating creatures were not at all. Convergent evolution being a major reason for stuff they got wrong, making morphology a poor approximation of phylogeny.

Even if 2 distinct species eons apart ended up literally identical and sexually compatible thru some cosmic luck, they wouldn't show up as a loop since we could tell how far apart they are by looking at how mutated some useless and asexually transmitted gene is (mitochondria DNA provides that for eukaryotes, say cytochrome oxidase I gene). The key part is that the gene doesn't receive any evolutionary pressure, so that it only drifts thru random mutations (considering they can't provide a reproductive advantage/increase "fitness"). If you know the mutation rate (which we do), you can date how far back a common ancestor goes between two species' by basically running a diff script on their COI gene sequences.

Considering every known form of life descends from a single ancestor (LUCA [1]), a tree (or root system) analogy is quite apt.

1: https://en.wikipedia.org/wiki/Last_universal_common_ancestor

[samus]

On a large scale, there can not be loops, unless we accept the existence of time travel. But there are indeed some complications that make the concept of a tree of life surprisingly subtle:

- ring species

- species complexes

- hybrid species (turning the tree into a Directed Acyclic Graph)

- horizontal gene transfer via plasmids (between bacteria) or viruses

(edited for formatting)