Experimental evolution · stable product v1.0

When does a group become an individual?

Across 600 transfers, five populations of snowflake yeast evolved larger bodies made from increasingly elongated cells. Replay the published trajectories, then test the geometry without mistaking the model for the experiment.

Enter the evolution chamber Read the result

5: independent anaerobic lines
600: daily transfers
64: paired published means

Frozen question

Does cell shape track cluster size in every lineage?

Criterion passed 5 / 5 positive

Median within-line Spearman ρ = 0.956

Interactive laboratory

Replay a lineage

Time and cell shape follow the published population means. The animated arrangement explains how elongation can relieve packing strain and how entangled branches can remain connected after a bond breaks. It does not recreate measured cluster images.

Population

Transfer0

Published measurement

Mechanism view

Mean aspect ratio1.25

Mean radius16.75 µm

Relative radius1.0×

Modelled packing relief0%

Mechanism stateConnected tree

modelled cells chitinous tree junction steric retention · not a bond severed junction

Connected clonal tree Every cell has one modelled parent except the founder.

Measured mean cell aspect ratio · mean cluster radius

Modelled branch topology · contacts · fractures · packing cue

What the model is doing

A fixed PA1 seed generates one repeatable illustrative topology. The published means determine its scale and cell elongation.

The mechanism views operationalize the paper’s proposed sequence—elongation reduces packing strain; later branch entanglement can preserve cluster integrity after cellular bonds fail. A tree junction is a permanent chitinous parent–daughter connection; steric retention is physical interlocking after a cut, not adhesion or a repaired bond. These are explanatory stress tests, not additional observations at the selected transfer.

Published figure-source data

Five paths, one direction

Select a measure. Each line is a biological population; points are published means, not individual cells. PA2 radius is missing at transfer 400 and is not imputed.

The oxygen contrast

Size selection did not produce the same outcome everywhere

At transfer 600, anaerobic lines were far larger than the obligately aerobic and mixotrophic lines. These are five replicates per evolved treatment, shown individually.

Evidence ladder

What changed—and what remains unresolved

Cells elongated

Longer cells reduce dense packing and delay strain-driven fracture.

Branches entangled

Interlocking branches can keep a cluster intact after individual bonds break.

Genomes duplicated

Tetraploidy evolved early, but its presence alone did not guarantee macroscopic size.

Transport emerged

Large clusters can generate fluid flows that move nutrients beyond simple diffusion.

New v1.0 evidence

Genome duplication helps. It is not enough.

Engineered tetraploidy produced larger cells and clusters in both tested backgrounds. But all ten evolved lines were already approximately tetraploid while only the five anaerobic PA lines later became macroscopic. Explore the intervention and the longitudinal sufficiency test without merging those two levels of evidence.

Published chromosome copy number

PA1 · transfer 200

Reference baseline: 4 copies per chromosome

Bar height and colour encode reported integer copy number. Aneuploidy burden is the sum of absolute deviations from four—not a fitness score.

Engineered intervention · four replicate strains per group

PA: immediate 2N → 4N effect

24 h cluster radius——

Cell volume——

Cell aspect ratio——

Exact replicate-level inference loading…

Longitudinal sufficiency test

Same early genome shift, different later scale

At transfer 1,000, all five PA lines exceeded 300 µm; the largest PM line was 53.226 µm. This rejects tetraploidy as a sufficient explanation in these treatments. It does not identify which later genomic or mechanical change completed the transition.

Measured here: ploidy proxy, chromosome copy number, cell geometry, and cluster radius. Not measured here: a joined, quantitative time-resolved entanglement endpoint. The proposed coupled threshold remains a future hypothesis.

Next registered question

A coupled-threshold hypothesis

Macroscopic multicellularity may require genomic size amplification and a mechanically entangling geometry to coincide. The v1.0 result shows that genome duplication helps but is not sufficient. A future protocol still needs a quantitative, time-resolved mechanical endpoint before the combined threshold can be tested.

Genome duplication
larger, longer cells

Entangling geometry
damage tolerance

Macroscopic body?
future test

Stop at the boundary

What this result cannot say

Association through time does not prove that elongation alone caused larger size.
Five selected laboratory populations do not define a universal evolutionary law.
Snowflake yeast are undifferentiated cell clusters, not miniature animals or plants.
The experiment does not reconstruct the historical origin of any living lineage.
The largest size and shape jumps do not occur in a consistent temporal order.

Inspect the evidence

Sources and reproduction

Primary paperBozdag et al., Nature (2023) Source dataPublic workbook and analysis code Frozen protocolQuestion, endpoint, exclusions, boundaries Research reportResult, sensitivities, interpretation Genome-duplication paperTong et al., Nature (2025) Frozen v1.0 protocolIntervention, sufficiency gate, boundaries v1.0 research reportExact tests, estimates, interpretation