The image of a single branching tree of life, inherited from Darwin and formalized bymolecular phylogenetics, sits uneasily with what genomics now shows about prokaryotes: thatgenes move sideways across lineages so pervasively that no single gene tree reliably represents anorganism's whole history. This article argues that the tree-versus-web dispute has been miscast asa choice between two incompatible topologies and that both camps are partly right, because theanswer depends on which genes one asks about. I develop a gene-class-stratified reconciliationframework that models a genome as a vertically inherited informational core — the ribosomaland transcription-translation machinery that remains stubbornly tree-like — embedded in a horizontally acquired operational and stress-response shell that is genuinely network-like. On thisaccount the tree of life is neither a tree nor a web but a tree of cores threaded through a networkof shells, and the balance between the two is not fixed but varies across taxa and environments.From the framework I advance the falsifiable Extremity–Reticulation hypothesis: the adaptivecontribution of the horizontal shell, quantified by a Reticulation Index that partitions a genome'sadaptive genes by functional class and provenance, scales with environmental extremity, so thatextremophiles occupy the maximally reticulate end of a continuum. Heuristic scoring across anextremity gradient shows the informational core holding near a Reticulation Index of 0.05–0.08regardless of environment, while the stress-resistance shell climbs from roughly 0.20 in mesophiles to near 0.60 in polyextremophiles. Extremophiles, in this reading, are not marginal curiosities but the clearest demonstration that adaptive evolution in prokaryotes is driven substantiallyby acquisition rather than only by mutation, and the tree of life must be reconceived accordingly— as a statement about the informational core alone, explicitly qualified for everything else.