= Multicellular organism
{wiki}
A <multicellular organism> has a body of multiple <biological cells> whose growth and activities are integrated into the life of the whole <organism>. Extensive <cell differentiation> is common in complex examples, but specialized <biological cells> are not a prerequisite for every simple form. Ask how the body develops, how its parts cooperate and how it produces descendants, rather than counting attached <biological cells> alone. The boundary with a <biological colony> depends on the degree of integration and the criteria being investigated; there is no universal cell-count threshold. See the Queller and Strassmann account cited under <unicellular organism>.
Specialization makes the whole capable of combinations that one general-purpose <biological cell> cannot achieve in the same way. In an <animal>, a <neuron> transmits signals while a <red blood cell> carries <oxygen>; each still requires a working <cell membrane> and <metabolism>. A mature mammalian <red blood cell> lacks a <cell nucleus> and <mitochondria>, illustrating how a useful part need not retain the equipment of a self-reproducing <organism>. See https://openstax.org/books/biology-2e/pages/40-2-components-of-the-blood[Clark, Douglas and Choi, Biology 2e, 40.2: red blood cell specialization]. <Cell differentiation> generally involves different <gene expression>, rather than a different inherited <genome> for each function. <Biological development> must coordinate growth, position, interactions and specialization: increasing the number of <biological cells> is only one component. See https://www.ncbi.nlm.nih.gov/books/NBK21046/[Alberts et al., Molecular Biology of the Cell, chapter 21 (2002): construction of multicellular bodies].
\b[A motivating experiment: select for a property of groups.] Ratcliff, Denison, Borrello and Travisano's 2012 study began with <Saccharomyces cerevisiae> growing as <unicellular organisms>. Daily transfer favored rapidly settling material. Their initial protocol allowed a $10\,\mathrm{mL}$ culture to settle for $45\,\mathrm{min}$, discarded the upper $9.9\,\mathrm{mL}$, and transferred the bottom $100\,\mathrm{\mu L}$ into fresh medium. After seven days, they instead sampled $1.5\,\mathrm{mL}$ and centrifuged it at $100$ times gravitational acceleration for $10\,\mathrm{s}$, again transferring the bottom $100\,\mathrm{\mu L}$. Faster settling made it more likely that a lineage would seed tomorrow's culture; the procedure selected a physical outcome without specifying how it must arise. These protocol details are explicitly recounted in https://academic.oup.com/evolut/article/73/5/1012/6727309[Gulli, Herron and Ratcliff, Evolution of altruistic cooperation among nascent multicellular organisms (2019): original settling protocol].
\Image[/unicellular-multicellular-settling-selection.png]
{title=Settling selection and an explicitly stipulated enrichment example}
{description=Original qualitative transfer schematic with toy numerical bars. The workflow represents the initial settling protocol described above. The lower panel assumes equally weighted tracked units and transfer probabilities $0.8$ and $0.1$; these values and the resulting frequencies are invented for explanation, not fitted to <snowflake yeast> measurements. The equal-transfer comparison is a model control, distinct from the experimental comparisons described below.}
The primary 2012 paper reports <snowflake yeast> dominating all ten initial replicate populations by 60 transfers. To distinguish the joining of previously separate <biological cells> from <clonal development>, the researchers enzymatically separated clusters and followed individual <biological cells> for $16\,\mathrm{h}$: attached descendants rebuilt clusters, with no such joining observed. Staining localized attachments at bud scars. Time-lapse observations also showed clusters releasing smaller multicellular offspring, called propagules; small clusters grew before they reproduced. See https://pmc.ncbi.nlm.nih.gov/articles/PMC3277146/[Ratcliff et al., Experimental evolution of multicellularity (2012), figures 1–3: emergence, development and reproduction].
The experimental comparisons ask different questions. Competition with the <unicellular organism> ancestor showed a reported $34\%$ advantage under settling selection. Without settling selection, the estimated $10\%$ disadvantage had a one-sided $P=0.06$, so it was suggestive rather than decisive evidence of a cost. Separately, three populations derived from one evolved isolate retained the cluster form through 35 transfers without settling selection. That persistence tests stability, not whether clusters always outperform single <biological cells>. See Ratcliff et al. (2012), figure 2 and the cluster-formation results cited above.
A 2015 follow-up added a causal test: deleting the <gene> ACE2 in the <unicellular organism> ancestor produced <snowflake yeast>, while supplying functional ACE2 to one evolved isolate restored separation into single <biological cells>. Sequence changes altering the encoded <protein> were found in ACE2 in five of ten examined lineages; the study did not identify that same sequence-change mechanism in every lineage. The deletion and restoration comparisons support a mechanism beyond the observation that clusters settle quickly. See https://www.nature.com/articles/ncomms7102[Ratcliff, Fankhauser, Rogers, Greig and Travisano, Origins of multicellular evolvability in snowflake yeast (2015), figure 1].
\b[What follows, and what does not.] The evidence supports inherited cluster formation, <clonal development> and a reproducing collective whose properties can respond to <natural selection>. Calling these simple <multicellular organisms> recognizes those demonstrated properties; it does not turn every clump into an <animal> or establish animal-like <cell differentiation>. This laboratory route also does not reconstruct the actual ancestry of <animals> or show that complex <developmental integration> must follow. The interesting achievement is narrower and testable: a change in how descendants remain connected creates a new scale at which growth and <reproductive success> can be measured.
For the toy figure, let $p$ be the fraction of tracked units of type C before transfer and let $s_C,s_U$ be the respective <probabilities> of transfer for C and U. Assume equally weighted units and equal multiplication between transfers. When overall transfer <probability> is nonzero, conditioning on transfer gives
$$
p_{\mathrm{after}}=\frac{p s_C}{p s_C+(1-p)s_U}.
$$
With $p=1/5$, $s_C=4/5$ and $s_U=1/10$, the numerator is $4/25$ and the denominator $6/25$, giving $p_{\mathrm{after}}=2/3$. If both types have the same nonzero transfer <probability>, the fraction stays $1/5$. This is <conditional probability>, not a model of the full experimental life cycle: real cluster sizes, growth and fragmentation affect which quantity a frequency measures. In particular, a fraction of clusters is not automatically a fraction of <biological cells>.
Back to article page
