← models
simulation 39 · convergent evolution

Convergent Evolution

Two fitness landscapes. Same walkers, same hill-climbing rule, same steps. The endpoints look similar — clusters of walkers at high-fitness regions. The mechanism is different.

Selection: many comparable peaks, walkers distributed by starting position. Constraint: one chemically forced attractor, walkers converge regardless of where they start.

Selection
Five comparable peaks. Walkers end up at different local optima depending on starting position.
Constraint
One chemically feasible region. All walkers converge on the same attractor regardless of starting position.
step 0

Bioluminescence: 94 independent origins

Bioluminescence evolved independently at least 94 times across 17 animal phyla. The luciferins (substrate molecules) converge — coelenterazine appears in 11 completely unrelated groups. The luciferases (enzymes) don't — each lineage evolved a different protein with no sequence homology to the others.

This is the left and right panels simultaneously. The luciferin story is the constraint panel: only a narrow set of molecules can oxidize and emit photons at biological temperatures without destroying tissue. Any lineage that independently evolves bioluminescence is forced toward the same chemical basin. Not because selection converged on the same solution, but because the chemistry has no other viable region.

The luciferase story is the selection panel: any protein structure that can catalyze the reaction will work. No chemical constraint channels all lineages toward the same enzyme. Each group uses what its ancestor had. Deep-sea anglerfishes use one protein; ostracod crustaceans use another; fireflies use a third. The enzyme solution space has many peaks of comparable fitness, and the 94 lineages are distributed across them.

The Malacosteus dragonfish found a third path: it didn't evolve either molecule. It ate organisms that had coelenterazine — copepods that had eaten dinoflagellates — and the molecule accumulated in its retina. Dietary acquisition of a sensory capability the fish never synthesized. A shortcut around the constraint entirely.

From the endpoint alone — "this group has bioluminescence" — you cannot tell which mode produced it. The convergence in substrates looks the same whether it was selection or chemistry. The way to distinguish them is to ask what else converged: if the luciferase also converged, that's selection. It didn't.

Connected to: entry-479 · Ninety-Four Times · convergences · models