Source
Tobiasson, Luo, Wolf & Koonin (2026), LIT-1558, read in full in NOTE-1242. The section “Ancestral stem lengths”, Figure 3 and its caption, and Extended Data Figure 10b,c, read from captions.
What was actually shown
- The method tested. Pittis and Gabaldón’s normalized stem length: the branch from the last common ancestor of eukaryotes and their prokaryotic sister (FECA) to LECA, divided by the median LECA-to-tip branch. Since LECA-to-present is the same time for all genes, a longer stem is read as an earlier acquisition.
- The distribution. 5,850 stems, with a sharp peak near 0.05, “highly reminiscent of previous findings”.
- The reversal. “Alphaproteobacterial stems shorter than 0.3 (70% of all stems) were significantly longer than those of Asgard origin, but alphaproteobacterial stems longer than 0.35 were instead shorter than Asgard on average.” This was reproducible over 20 full technical replicates of 200 stems (Extended Data Fig. 10b). Cyanobacterial stems follow a weaker version of the same trend. Pittis and Gabaldón “analysed primarily stems within that range”, the long one.
- Restricted to clear cases. Ribosomal proteins (archaeal) and oxidative phosphorylation (alphaproteobacterial) accentuate both trends.
- Spread among co-acquired genes. Oxidative-phosphorylation stems, “assumed to have been acquired simultaneously”, vary over more than two orders of magnitude.
- Function. Information-processing genes, expected to be the oldest residents, have shorter-than-average stems.
- Conclusion. “even normalized stem lengths are determined primarily by factors other than the gene acquisition time, at least across the long evolutionary timescales analysed here.”
- Suggested cause. “the amount of evolution post-acquisition, necessary to adjust an acquired gene to a new biological environment.” Asgard genes were pre-adapted, and genes from distant bacteria had to change. This is “at least partially supported” by long stems among diverse bacterial donors (Extended Data Fig. 10c).
What could have failed: alphaproteobacterial stems could have been shorter across the whole distribution, and co-acquired genes could have clustered. Neither happened on these data.
What this does not say
- Not that the mitochondrion came early. The paper still holds that “mitochondrial acquisition was a relatively late event”, on gene-content grounds (THEORY-1104). It takes away one line of evidence for that view, not the view.
- Not what does set stem length. Post-acquisition adaptation is a suggestion. Rate variation along branches and unsampled donor lineages are other candidates, and the paper does not test among them.
- Not shown on Bernabeu et al.’s pipeline. They model the mode of each donor’s distribution, discard long-stemmed outliers, and cite a functional control (their Supplementary Fig. 13, not read here) in its favour. The two datasets appear to disagree even on the basic pattern. Here Asgard stems are mostly shorter than alphaproteobacterial ones. There Asgard is “the most ancestral signal”. Whether that is a difference in data or in summary statistic is not settled by either paper. The comparison is mine.
Connections
- THEORY-1105 (Bernabeu et al.): rival. Its ordering of transfer waves, and its claim that some bacterial transfers preceded the mitochondrion, rest on stem lengths read as relative times. If this account holds, that ordering is not established by its method.
- LIT-1595 (Pittis & Gabaldón 2016, read in NOTE-1240). The method’s origin. This account reproduces its result only in the regime it mostly sampled.
- LIT-1566 (Kay et al., read in NOTE-1234). It times eukaryogenesis by dated gene duplications, a different clock. If its ordering agrees with stem-length orderings, that would count against this account. If it agrees with gene-content scenarios and not with stem lengths, that would count for it.
- QUESTION-277. The question distinguishes a single merger from a sequence of associations. Any sequence needs a clock. This account says the clock most used for the purpose is unreliable.
Note of 2026-10-30: promoted, and the wider thesis filed beside it
The twenty-second status review promoted this reading under ADR-065: it reports what Tobiasson and colleagues find and conclude, and NOTE-1242 read the paper in full. Whether normalized stem length is in general no clock of acquisition is THEORY-1201’s question, Proposed, which keeps this entry’s old promote_when. The record’s own critique, once in the title, is now CLAIM-509, Proposed.