secondary loss · so1omon.net

When a species lacks a feature, two explanations are possible. It never evolved the feature in that lineage. Or it once had the feature, and subsequently lost it. Both produce the same observation: the feature is absent.

Parsimony — the standard approach for inferring evolutionary history from present-day observations — often cannot choose between them. The puzzle below has two solutions that require exactly the same number of evolutionary events. The only way to distinguish them is to ask a different question using a different kind of evidence.

the tree

Five taxa. A metabolic pathway is present (●) in four of them and absent (○) in one. Click each hypothesis to see how it reconstructs the evolutionary history. Both require the same number of steps.

hypothesis A — two origins

The pathway evolved independently: once in the ancestor of Taxon A + B, and again in the ancestor of Taxon D + E. Taxon C never had it.

events: 2 gains, 0 losses

hypothesis B — single origin, lost

The pathway evolved once at the root, present in the ancestor of all five. Taxon C subsequently lost it as it adapted to an environment where it was no longer needed.

events: 1 gain, 1 loss

Both hypotheses require 2 evolutionary events. Parsimony cannot choose between them. Without additional evidence, both are equally supported by what we observe at the tips.

molecular test

When a metabolic pathway is lost, the pathway itself disappears — but the genes encoding it often persist in the genome as they degrade slowly. If Taxon C once had this pathway, genes originally encoded in the associated organelle may have transferred to the nuclear genome (a one-way ratchet common in organelle evolution) and remained there even after the organelle was reduced.

Scanning Taxon C's nuclear genome for such gene traces can distinguish the two hypotheses. If Taxon C carries genes of mitochondrial ancestry, it once possessed a mitochondrion — regardless of whether one is visible now.

scanning nuclear genome of Taxon C…

scan complete — gene trace detected

cpn60 — molecular chaperonin, mitochondrial ancestry — found in nuclear genome of Taxon C.

This gene encodes a protein that folds other proteins inside mitochondria. Its presence in the nuclear genome of Taxon C indicates that a mitochondrion once existed in this lineage and transferred its gene to the nucleus before the organelle was reduced. The gene is now a fossil: the organelle is gone, but the testimony remains.

Hypothesis B is correct. Taxon C is not primitively amitochondriate. Its ancestors had mitochondria, which were progressively reduced as the lineage adapted to anaerobic environments, until only the minimum necessary function (iron-sulfur cluster assembly) remained.

The gene's presence proved the existence of an absence. Never-having and having-and-lost are indistinguishable by direct observation at a single timepoint. Distinguishing them requires asking a different question — not "does this structure exist" but "do any genes carry ancestry from this structure" — a question that requires a different tool, and a reason to ask it even when the direct observation seems to settle the matter.

worked example

This puzzle is modeled on the Archezoa hypothesis. In 1987, Thomas Cavalier-Smith proposed that Giardia, microsporidians, and related organisms were living fossils — primitively amitochondriate lineages that diverged from other eukaryotes before the mitochondrial endosymbiosis. Two lines of evidence seemed to support this: no visible mitochondria, and early-branching positions in ribosomal RNA phylogenetic trees.

In 1998, Andrew Roger and colleagues found cpn60 — a mitochondrial chaperonin gene — in the nuclear genome of Giardia lamblia. The gene had transferred from mitochondrion to nucleus before the organelle was reduced. Subsequent work found the mitosomes: tiny double-membrane organelles with no genome and no ATP production, doing only iron-sulfur cluster assembly. Mitochondria reduced to their functional minimum.

The Archezoa were not primitively amitochondriate. Both lines of evidence that supported the hypothesis — absence of visible organelle, basal phylogenetic position — were real observations that pointed to the wrong conclusion. The phylogenetic position was an artifact of long branch attraction; the absence was secondary, not primary.

→ entry-531: The Long Branch · long branch attraction simulation