Clonal-aggregative multicellularity tuned by salinity in a choanoflagellate

Ros-Rocher, Núria; Reyes-Rivera, Josean; Horo, Uzuki; Combredet, Chantal; Foroughijabbari, Yeganeh; Larson, Ben T.; Coyle, Maxwell C.; Houtepen, Erik A. T.; Vermeij, Mark J. A.; Steenwyk, Jacob L.; Brunet, Thibaut

doi:10.1038/s41586-026-10137-y

Download PDF

Article
Open access
Published: 25 February 2026

Clonal-aggregative multicellularity tuned by salinity in a choanoflagellate

Nature (2026)Cite this article

Subjects

Abstract

Multicellularity evolved independently multiple times in eukaryotes^1,2,3,4. Two distinct mechanisms underpin multicellularity⁵: clonality (serial cell division without sister-cell separation) and aggregation (whereby independent cells assemble into a multicellular entity). Clonal and aggregative multicellularity are traditionally considered to be mutually exclusive^1,6,7,8, with rare exceptions⁹, and evolutionary hypotheses have addressed why multicellularity might diverge towards one or the other extreme^3,4. Both animals and their sister group, the choanoflagellates, are currently known to acquire multicellularity only clonally^4,10,11. Here we show that the choanoflagellate Choanoeca flexa¹² forms motile and contractile cell monolayers (sheets) through multiple mechanisms—C. flexa sheets can form purely clonally, purely aggregatively or through a combination of both processes. We characterize the life history of C. flexa in its natural environment—ephemeral splash pools on the island of Curaçao—and show that C. flexa undergoes reversible transitions between unicellularity and multicellularity during evaporation–refilling cycles. Different splash pools house genetically distinct strains of C. flexa and kin recognition constrains aggregation between them. We show that clonal-aggregative multicellularity is a versatile strategy for the robust establishment of multicellularity in this variable and fast-fluctuating environment. Our findings challenge former generalizations about choanoflagellates and expand the option space of choanozoan multicellularity.

Main

Multicellularity evolved independently many times across eukaryotes¹³, with multiple origins of both clonality and aggregation^2,4. Efforts to reconstruct the origin of animal multicellularity have benefited from the study of their closest living relatives, the choanoflagellates^2,4,11,14,15 (Fig. 1a). Choanoflagellates are bacterivorous aquatic microeukaryotes bearing an apical flagellum surrounded by a collar of actin-filled microvilli¹¹ (Fig. 1b,c). Moreover, many species display facultative multicellularity. The best-characterized choanoflagellate, Salpingoeca rosetta¹⁶, forms colonies exclusively clonally¹⁰, and clonal multicellularity has classically been assumed to be a general choanoflagellate feature^4,11,17. However, this assumption remains to be tested across choanoflagellate diversity. Notably, although animal multicellularity is purely clonal, other close relatives of animals exhibit diverse forms of multicellularity, including aggregation in filastereans^18,19 and cellularization of multinucleated cells^20,21 or cleavage-like serial cell divisions^21,22,23,24 in ichthyosporeans.

Here we describe an unconventional mode of multicellularity in the choanoflagellate C. flexa¹² that challenges prevailing assumptions about choanoflagellates. We show that C. flexa colonies form by serial cell division, aggregation or a combination of both, a mechanism that we refer to as clonal-aggregative multicellularity. We propose that this mixed multicellularity represents an adaptation to the dynamic natural environment of C. flexa: ephemeral splash pools that undergo extreme salinity fluctuations during cycles of evaporation and refilling.

C. flexa sheets can form clonally

C. flexa forms curved monolayers of polarized cells (or sheets) connected by direct collar–collar adhesions¹² (Fig. 1b–d). Sheets can reversibly invert their curvature in response to light-to-dark transitions, switching between feeding and swimming states^12,25 (Fig. 1b–d). In an earlier study, we established laboratory cultures of C. flexa sheets from a single cell, indicating that sheets can arise from individual cells¹². Nevertheless, the mechanisms that establish C. flexa multicellularity remain unclear.

In an attempt to image colony formation, we isolated single cells from mechanically dissociated sheets and monitored them using time-lapse microscopy. We observed formation of sheets by serial cell divisions approximately every 8–10 h (Fig. 1e,f and Supplementary Video 1) giving rise to small, polarized cell monolayers with the signature curved morphology of C. flexa sheets (Fig. 1b,e and Supplementary Video 1). We also monitored small and medium-sized colonies and captured events of colony expansion by cell division (Fig. 1g, Supplementary Fig. 1 and Supplementary Videos 2 and 3). However, we also noticed instances of free-swimming single cells or doublets meeting colonies, attaching and seemingly integrating within sheets (Fig. 1g, Supplementary Fig. 1 and Supplementary Videos 2 and 3). This showed that colonies can form purely clonally but suggested that they might also expand by aggregation. We therefore set out to test whether C. flexa is capable of purely aggregative multicellularity (Fig. 1h).

C. flexa sheets can form by aggregation

To test for aggregation, we performed live imaging of free-swimming single cells from mechanically dissociated colonies (Fig. 2a–d and Supplementary Videos 4–6). Notably, cells aggregated within minutes into doublets connected by collar–collar contacts (Fig. 2c and Supplementary Video 5), and progressively formed larger groups by incorporation of additional cells and fusion between groups (Fig. 2d and Supplementary Video 6). Early aggregates had irregular shapes, reflecting initial collision and adhesion in variable orientations. Over time, these aggregates underwent morphological maturation through cellular reorientations, ultimately forming polarized monolayers with canonical sheet morphology within 24 h (Fig. 2d and Supplementary Video 6).

To characterize aggregative morphogenesis, we imaged aggregates fixed at multiple stages (Fig. 2e–g, Supplementary Fig. 2 and Supplementary Videos 7–11). We quantified two metrics: the angles between the collars of neighbouring cells (or collar–collar angle) and the proportion of cells with aligned apicobasal polarity (Fig. 2f and Supplementary Fig. 2a). Both metrics initially showed high variance, consistent with early variable orientations, and progressively converged towards stereotypical values during maturation (Fig. 2e–g, Supplementary Fig. 2 and Supplementary Videos 7–11). Although early aggregates showed diverse types of intercellular contacts (collar–collar, collar–cell body and cell body–cell body; Fig. 2e and Supplementary Fig. 2b), cells in mature colonies were connected almost exclusively by their collars (Fig. 2e, Supplementary Fig. 2b and Supplementary Video 11). At 24 h, colonies averaged around 50 cells, with some reaching about 120 cells (Fig. 2g (bottom) and Supplementary Fig. 2d). These numbers confirmed that sheets could not have formed exclusively through cell division within 24 h, given the cell cycle duration in C. flexa (over 8 h; Fig. 1f), which allows at most 16 cells, assuming maximal and synchronized proliferation.

To independently confirm that sheets observed 24 h after dissociation resulted from early aggregate maturation, we combined two cell populations stained with distinct fluorophores (Fig. 2h, Supplementary Fig. 3 and Supplementary Videos 12 and 13). Cells of different colours readily aggregated (Supplementary Fig. 3a and Supplementary Video 12) and formed morphologically regular chimeric sheets within 24 h (Fig. 2h, Supplementary Fig. 3b,c and Supplementary Video 13). Furthermore, treatment with the cell cycle inhibitor aphidicolin did not abolish aggregation (Fig. 2i and Supplementary Fig. 4), confirming that aggregation is independent of cell division. Finally, we wondered whether aggregation was an active process or whether it might result from passive cell stickiness. Whereas live cells readily aggregated, fixed cells did not (even under orbital agitation forcing cell encounters), suggesting that aggregation requires living cells and is therefore an active process (Fig. 2j,k).

Taken together, this shows that C. flexa sheets can form by pure aggregation. This process is active, independent of cell division and occurs in multiple steps: random cell collisions create clumps of variable morphology, which then mature into polarized monolayers through cell reorientation.

This mixed multicellularity was a surprise, as clonal and aggregative multicellularity are classically considered to occupy different ecological niches: aggregation is often an emergency response to sudden stress^3,4, whereas clonality requires sufficiently stable environmental conditions to sustain cell division. This prompted us to investigate the natural context of clonal-aggregative multicellularity in C. flexa.

Salinity limits sheet presence in nature

C. flexa was discovered on the tropical island of Curaçao¹² (Fig. 3a) and has been repeatedly reisolated in the field since its discovery, allowing it to be studied in its native biotope. C. flexa sheets are found in coastal pools that undergo cycles of evaporation and refilling by splash, waves or rainfall²⁶ (Fig. 3a–c, Extended Data Fig. 1a,b and Supplementary Video 14). Splash pools are therefore ephemeral habitats in which organisms are exposed to extreme and recurrent hypersaline and hyperosmotic stress^26,27. We set out to investigate how this dynamic habitat influences the life history and multicellularity of C. flexa.

We investigated 150 splash pools in Shete Boka National Park across two field expeditions, Exped-A and Exped-B (Fig. 3d, Extended Data Fig. 1 and Supplementary Table 1). During Exped-A, we sampled 79 splash pools and found that splash pool salinity ranged from 15 parts per thousand (ppt) to saturation (≥280 ppt) (Fig. 3e and Extended Data Fig. 1d). By contrast, seawater from adjacent inlets (or bokas) had a stable salinity of around 40 ppt. In 10 out of 79 splash pools, we found choanoflagellate sheets that we identified as C. flexa based on morphology, behaviour and 18S ribosomal DNA (rDNA) sequencing (Fig. 3f, Extended Data Fig. 2, Supplementary Video 15, Supplementary Tables 1 and 2 and Supplementary Data 1–4). No sheets were observed in the other 69 splash pools. Although splash pool salinity ranged from 15 to ≥280 ppt, sheets were found only in splash pools below 73 ppt in this survey. To independently test whether salinity constrained sheet presence, we defined a 4 m × 10 m quadrant at a randomly chosen location in a second expedition (Exped-B; Methods) and exhaustively sampled all splash pools therein (n = 71) (Fig. 3d (inset), Extended Data Fig. 1c and Supplementary Table 1). We observed sheets in 14 splash pools, with a significantly lower salinity than average (≤94 ppt; Fig. 3e and Extended Data Fig. 1d). Across both expeditions, the average salinity of splash pools with observable sheets was 62.1 ± 24.8 ppt, significantly below that of splash pools without visible sheets (146.3 ± 95.7 ppt; P = 1.7 × 10⁻⁶, Mann–Whitney U-test). We never observed actively swimming and inverting sheets over 94 ppt salinity.

C. flexa persists in soil in a cryptic form

We next examined how evaporation–refilling cycles affected sheet occurrence over time. We monitored ten splash pools containing C. flexa sheets and five additional randomly selected splash pools from Exped-A over 8 days (Fig. 3e,g,h, Extended Data Fig. 3 and Supplementary Table 1). All 15 splash pools underwent measurable evaporation, as indicated by decreasing depth and increasing salinity. We also noted six desiccation events and four refilling events. Active sheets were never observed after salinity exceeded 128 ppt, consistent with earlier results. Notably, we did observe two individual colonies with an apparently stressed phenotype (irregular outlines, loose cell packing, no flagellar beating and no inversion behaviour) in two distinct splash pools at 200 and 212 ppt. These inactive colonies were no longer observed at later timepoints, suggesting that they had died or dissociated (Fig. 3h, Extended Data Fig. 3f and Supplementary Table 1). Similarly, gradual evaporation in the laboratory of a splash pool sample containing sheets led to sheet disappearance (Extended Data Fig. 4). These observations confirm that the multicellular form of C. flexa does not tolerate high salinity.

During this time course, two dry splash pools underwent refilling followed by sheet reappearance (Fig. 3h and Extended Data Fig. 3e,f). The newly observed sheets may have originated from the ocean or from neighbouring splash pools. Alternatively, C. flexa might have persisted in the soil of desiccated splash pools in a cryptic resistant form, such as unicellular cysts (as described in other choanoflagellates^28,29,30). To test this, we collected soil samples from six desiccated splash pools that had contained sheets and rehydrated them in the laboratory. We successfully recovered sheets from the rehydrated soil of two pools—implying that a resistant form of C. flexa had survived desiccation (Fig. 3i–k, Supplementary Video 16 and Supplementary Table 1).

To further investigate desiccation resistance in C. flexa, we set out to replicate evaporation–refilling cycles in the laboratory.

Hypersalinity induces sheet dissociation

We subjected C. flexa cultures to artificial evaporation (Fig. 4a), starting from the salinity of artificial seawater (35 ppt; hereafter, 1× salinity), and mirroring the temperature (30 °C) and evaporation rate of natural splash pools (Fig. 4b and Supplementary Fig. 5a). This resulted in desiccation after 4 days, during which C. flexa sheets gradually dissociated into non-motile single cells (Fig. 4c,d, Supplementary Fig. 5b and Supplementary Video 17). Control cultures without evaporation remained multicellular throughout the experiment (Supplementary Fig. 5c). We confirmed that the loss of multicellularity was caused by increased salinity by directly adding seawater salts to C. flexa cultures, which likewise resulted in sheet dissociation (Extended Data Fig. 5).

We next tested whether single cells resulting from evaporation-induced sheet dissociation could survive desiccation. We rehydrated samples by adding artificial seawater 3 h after desiccation, thereby mimicking refilling by waves. This restored salinity to around 50 ppt and was followed by the reappearance of sheets within 24 h (Fig. 4c,d). We assessed the mechanism of colony reformation using time-lapse microscopy after rehydration and observed unicellular flagellates engaging in both clonal division and aggregation (Fig. 4e and Supplementary Video 18). This shows that C. flexa sheets dissociate into desiccation-resistant, non-motile single cells during gradual evaporation, and that sheets reform after rehydration through both clonality and aggregation.

Hypersalinity induces encystation

We next investigated the phenotype of desiccation-induced single cells. In diverse protists, resistance to desiccation is achieved through differentiation into cysts³¹: dormant cells with reduced metabolism and proliferation. Encystation often entails morphological changes, including cell rounding, flagellar loss and cell wall formation²⁸. We monitored cellular morphology during evaporation and observed asynchronous structural changes that produced a heterogeneous cell population (Fig. 4f–i). At 3× salinity, most cells had dissociated from colonies and appeared round, without flagellum and/or microvilli but with occasional filopodia-like protrusions (Fig. 4g–i). Membrane and F-actin staining of desiccation-resistant cells (hereafter, cysts) confirmed the absence of a collar complex and revealed a transient F-actin cortex between 3× and 6× salinity (Fig. 4j–l and Supplementary Fig. 6), potentially protecting the plasma membrane against osmotic stress during early differentiation³². Cysts did not proliferate (Fig. 4m and Supplementary Fig. 7) and showed a higher nucleus-to-cytoplasm ratio than flagellates (Supplementary Fig. 8)—a common quiescence signature³³. Finally, direct addition of seawater salts to C. flexa sheets also induced encystation, confirming it as a response to salt stress (Extended Data Fig. 5). These morphological changes probably prepare cysts to survive desiccation (Fig. 4n).

We hypothesized that sheet dissociation during encystation was caused by retraction of the microvilli that connect cells within colonies¹². To test this, we induced microvillar retraction (without encystation) with the actin-depolymerizing compound latrunculin B, and colonies indeed dissociated within minutes (Extended Data Fig. 6a–d). Similarly, treatment of single cells with latrunculin B prevented aggregation (Extended Data Fig. 6e–h). These results confirm that microvillar integrity is necessary for multicellularity in C. flexa and support the idea that hypersalinity causes a coupled dissociation–encystation process.

The alternation between multicellular sheets and unicellular cysts in response to natural conditions suggests that these phenotypes each confer an adaptive advantage in their respective environments (Fig. 4n). We next sought to identify these putative advantages.

Adaptive benefits of sheets versus cysts

We first compared the desiccation resistance of sheets and cysts. We produced cysts by gradual evaporation and desiccation of sheets over 72 h. In parallel, we subjected sheets to rapid evaporation, causing desiccation over 20 h only (Methods). Under these conditions, cells did not visibly encyst but, instead, retained their flagellum, collar and multicellular morphology, even when completely desiccated (Fig. 4o). This indicates that encystation requires gradual evaporation comparable to that of natural splash pools. After desiccation, we rehydrated both types of cell by adding artificial seawater. Sheets desiccated by rapid evaporation never gave rise to viable cells after rehydration, whereas rehydrated cysts consistently produced viable and active sheets (Fig. 4p). Thus, cysts, but not sheets, can survive hypersalinity and desiccation and encystation is advantageous during evaporation.

We then examined whether multicellularity was, by contrast, advantageous in the other phase of the evaporation–refilling cycle, marked by low salinity. Multicellular choanoflagellates have been proposed to feed more efficiently thanks to cooperation between flagella³⁴, although tests in S. rosetta have yielded conflicting results^35,36,37. To assess this in C. flexa, we quantified bacterial capture in sheets and in dissociated single flagellates (Fig. 4q,r). Sheets captured over twice as many fluorescent bacteria per cell compared with single flagellates, suggesting that multicellularity confers a prey-capture advantage at salinities compatible with collar maintenance and therefore with feeding.

The versatile clonal-aggregative spectrum

Although splash pools all undergo evaporation–refilling cycles, they vary in size, evaporation rate and salinity after refilling. We wondered whether these environmental parameters might modulate the relative contributions of clonality and aggregation. Notably, we reasoned that conditions that impair cell division might favour aggregation, whereas conditions that limit cell–cell encounters should favour clonal multicellularity.

We first examined the influence of salinity within the range compatible with multicellularity (that is, below the dissociation–encystation threshold). At 1× salinity, both cell division and aggregation occurred (Fig. 5a and Extended Data Fig. 7a). By contrast, medium-high salinity (2×) arrested cell division without affecting aggregation (Fig. 5a and Extended Data Fig. 7a). Thus, sheets formed through mixed clonal-aggregative multicellularity at 1× salinity, but through pure aggregation at 2× salinity.

**Fig. 5: The versatile clonal-aggregative spectrum.**

We next examined the effect of cell density, reasoning that it might influence aggregation efficiency by modulating encounter rates between cells. Field measurements across 12 splash pools revealed densities ranging from 30 to 4.4 × 10⁴ cells per ml (Methods, Extended Data Fig. 7b and Supplementary Table 1). This range overlapped with cell concentrations used in laboratory aggregation experiments (from 1.0 × 10⁴ to 6.7 × 10⁵ cells per ml), confirming that C. flexa can reach densities sufficient for aggregation in nature. To quantify the effect of cell density, we seeded dissociated cells at 10² to 10⁵ cells per ml under both 1× and 2× salinity. We assessed the aggregation efficiency by measuring the final colony size under purely aggregative conditions (2× salinity), and measured the clonality efficiency by quantifying additional colony growth at 1× salinity over the 2× baseline (Methods). Measurements were taken during early sheet formation to avoid metric saturation. The aggregation efficiency increased monotonically with cell density and peaked at the highest density tested (10⁵ cells per ml) (Fig. 5b and Extended Data Fig. 7c). By contrast, clonality efficiency remained constant across intermediate densities (10²–10⁴ cells per ml) but decreased at the highest density (10⁵ cells per ml), possibly due to depletion of bacterial prey limiting proliferation (Fig. 5b and Extended Data Fig. 7c). Thus, low densities favour clonal multicellularity, high densities favour aggregative multicellularity and intermediate densities support mixed clonal-aggregative formation.

Taken together, this shows that clonal-aggregative multicellularity spans a continuum along which environmental conditions set the relative contributions of clonality and aggregation (Fig. 5c). Importantly, certain conditions (medium-high salinity and extreme cell densities) suppress one mode, resulting in purely clonal or purely aggregative multicellularity. This versatility allows C. flexa to achieve multicellularity across a broader range of conditions in the variable environment of Curaçao splash pools.

Control and aggregative sheets are equivalent

Having identified conditions that modulate the balance between clonal and aggregative multicellularity, we addressed whether clonal and aggregative colonies were functionally equivalent. We produced purely aggregative sheets by seeding dissociated cells at high density under 2× salinity and compared them with control sheets generated under conditions permissive for clonality (1× salinity and a low initial density). Control and aggregative sheets had similar morphology (Fig. 5d,e and Extended Data Fig. 7d–g) and were statistically indistinguishable in size (Extended Data Fig. 7h), proportion of aligned cells (Extended Data Fig. 7i), collar–collar angles (Fig. 5f and Extended Data Fig. 7i) and circularity (Fig. 5f (right) and Extended Data Fig. 7j).

To confirm that pure aggregation could also support growth of preformed colonies, we fluorescently labelled single cells and added them to preformed sheets under 2× salinity (Fig. 5g). Labelled cells initially adhered at the periphery of sheets in variable orientations (Fig. 5h,j) but, after 24 h, were robustly incorporated into the colonies, displaying collar–collar contacts and aligned apico-basal polarity with their unlabelled neighbours (Fig. 5i,j). Together, these observations show that aggregation is sufficient to support formation and growth of morphologically regular sheets.

Finally, we assessed the behaviour of purely aggregative sheets. Control and aggregative sheets exhibited equivalent inversion behaviour in response to light-to-dark transitions (Fig. 5k and Extended Data Fig. 8) and comparable prey capture advantage over single cells (Fig. 5l). We also generated sheets by aggregation of cells labelled with two different colours (as in Fig. 2h) and confirmed that the resulting chimeric sheets robustly inverted in response to light-to-dark transitions (Supplementary Video 19). Thus, colonies formed through aggregation alone are functional. Overall, although environmental parameters modulate the strategy by which C. flexa becomes multicellular, the morphology and behaviour of the resulting sheets appear largely uncoupled from their formation mechanism.

Kin recognition constrains aggregation

To further assess the relevance of aggregation to the life history of C. flexa, we tested for biological signatures of aggregative multicellularity. We first assessed species specificity—a frequent feature of aggregation³⁸—by mixing equal quantities of C. flexa and S. rosetta single cells. C. flexa formed aggregative sheets that completely excluded S. rosetta, demonstrating species-specific aggregation (Extended Data Fig. 9). Notably, S. rosetta cells did not aggregate with each other under these conditions, confirming that aggregation is specific to C. flexa rather than a general behaviour of choanoflagellates at sufficient density.

A second frequent feature of aggregation is kin recognition—that is, the ability to discriminate between kin and non-kin and to preferentially aggregate with closely related strains within the same species. Kin recognition increases relatedness in the resulting multicellular entity, which might facilitate the evolution of coordinated collective behaviour³⁹ and/or restrict the spread of cheater mutants⁴⁰. We assessed kin recognition among three C. flexa strains isolated from different splash pools: strains 1, 2 and 3 (Fig. 6a,b). Each strain was established by manual isolation of a single sheet, followed by amplification in laboratory cultures (referred to as ‘single-sheet-bottlenecked’ cultures). Single-cell isolation from each single-sheet-bottlenecked culture enabled the establishment of clonal strains.

**Fig. 6: Kin recognition constrains aggregation.**

To test for genetic divergence between these strains, we sequenced, assembled and annotated the C. flexa reference genome using a combination of long-read and Omni-C sequencing. The final assembly comprised 56 Mb and 14,084 genes across 528 scaffolds, with a BUSCO completeness score of 82.8%. This represents the third high-quality choanoflagellate genome after Monosiga brevicollis⁴¹ and S. rosetta⁴². We then performed short-read sequencing of strains 1, 2 and 3, as well as of three clonal descendants of each strain. Across all of the samples, we detected 193,564 single-nucleotide polymorphisms (SNPs), enabling us to reconstruct a phylogenomic tree in which all three strains were clearly delineated from each other (Fig. 6c and Supplementary Data 5 and 6). All clones clustered with their strain of origin and contained similar levels of genetic diversity as their parental strain (Supplementary Fig. 9). These results suggest either clonal formation of the originally isolated sheets (as in Fig. 1e–g) or loss of genetic diversity during sheet isolation and establishment of laboratory cultures.

We next tested kin recognition by dissociating sheets from all strains, staining them with different fluorophores and assessing aggregation across pairwise combinations. Notably, cells of strains 1 and 2 preferentially aggregated with their own strain when mixed, indicating kin recognition (Fig. 6d–g and Supplementary Fig. 10). Other pairwise combinations of strains showed no quantifiable preference for self-aggregation and readily formed chimeric sheets, as did self-mixing controls, confirming that aggregation is a widespread behaviour in C. flexa.

These results support the existence of kin recognition in C. flexa, similar to other aggregative organisms such as social amoebae. In the latter, kin recognition is mediated by polymorphic surface receptors rich in adhesion domains⁴³. In Dictyostelium, these receptors were first identified by their high ratio of non-synonymous to synonymous substitutions (K_a/K_s), a signature of diversifying selection⁴³. We reasoned that, if C. flexa indeed possesses a comparable kin-recognition mechanism, candidate kin-recognition loci might be revealed by screening the genomes of strains 1 and 2 for high K_a/K_s loci (K_a/K_s > 2). Across all pairwise strain comparisons, we identified 840 genes with high-K_a/K_s regions (Fig. 6h), including multiple transmembrane proteins (Supplementary Fig. 11a). Between strains 1 and 2, these high-K_a/K_s genes encoded predicted proteins with a significant overrepresentation of domains with adhesion (cadherins, TM132) or signalling (protein kinase) functions (Fig. 6i and Supplementary Fig. 11b). The top candidates notably included one predicted cadherin (FUN_000880; Fig. 6i,j) and one predicted receptor protein kinase (FUN_003087; Fig. 6i and Supplementary Fig. 11b,c).

These findings indicate that polymorphic receptor candidates potentially mediating kin recognition can be readily identified in C. flexa. However, some of these proteins may perform other functions (such as bacterial prey capture), and determining their precise roles will require functional genetic tools, which are not yet available for C. flexa.

Together, species specificity and kin recognition support the view that aggregation is a physiologically relevant process in C. flexa.

Discussion

Our findings suggest that clonal-aggregative multicellularity is regulated by evaporation–refilling cycles of splash pools, the natural habitat of C. flexa (Fig. 6k). Sheets are found in splash pools with moderate salinity, and dissociate into quiescent cysts after evaporation. Cysts persist in dry soil and, after rehydration, differentiate into flagellates that then reform sheets through clonal-aggregative multicellularity. This mixed mechanism confers two advantages in this fluctuating environment: (1) robust re-establishment of multicellularity across a broad range of conditions; and (2) fast acquisition of multicellularity by simultaneous action of both mechanisms. Indeed, aggregation generally provides a speed advantage over clonality, as aggregative eukaryotes often form multicellular colonies within a few hours⁴⁴, whereas clonal multicellularity is constrained by cell cycle duration (from 1 to 11 days in diverse microorganisms in the wild; Extended Data Table 1). Thus, we propose that clonal-aggregative multicellularity represents an adaptation to the variable and ephemeral splash pool environment.

The salinity-regulated multicellularity of C. flexa is consistent with recent studies on environmentally entrained life cycles, including field characterizations of multicellular bacteria (cyanobacteria inhabiting brackish environments⁴⁵ and cave-dwelling bacteria subjected to periodic flooding⁴⁶); laboratory experiments in yeast⁴⁷; and theoretical models^5,48,49,50. This suggests that environmentally entrained life cycles driven by cyclical environmental transitions might have preceded deterministic developmental programs in evolution²². More broadly, this supports the emerging concept of a premetazoan origin for complex life histories^14,51.

The clonal-aggregative multicellularity of C. flexa was a surprise, given that clonal and aggregative multicellularity are often depicted as mutually exclusive^2,4. However, clonality and aggregation cooperate in the formation of diverse structures. These include bacterial biofilms⁵², experimentally evolved clusters of bacteria⁵³, predator-induced groups of algae (which can even combine different species)⁹, yeast clusters⁵⁴ and syncytial amoebae¹³. While these diverse processes may not all represent bona fide multicellularity⁵⁵—as the resulting structures often lack multicellular-level adaptations such as controlled shape, size or collective behaviour—they demonstrate that aggregation and incomplete cell division can coexist. The clonal-aggregative multicellularity of C. flexa specifically challenges the canonical view that choanoflagellates exhibit only clonal multicellularity¹¹. This suggests that the diversity of multicellularity in choanoflagellates, a pivotal clade for reconstructing animal origins, warrants more systematic re-exploration. Broader taxonomic sampling in the future will clarify whether clonal-aggregative multicellularity potentially contributed to animal origins (as proposed previously¹⁷) or whether it represents a specific innovation of the C. flexa lineage.

Aggregation presents a well-known evolutionary challenge: a single aggregate can combine different genotypes, raising the potential of genetic conflict^54,56 and, importantly, limiting the gene coevolution required for the emergence of complex multicellular behaviours^39,57. The risks of chimerism can be mitigated by restricting aggregation to close relatives—either actively, through kin recognition, or passively, through a spatially structured environment limiting dispersal⁵⁸. The latter might be relevant to C. flexa, as splash pools are physically disconnected environments that might promote geographical divergence of genotypes (as for Daphnia in tide pools⁵⁹). Splash pools may therefore have favoured the evolution of aggregative multicellularity by initially relaxing the need for strong kin-recognition mechanisms (which C. flexa nonetheless eventually evolved). This scenario aligns with the concept of ecological scaffolding, which proposes that patchy environments can facilitate the emergence of multicellularity by fostering local cooperation⁶⁰. In the future, these questions will be informed by deeper characterization of dispersal, molecular kin-recognition mechanisms and natural genetic diversity at different scales (within and between colonies, within and between splash pools).

Finally, our study establishes C. flexa as a promising model to study multicellularity in a close relative of animals within its natural context. This contrasts with other well-characterized facultatively multicellular holozoans, such as S. rosetta¹⁶ and Capsaspora owczarzaki¹⁸, which were each isolated only once from their natural environment over two decades ago, inevitably restricting studies of unicellular-to-multicellular transitions to laboratory settings. In the future, we expect the continued dialogue between field and laboratory studies of C. flexa to continue clarifying key questions, such as the selective advantage(s) and ecological consequences of multicellularity versus unicellularity; the short-term plasticity and long-term evolvability of clonal-aggregative multicellularity in diverse environments and under diverse selective pressures; and the molecular mechanisms of kin recognition.

Methods

A full description of the methods is provided in the Supplementary Information. All dilution percentages are v/v unless specified otherwise. No statistical methods were used to predetermine sample size. No blinding and randomization were used.

Cell strains and growth conditions

C. flexa clonal monoxenic cultures (strain ChoPs7, hereafter strain 1), established as described previously¹² from single-sheet cultures isolated near Boka Wandomi in Shete Boka National Park (the location reference is provided in Fig. 6a), were grown in 25 cm² tissue-culture-treated flasks (130189, Thermo Fisher Scientific) in either in 1% seawater complete (SWC) medium or 5% cereal grass medium 3 (CGM3) diluted in artificial seawater (ASW), following standard culture medium and culture protocols for choanoflagellates^12,25 with minor modifications (Supplementary Information).

Tracking cell division of single cells

The cell concentration of an exponentially growing ChoPs7 culture grown in SWC medium was estimated using a LUNA-II automated cell counter (LogosBiosystems). The culture was then diluted to 1 cell per µl in 5% SWC supplemented with 1% Halopseudomonas oceani resuspended food pellet (20 mg ml⁻¹ in ASW). A 1 µl aliquot of the diluted culture was pipetted onto the centre of a well in a black 96-well ibiTreat µ-plate (89626, Ibidi) and covered with 400 µl of Ibidi anti-evaporation oil (50051, Ibidi). The sample was imaged using a Plan-Apochromat ×20/0.8 M27 Zeiss objective on the Zeiss Axio Observer Z.1 inverted microscope, using tile scan mode to cover the entire droplet surface, Definite Focus and a ColorBand filter (FGL610, Thorlabs).

Time-lapse imaging of mixed clonal-aggregative multicellularity

Colonies from ChoPs7 cultures grown in CGM3 medium were transferred to a FluoroDish (FD35-100, World Precision Instruments) and incubated for 30 min to allow them to settle at the bottom of the dish. Colonies were imaged every 5 min by differential interference contrast (DIC) microscopy using a Plan-Apochromat ×63/1.4 oil-immersion Zeiss objective mounted on a Zeiss Observer Z.1 inverted microscope equipped with a Hamamatsu ORCA-Flash 4.0 V2 CMOS camera (C1140-22CU) in timelapse mode.

Aggregation dynamics over time

Two types of time-lapse videos of aggregation were generated: low-magnification overviews capturing a large number of cells imaged with a ×5 objective (Fig. 2b and Supplementary Video 4) and high-magnification, high-frequency videos of a smaller number of cells imaged with a ×63 objective to enable cell tracking (Fig. 2c,d and Supplementary Videos 5 and 6). The experiment was performed in four independent biological replicates (n = 4). For both types of experiments, 3 to 10 ml of dense ChoPs7 cultures were collected by centrifugation at 4,700g for 20 min at 4 °C, resuspended in supernatant and dissociated by vortexing for 30 s. The resulting cell suspension was transferred into a well of a Corning 96-well plate (13539050, Thermo Fisher scientific) for imaging and imaged overnight by DIC microscopy on the Zeiss Axio Observer Z.1 inverted microscope (details are provided in the Supplementary Information).

Staining and Airyscan imaging of aggregates at different stages

Approximately 10–20 ml of exponentially growing ChoPs7 cultures grown in SWC medium was collected by centrifugation at 3,300g for 15 min at room temperature and washed twice with 20 ml of 1× ASW. Cells were counted, seeded in 1 ml of ASW in a 24-well plate and incubated for 10 min, 30 min, 2 h, 6 h and 24 h at 25 °C. After incubation, 45 µl of cells was transferred into a 96-well plate (89626, Ibidi), fixed by adding ice-cold paraformaldehyde (PFA) (15710, Electron Microscopy Sciences) to a final concentration of 4%, permeabilized for 5 min with 0.1% Triton X-100 (A16046, Thermo Fisher Scientific) and stained with 1:1,000 FM 4-64X (F34653, Invitrogen) and 1:100 Alexa Fluor 488 Phalloidin (A12379, Invitrogen). The samples were imaged on the Zeiss Axio Observer Z1/7 microscope with LSM900 Airyscan 2.

Production, staining and imaging of dual-labelled chimeric aggregates

ChoPs7 cultures were collected as described above, resuspended in around 1 ml of ASW, dissociated by vortexing and divided into two 1.5 ml microcentrifuge tubes. Cells in each tube were stained 1:1,000 with either CellTrace CFSE (green; C34570, Thermo Fisher Scientific) or CellTrace Far Red (magenta; C34572, Thermo Fisher Scientific). Cells were collected again at 3,300g for 15 min at room temperature, washed once with 1 ml of 1% BSA in ASW, washed once more with 1 ml of ASW and resuspended in 500 µl of ASW. Cells were counted and mixed at a 1:1 ratio in 1 ml of SWC medium in a 24-well plate. Plates were incubated overnight at 25 °C, and non-mixed single-labelled populations were seeded in parallel as controls. After incubation, 100 µl of the sample was transferred into a 96-well plate, fixed with 4% PFA, permeabilized with 0.1% Triton X-100 and stained with 1:1,000 Alexa Fluor Plus 405 Phalloidin (A30104, Invitrogen). The samples were imaged using the LSM900 Airyscan 2 system as described above.

Quantification of aggregation in aphidicolin-treated cells

Chops7 cultures grown in CGM3 medium were treated overnight with aphidicolin or DMSO (drug vehicle control), dissociated by vortexing and transferred into an µ-Slide 8-Well chamber (80826, Ibidi). The samples were imaged every 30 min for 2 h on the Zeiss Observer Z.1 inverted microscope equipped with a Hamamatsu ORCA-Flash 4.0 V2 CMOS camera (C1140-22CU).

Aggregation of fixed versus live cells

ChoPs7 cultures were collected by centrifugation, resuspended in ASW, dissociated by vortexing, counted, vortexed again to ensure complete dissociation, seeded in a 24-well plate and immediately fixed with 4% PFA. Cells were then placed on an orbital shaker (Rotamax 120, Heidolph) at 50 rpm for 24 h at room temperature. Control plates containing live (non-fixed) cells and static (non-agitated) conditions were prepared in parallel. After 24 h, cells were imaged using transmitted light bright-field microscopy on a Zeiss Axio Observer Z.1 inverted microscope.

Field sampling

Fieldwork data were collected in Shete Boka National Park (12°22′5.718′ N, 69°06′56.916′ W) in Curaçao, in three independent expeditions during July–August 2023 and July–August 2024. The park spans nearly 10 km of rocky, wave-exposed coastline and contains approximately 10 pocket bays, or bokas.

Exped-A sampling and monitoring

For Exped-A, at least 10 ml of seawater was collected from n = 79 different splash pools (Sp1–Sp79) using 25 cm² tissue-culture-treated flasks along around 2 km of coastline. Fifteen of these splash pools (n = 15) were selected for daily monitoring of evaporation and refilling cycles over an 8-day time course. Each splash pool was uniquely identified using a physical tagging system and its GPS coordinates were recorded using an iPhone 12 Mini (Apple). A photograph of each splash pool and its surrounding environment was also taken using the same device. The following parameters were measured in situ: salinity using a refractometer (B07FQPFJGX (ASIN), Gain Express), temperature using a thermometer (B07CB8JG21 (ASIN), ThermoPro) and depth using measuring tape. Splash pool seawater or soil (in dry splash pools) were collected for rehydration experiments. The presence of sheets was visually assessed at the CARMABI biological station in Curaçao using the Leica DM IL LED inverted microscope equipped with the Nikon Z 50 camera. As controls, open sea salinity and temperature were measured in the bokas of Boka Wandomi and Boka Kalki.

Exped-B sampling

For Exped-B, a random-number generator was used to select a randomized sampling location between 150 m and 250 m upstream Boka Wandomi, avoiding sites that were previously sampled during Exped-A. At the selected location, an area measuring 10 m by 4 m was mapped and defined. All splash pools containing at least 5 ml of seawater within this area (n = 71) were collected and analysed as described above (M1–M77). New C. flexa strains (strain 2 and strain 3) were isolated from splash pools M44 and M60, respectively (see the ‘Manual isolation of sheets collected in the field’ section in the Supplementary Methods).

Exped-C sampling

For Exped-C, at least 10 ml of seawater was collected from n = 12 different splash pools (SpA–SpL) located near Boka Wandomi and Boka Pistol and analysed as described above. To maximize the likelihood of finding C. flexa sheets, the salinity of each splash pool was measured before sampling. Splash pools with salinity values within the permissive range for sheet occurrence (15–128 ppt) were selected for collection. The samples were later inspected using an inverted microscope to estimate C. flexa cell density based on the number of observed sheets (Supplementary Information).

Soil rehydration

A total of n = 32 soil samples was rehydrated across two independent fieldwork expeditions (Exped-A and Exped-C). Soil samples of n = 6 splash pools that underwent complete desiccation during Exped-A were scraped and collected daily for 8 days using a spatula into 25 cm² tissue-culture-treated flasks, ensuring that soil was sampled from multiple areas within each splash pool. Soil samples were rehydrated in the laboratory with 50–100 ml of sterile-filtered seawater collected from Boka Wandomi, adjusting the volume to reach a salinity of around 40 ppt. At least three independent rehydrations were performed for each sample. The presence of sheets was monitored daily during the next five days.

Soil samples of n = 26 additional splash pools (Soil1–Soil26) surveyed during Exped-C were collected, treated and monitored as described above with minor adjustments (Supplementary Information) as an independent replicate experiment.

Artificial evaporation

In total, 3 ml of dense ChoPs7 cells were seeded into four separate six-well plates (130184, Thermo Fisher Scientific) and transferred onto a grid inside a 30 °C incubator. One six-well plate was kept with its lid on (low-evaporation control), while the remaining plates were left uncovered (gradual-evaporation condition) over a 9-day time course. A plastic box (dimensions: 35.7 cm × 23.5 cm × 13.5 cm) was placed over the plates to allow for air exchange while minimizing contamination from airborne particles (Fig. 4a). At each timepoint, salinity was measured using a refractometer and a 100 µl sample was transferred into a 96-well plate for imaging on the Zeiss Axio Observer Z.1 inverted microscope and manual counting of cells in unicellular versus multicellular forms.

Cyst imaging under gradual evaporation

Cyst formation was monitored daily over a 4-day period using DIC microscopy on the Zeiss Observer Z.1 inverted microscope equipped with the Hamamatsu ORCA-Flash 4.0 V2 CMOS camera (C1140-22CU). Then, 200 ml of Chops7 cultures was transferred into a Bio-Assay Dish (240845, Thermo Fisher Scientific) and evaporated at 28 °C with the lid open for 4–6 h until the salinity reached 60 ppt. The temperature was increased to 29 °C and the lid was closed overnight to allow cells to adapt to the new salinity. The lid was partially reopened the next morning to resume gradual evaporation until the salinity reached 80 ppt; the lid was closed overnight, and the temperature was increased to 30 °C. The lid was reopened the next morning until the salinity reached 100–110 ppt. The culture was then maintained at that salinity with the lid closed for 24 h. On the morning of day 4, cells were imaged in a FluoroDish (FD35-100, World Precision Instruments).

Cyst F-actin and membrane staining

Cysts were produced using gradual evaporation at 30 °C to allow gradual evaporation (see the ‘Artificial evaporation experiment’ section in the Supplementary Methods), fixed with 4% PFA, transferred into a 96-well plate, permeabilized with 0.1% Triton X-100 and stained with 1:1,000 FM 4-64X, 1:100 Alexa Fluor 488 Phalloidin, and 1:100 Hoechst (H21486, Invitrogen). Samples were imaged using a Zeiss LSM900 Airyscan 2 as above.

Quantification of prey capture

A H. oceani food pellet (20 mg) resuspended in 1 ml ASW was stained with BactoView-Live Green (FITC) (40102, Biotium) at 4 µl ml⁻¹ and incubated for 30 min at room temperature in the dark, centrifuged at 2,750g for 5 min and resuspended in 1 ml ASW. In parallel, 200 µl of dense ChoPs7 cultures, or of single cells dissociated by vortexing, was transferred into 8-well chambers (80826, Ibidi). Then, 100 µl of stained bacteria diluted 1:20 in ASW was added to each well containing colonies or single cells. C. flexa was incubated with bacteria for 1 min before fixation with 4% PFA. After fixation, the samples were imaged using DIC and green fluorescence microscopy using the Zeiss Observer Z.1 microscope. The capture efficiency was calculated as the ratio between the number of bacteria attached to the collars of choanoflagellate cells to the total number of choanoflagellate cells in each image.

Normalized growth and efficiency of aggregation over 2 h

To calculate the normalized growth under the 1× and 2× salinity conditions (Fig. 5a), the growth rates obtained from experiment shown in Fig. 4m (see the ‘Growth quantification after gradual evaporation’ section in the Supplementary Methods) were normalized as follows:

Normalized growth = (growth rate)/(maximum growth rate across all conditions)

The efficiency of aggregation (E_a) for each individual experiment was calculated based on aggregation time lapses (see the ‘Time-lapse imaging of mixed clonal-aggregative multicellularity’ section above) at the 2 h timepoint as the ratio of the mean area of sheets obtained in that individual experiment over the maximal area observed under 2× salinity (Supplementary Information).

Efficiency of clonality and aggregation at different cell densities

Approximately 20 ml of dense ChoPs7 culture was collected and dissociated as described above. Then, 10², 10³, 10⁴ and 10⁵ cells were seeded in 1 ml of either 1× ASW or 2× ASW supplemented with 1% SWC and 0.5% H. oceani in a 24-well plate. Cells were incubated at 30 °C and imaged after 24 h, 48 h and 72 h incubation using bright-field microscopy on the Zeiss Axio Observer Z.1 microscope.

We defined the efficiency of aggregation (E_a) at a given cell density as the mean size reached at 24 h at that cell density under 2× salinity, normalized to the largest size measured at 2× salinity (across all cell densities) at 24 h. We defined the efficiency of clonality (E_c) as the additional growth allowed by cell division at 1× salinity relative to the aggregative baseline measured at 2× salinity (equations and technical details are provided in the Supplementary Information). As a control, we verified that our metrics for efficiency of aggregation and clonality were not saturated (Supplementary Information).

Staining and Airyscan imaging of aggregative and control sheets

For control sheets

Approximately 20 ml of dense ChoPs7 cells were collected and counted as above and seeded into 1 ml SWC medium in a 24-well plate at low density (~200 cells per ml) to favour clonal over aggregative formation of sheets. Cells were incubated for 3 days at 30 °C.

For aggregative sheets

Approximately 20 ml of dense ChoPs7 culture was collected, dissociated and counted, and 10⁵ cells were seeded in 1 ml of 2× ASW (with 1% SWC and 1% H. oceani) in a 24-well plate. Cells were incubated for 24 h at 30 °C.

Both types of sample were fixed with 4% PFA, stained with 1:1,000 FM 4-64X and 1:1,000 Alexa Fluor 488 Phalloidin as described above, and imaged using the Zeiss LSM900 Airyscan 2 microscope as described above. Cells per sheet were counted and collar–collar angles were quantified using Imaris v.9.9.1 (Bitplane, build 61122 for x64). Circularity was quantified using Fiji⁶¹ v.2.14.0/1.54f (Supplementary Information).

Light-regulated inversion in aggregative and control sheets

Aggregative and control sheets were produced as described above, transferred into a 96-well plate and imaged on the Leica Stellaris 5 confocal microscope with the HC PL FLUOTAR 10X/0.30 DRY objective in the FRAP mode. A white-light laser was used to create a 488 nm laser line with 1% intensity and a 633 nm laser line with 1% intensity. For light-to-dark stimulation, the 488 nm laser line was turned on during the pre- and post-bleaching frames and turned off during bleaching frames, while the 633 nm laser line remained continuously on. Bright-field images were acquired using the Trans PMT detector, and fluorescence signals were detected using two HyD type S, covering 495 nm–605 nm spectra for 488 nm excitation and 635 nm–745 nm spectra for 633 excitation, respectively. The imaging interval was around 0.19 s. Acquired images were exported in ‘ImageJ TIFF’ format using Leica Application Suite X.

Ten colonies from each replicate were subjected to light-to-dark transition to assess light-sensitive inversion behaviour (total of n = 60 sheets). For quantitative analysis, five representative colonies per condition were analysed to measure changes in colony area during inversion. Cell segmentation was performed on bright-field images using a custom-trained model of Cellpose (v.2.2.3)^62,63. The resulting colony masks were quantified using the label and regionprops_table functions from the measure module of the same package. The quantified area was normalized to the mean area of the first 20 frames before the light-to-dark transition and plotted using tidyverse (v.2.0.0) in R (v.4.1.1) and RStudio (v.2021.9.0.351).

Sequencing and assembly of the C. flexa reference genome

ChoPs cultures for genome sequencing and assembly were established by thawing a low-passage, previously reported C. flexa culture¹². Monoxenicity was established as previously described¹² by antibiotic treatment and addition of live H. oceani bacteria. The resulting ChoPs strain (ChoPs8) was grown to maximal density (around 1 × 10⁶ cells per ml) in 5% CGM3 medium. Bacteria were removed by spinning three times at 3,000g for 15 min, washing each time with 45 ml ASW. The final pellet was snap-frozen in liquid nitrogen and sent to Dovetail Genomics (Scotts Valley) for genomic DNA extraction, Omni-C+PacBio sequencing and genome assembly, which were performed by Dovetail Genomics staff using in-house protocols (details are provided in the Supplementary Information).

The genome of C. flexa was annotated using an integrated pipeline combining the results from several gene-calling algorithms with clues from C. flexa transcriptomic data and predicted proteomes from previously sequenced choanoflagellates⁶⁴. In total, 688,088 protein sequences were used as homology-based evidence for gene annotation. Protein sequences were aligned to the C. flexa genome using DIAMOND (v.2.1.8)⁶⁵ and exonerate (v.2.4.0)⁶⁶. DIAMOND identified 269,722 putative alignments and exonerate identified 2,679. Three gene prediction software packages were used to define 45,273 putative gene models. First, Augustus (v.3.3.2)⁶⁷ was run using Toxoplasma parameters, resulting in 6,822 high-quality predictions (>90% exon evidence) and 9,196 gene models without quality thresholding (15,676 total Augustus gene predictions). Second, SNAP⁶⁸, was trained using 194 eukaryotic BUSCO genes (v.2.0)⁶⁹ identified in the genome, yielding 15,083 gene models. Third, GlimmerHMM⁷⁰, trained on the same set of eukaryotic BUSCO genes, identified 14,172 gene models. All putative gene models were combined using a weighted consensus approach using the EvidenceModeler software⁷¹. This resulting set of 16,832 gene models was further filtered to remove sequences shorter than 50 amino acids in length, repetitive elements like transposons or spanned gaps. This filtering resulted in the removal of 186 gene models, and 16,646 total gene models remained. Workflow orchestration was performed using funannotate (v.1.8.16)⁷². The completeness of the gene models was assessed using BUSCO v.4.0.4, with the Eukaryota database of near-universally single-copy orthologous genes from OrthoDB (v.10)^73,74. This analysis revealed that the genome annotation is 83.6% complete. Lastly, tRNAs were identified in the genome assembly using tRNAscan-SE (v.2.0.9)⁷⁵, resulting in 237 tRNA models.

A final decontamination step was performed to remove putative bacterial contaminant sequences from H. oceani (NCBI accession: GCF_963677335.1), the bacterial food source used in C. flexa cultures. Three scaffolds were identified as probable contaminants by BLAST and removed from the final assembly (Supplementary Information).

The final C. flexa genome assembly spanned 56,404,751 base pairs (bp) across 528 scaffolds, with a GC content of 50.83%. The assembly N50 and L50 values were 1,302,044 bp and 16 scaffolds, respectively, and the largest scaffold measured 3,795,244 bp, indicating high assembly contiguity. The final genome annotation encoded 14,084 genes with an overall BUSCO completeness of 82.8%.

Whole-genome short-read sequencing of C. flexa strains isolated in the field

Cultures of strains 1, 2 and 3 were established after a single-cell bottleneck. gDNA was collected from dense cultures using either a lysis buffer coupled with ethanol and sodium acetate precipitation, or the Blood & Cell Culture DNA Mini Kit (13323, Qiagen) (Supplementary Information). All gDNA samples were shipped to Eurofins Genomics for INVIEW Resequencing (10 million paired-end reads, Illumina 150 bp sequencing). Paired-end reads were quality assessed and trimmed using FASTP (v.0.20.1)⁷⁶. Qualified reads were aligned to the C. flexa reference genome using BWA-MEM (v.0.7.17)⁷⁷. The mapped reads were converted to BAM format and sorted using Samtools (v.1.18)⁷⁸. Duplicate reads were marked using GATK (v.4.1.9.0)⁷⁹, and BAM files were indexed using Samtools v.1.18. Variant calling was performed using GATK HaplotypeCaller, applying different ploidy assumptions (ploidy = 1, 2 and 4) to detect potential polymorphisms across samples. The resulting variants were jointly genotyped for each ploidy condition using the GATK GenotypeGVCF function⁸⁰.

Phylogenomic tree construction

We realized that, although strains 2 and 3 seemed to be haploid, strain 1 (ChoPs7) exhibited a diploid-like pattern of allelic frequencies in scaffold_1, as assessed using ploidyNGS⁸¹. To accommodate samples with potentially differing ploidy levels, we used SNPs called under a diploid assumption and found homozygous across all samples for downstream analysis. Variants were filtered using BCFtools in Samtools (v.1.18)⁸² with the following criteria: quality score > 30, filtered read depth > 4, variant type = ‘SNP’, minimum and maximum allowed alleles = 2 and homozygous genotypes across all samples. The resulting VCF files were converted to PHYLIP format as input for IQ-TREE using the vcfR package⁸³ in R (v.4.1.1). A phylogenomic tree of all samples was constructed based on the identified SNPs using IQ-TREE (v.2.3.2) with the general time reversible (GTR) substitution model, gamma-distributed rate variation and ascertainment bias correction^84,85. The output tree structure was visualized using iTOL (v.6.9.1)⁸⁶.

Identification of polymorphic sites under putative diversifying selection

Coding sequences for each strain were inferred from the variant data using vcf2fasta (https://github.com/yeeus/vcf2fasta). Inferred coding sequences were programmatically checked across all reference sequences, and genes with incorrect inferences were removed from the analysis. Variants were jointly called for all strains using the GATK (v.4.1.9.0) and filtered according to the recommended parameters from the GATK team. The inferred coding sequences of each gene were checked for length integrity, retaining only those that were a multiple of 3, of equal length across strains and containing no gaps. Genes that did not satisfy those criteria were aligned using Clustal-omega (v.1.2.4), and those genes that introduced gaps that were not in multiples of 3 (causing frameshifts) were removed from the downstream analyses. To compare predicted protein sequences across strains, we computed the number of non-synonymous substitutions across the strains using Biopython (v.1.85)⁸⁷. The K_a/K_s ratio (that is, the dN/dS ratio) was calculated between strains to identify genes under putative diversifying selection. Moreover, we additionally computed sliding-window K_a/K_s ratios to capture localized signals of selection for subregions of each sequence, using KaKs_Calculator 2.0 with a window size = 114 bp and step size = 6 bp, and computed the K_a/K_s ratio using KaKs_Calculator 3.0 with the MYN method^88,89. We defined regions with K_a/K_s ratio > 2 for at least 30 bp as a high K_a/K_s region.

To explore functional enrichment, we computed InterPro signatures overlapping high K_a/K_s regions. InterPro signatures of the C. flexa predicted proteome were obtained using InterProScan (v.5.50-84.0)^90,91. The InterPro signature enrichment was performed using Fisher’s exact test in the base R stats package, comparing the frequency of each InterPro signature within versus outside high K_a/K_s regions. Enrichment analysis results were visualized using tidyverse (v.2.0.0)⁹².

Kin recognition experiments

Approximately 40 ml of dense cultures of strains 1, 2 and 3 were collected, washed and stained with CellTrace CFSE (green) and CellTrace Far Red (magenta) as described above. Green- and magenta-labelled single-cell populations from each strain were mixed in a 1:1 ratio (5 × 10³ cells of each colour) in 1 ml SWC medium in a 24-well plate and incubated overnight at 25 °C. After incubation, 100 µl of each sample were transferred into a 96-well plate and colonies were imaged in by DIC microscopy and epifluorescence microscopy on the Zeiss Axio Observer Z.1 inverted microscope. Green and magenta cells were manually counted, and kin recognition was quantified by calculating a segregation index between every pairwise strain combination defined in an previous study⁹³ (Supplementary Information).

Statistical analyses

The significance of differences in pairwise comparisons was tested using the non-parametric Mann–Whitney U-test. Shapiro–Wilk normality test and F-test were used to evaluate data normality and the differences in variances between conditions, respectively. All statistical analyses were performed in R Statistical Software (v.4.4.1)⁹⁴ using the base stats package (v.3.6.3).

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability

The annotated reference C. flexa genome and associated raw reads have been deposited at Zenodo⁹⁵ (https://doi.org/10.5281/zenodo.13837466) and in the NBCI Sequence Read Archive (BioProject: PRJNA1381315). Raw short reads for strains 1, 2 and 3 have been deposited at Zenodo⁹⁶ (https://doi.org/10.5281/zenodo.13837614) and in the NCBI Sequence Read Archive (BioProject: PRJNA1381491).

Code availability

The codes used for SNP analysis have been deposited at GitHub (https://github.com/uhoro/RRRR_SNP_analysis).

References

Grosberg, R. K. & Strathmann, R. R. The evolution of multicellularity: a minor major transition? Annu. Rev. Ecol. Evol. Syst. 38, 621–654 (2007).
Article Google Scholar
Sebé-Pedrós, A., Degnan, B. M. & Ruiz-Trillo, I. The origin of Metazoa: a unicellular perspective. Nat. Rev. Genet. 18, 498–512 (2017).
Bonner, J. T. The origins of multicellularity. Integr. Biol. 1, 27–36 (1998).
Article Google Scholar
Brunet, T. & King, N. The origin of animal multicellularity and cell differentiation. Dev. Cell 43, 124–140 (2017).
Article CAS PubMed PubMed Central Google Scholar
Staps, M., Van Gestel, J. & Tarnita, C. E. Emergence of diverse life cycles and life histories at the origin of multicellularity. Nat. Ecol. Evol. 3, 1197–1205 (2019).
Article PubMed Google Scholar
Márquez-Zacarías, P., Conlin, P. L., Tong, K., Pentz, J. T. & Ratcliff, W. C. Why have aggregative multicellular organisms stayed simple? Curr. Genet. 67, 871–876 (2021).
Article PubMed Google Scholar
Leger, M. M. & Ruiz-Trillo, I. in The Evolution of Multicellularity (eds Herron, M. D. et al.) 157–186 (CRC Press, 2022).
Fisher, R. M., Cornwallis, C. K. & West, S. A. Group formation, relatedness, and the evolution of multicellularity. Curr. Biol. 23, 1120–1125 (2013).
Article CAS PubMed Google Scholar
Kapsetakia, S. E., Tep, A. & West, S. A. How do algae form multicellular groups? Evol. Ecol. Res. 18, 663–675 (2017).
Google Scholar
Fairclough, S. R., Dayel, M. J. & King, N. Multicellular development in a choanoflagellate. Curr. Biol. 20, R875–876 (2010).
Article CAS PubMed PubMed Central Google Scholar
Leadbeater, B. S. C. The Choanoflagellates: Evolution, Biology and Ecology (Cambridge Univ. Press, 2015).
Brunet, T. et al. Light-regulated collective contractility in a multicellular choanoflagellate. Science 366, 326–334 (2019).
Article ADS CAS PubMed Google Scholar
Lamża, Ł. Diversity of ‘simple’ multicellular eukaryotes: 45 independent cases and six types of multicellularity. Biol. Rev. 98, 2188–2209 (2023).
Article PubMed Google Scholar
Ros-Rocher, N., Pérez-Posada, A., Leger, M. M. & Ruiz-Trillo, I. The origin of animals: an ancestral reconstruction of the unicellular-to-multicellular transition. Open Biol. 11, 200359 (2021).
Article CAS PubMed PubMed Central Google Scholar
Ros-Rocher, N. & Brunet, T. What is it like to be a choanoflagellate? Sensation, processing and behavior in the closest unicellular relatives of animals. Anim. Cogn. 26, 1767–1782 (2023).
Article PubMed PubMed Central Google Scholar
Booth, D. S. & King, N. The history of Salpingoeca rosetta as a model for reconstructing animal origins. Curr. Top. Dev. Biol. 147, 73–91 (2022).
Ruiz-Trillo, I., Kin, K. & Casacuberta, E. The origin of metazoan multicellularity: a potential microbial black swan event. Annu. Rev. Microbiol. 77, 499–516 (2023).
Article CAS PubMed Google Scholar
Sebé-Pedrós, A. et al. Regulated aggregative multicellularity in a close unicellular relative of metazoa. eLife 2, e01287 (2013).
Article PubMed PubMed Central Google Scholar
Ros-Rocher, N. et al. Chemical factors induce aggregative multicellularity in a close unicellular relative of animals. Proc. Natl Acad. Sci. USA 120, e2216668120 (2023).
Article CAS PubMed PubMed Central Google Scholar
Dudin, O. et al. A unicellular relative of animals generates a layer of polarized cells by actomyosin-dependent cellularization. eLife 8, e49801 (2019).
Article PubMed PubMed Central Google Scholar
Shah, H. et al. Life-cycle-coupled evolution of mitosis in close relatives of animals. Nature 630, 116–122 (2024).
Article ADS CAS PubMed PubMed Central Google Scholar
Olivetta, M., Bhickta, C., Chiaruttini, N., Burns, J. & Dudin, O. A multicellular developmental program in a close animal relative. Nature 635, 382–389 (2024).
Article ADS CAS PubMed Google Scholar
Shabardina, V. et al. Ichthyosporea: a window into the origin of animals. Commun. Biol. 7, 915 (2024).
Article PubMed PubMed Central Google Scholar
Marshall, W. L. & Berbee, M. L. Facing unknowns: living cultures (Pirum gemmata gen. nov., sp. nov., and Abeoforma whisleri, gen. nov., sp. nov.) from invertebrate digestive tracts represent an undescribed clade within the unicellular opisthokont lineage Ichthyosporea (Mesomycetozoea). Protist 162, 33–57 (2011).
Article PubMed Google Scholar
Reyes-Rivera, J. et al. Nitric oxide signaling controls collective contractions in a colonial choanoflagellate. Curr. Biol. 32, 2539–2547 (2022).
Article CAS PubMed Google Scholar
Garbary, D. in Algae and Cyanobacteria in Extreme Environments (ed. Seckbach, J.) 173–191 (Springer, 2007).
Amsalem, E. & Rilov, G. High thermal plasticity, and vulnerability, in extreme environments at the warm distributional edge: the case of a tidepool shrimp. J. Exp. Mar. Biol. Ecol. 545, 151641 (2021).
Article Google Scholar
Leadbeater, B. S. C. & Karpov, S. A. Cyst formation in a freshwater strain of the choanoflagellate Desmarella moniliformis Kent. J. Euk. Microbiol. 47, 433–439 (2000).
Article CAS PubMed Google Scholar
Stoupin, D. et al. Cryptic diversity within the choanoflagellate morphospecies complex Codosiga botrytis—phylogeny and morphology of ancient and modern isolates. Eur. J. Protistol. 48, 263–273 (2012).
Article PubMed Google Scholar
Jeuck, A., Arndt, H. & Nitsche, F. Extended phylogeny of the Craspedida (Choanomonada). Eur. J. Protistol. 50, 430–443 (2014).
Article PubMed Google Scholar
Schaap, P. & Schilde, C. Encystation: the most prevalent and underinvestigated differentiation pathway of eukaryotes. Microbiology 164, 727–739 (2018).
Article CAS PubMed Google Scholar
Di Ciano, C. et al. Osmotic stress-induced remodeling of the cortical cytoskeleton. Am. J. Physiol. Cell Physiol. 283, C850–C865 (2002).
Article PubMed Google Scholar
Balachandra, S., Sarkar, S. & Amodeo, A. A. The nuclear-to-cytoplasmic ratio: coupling DNA content to cell size, cell cycle, and biosynthetic capacity. Annu. Rev. Genet. 56, 165–185 (2022).
Article CAS PubMed PubMed Central Google Scholar
Koehl, M.aR. Selective factors in the evolution of multicellularity in choanoflagellates. J. Exp. Zool. B 336, 315–326 (2021).
Article CAS Google Scholar
Fenchel, T. Filter-feeding in colonial protists. Protist 170, 283–286 (2019).
Article PubMed Google Scholar
Kirkegaard, J. B. & Goldstein, R. E. Filter-feeding, near-field flows, and the morphologies of colonial choanoflagellates. Phys. Rev. E 94, 052401 (2016).
Article ADS PubMed PubMed Central Google Scholar
L’Etoile, N. J. & King-Smith, C. Rosette colonies of choanoflagellates (Salpingoeca rosetta) show increased food vacuole formation compared with single swimming cells. J. Euk. Microbiol. 67, 263–267 (2020).
Article PubMed Google Scholar
Sternfeld, J. Evidence for differential cellular adhesion as the mechanism of sorting-out of various cellular slime mold species. J. Embryol. Exp. Morphol. 53, 163–178 (1979).
CAS PubMed Google Scholar
Wolpert, L. & Szathmáry, E. Multicellularity: evolution and the egg. Nature 420, 745–745 (2002).
Article ADS CAS PubMed Google Scholar
Ho, H.-I., Hirose, S., Kuspa, A. & Shaulsky, G. Kin recognition protects cooperators against cheaters. Curr. Biol. 23, 1590–1595 (2013).
Article CAS PubMed PubMed Central Google Scholar
King, N. et al. The genome of the choanoflagellate Monosiga brevicollis and the origin of metazoans. Nature 451, 783–788 (2008).
Article ADS CAS PubMed PubMed Central Google Scholar
Fairclough, S. R. et al. Premetazoan genome evolution and the regulation of cell differentiation in the choanoflagellate Salpingoeca rosetta. Genome Biol. 14, R15 (2013).
Article PubMed PubMed Central Google Scholar
Benabentos, R. et al. Polymorphic members of the lag gene family mediate kin discrimination in Dictyostelium. Curr. Biol. 19, 567–572 (2009).
Article CAS PubMed PubMed Central Google Scholar
La Fortezza, M., Schaal, K. A. & Velicer, G. J. in The Evolution of Multicellularity (eds Herron, M. D. et al.) 89–110 (CRC Press, 2022).
Tang, S., Pichugin, Y. & Hammerschmidt, K. An environmentally induced multicellular life cycle of a unicellular cyanobacterium. Curr. Biol. 33, 764–769 (2023).
Article CAS PubMed Google Scholar
Mizuno, K. et al. Novel multicellular prokaryote discovered next to an underground stream. eLife 11, e71920 (2022).
Article CAS PubMed PubMed Central Google Scholar
Barrere, J., Nanda, P. & Murray, A. W. Alternating selection for dispersal and multicellularity favors regulated life cycles. Curr. Biol. 33, 1809–1817 (2023).
Article CAS PubMed PubMed Central Google Scholar
Pichugin, Y., Park, H. J. & Traulsen, A. Evolution of simple multicellular life cycles in dynamic environments. J. R. Soc. Interface 16, 20190054 (2019).
Article PubMed PubMed Central Google Scholar
Staps, M., Van Gestel, J. & Tarnita, C. E. in The Evolution of Multicellularity (eds Herron, M. D. et al.) 53–70 (CRC Press, 2022).
Isaksson, H., Lind, P. & Libby, E. Adaptive evolutionary trajectories in complexity: transitions between unicellularity and facultative differentiated multicellularity. Proc. Natl Acad. Sci. USA 122, e2411692122 (2025).
Article CAS PubMed PubMed Central Google Scholar
Brunet, T. & King, N. in The Evolution of Multicellularity (eds Herron, M. D. et al.) 251–278 (CRC Press, 2022).
Sauer, K. et al. The biofilm life cycle: expanding the conceptual model of biofilm formation. Nat. Rev. Microbiol. 20, 608–620 (2022).
Article CAS PubMed PubMed Central Google Scholar
Chavhan, Y., Dey, S. & Lind, P. A. Bacteria evolve macroscopic multicellularity by the genetic assimilation of phenotypically plastic cell clustering. Nat. Commun. 14, 3555 (2023).
Article ADS CAS PubMed PubMed Central Google Scholar
Pentz, J. T. et al. Evolutionary consequences of nascent multicellular life cycles. eLife 12, e84336 (2023).
Article CAS PubMed PubMed Central Google Scholar
Rose, C. J. & Hammerschmidt, K. What do we mean by multicellularity? The evolutionary transitions framework provides answers. Front. Ecol. Evol. 9, 730714 (2021).
Article Google Scholar
Hammerschmidt, K., Rose, C. J., Kerr, B. & Rainey, P. B. Life cycles, fitness decoupling and the evolution of multicellularity. Nature 515, 75–79 (2014).
Article ADS CAS PubMed Google Scholar
Jahan, I., Scott, T. J., Strassmann, J. E. & Queller, D. C. Testing the coordination hypothesis: incompatibilities in aggregative development of an experimentally evolved social amoeba. Evol. Lett. 9, 236–248 (2025).
Article PubMed Google Scholar
Smith, J., Strassmann, J. E. & Queller, D. C. Fine-scale spatial ecology drives kin selection relatedness among cooperating amoebae. Evolution 70, 848–859 (2016).
Article PubMed Google Scholar
Haag, C. R., Riek, M., Hottinger, J. W., Pajunen, V. I. & Ebert, D. Founder events as determinants of within-island and among-island genetic structure of Daphnia metapopulations. Heredity 96, 150–158 (2006).
Article CAS PubMed Google Scholar
Black, A. J., Bourrat, P. & Rainey, P. B. Ecological scaffolding and the evolution of individuality. Nat. Ecol. Evol. 4, 426–436 (2020).
Article PubMed Google Scholar
Schindelin, J. et al. Fiji: an open-source platform for biological-image analysis. Nat. Methods 9, 676–682 (2012).
Article CAS PubMed PubMed Central Google Scholar
Pachitariu, M. & Stringer, C. Cellpose 2.0: how to train your own model. Nat. Methods 19, 1634–1641 (2022).
Article CAS PubMed PubMed Central Google Scholar
Van Der Walt, S. et al. scikit-image: image processing in Python. PeerJ 2, e453 (2014).
Article PubMed PubMed Central Google Scholar
Richter, D. J., Fozouni, P., Eisen, M. B. & King, N. Gene family innovation, conservation and loss on the animal stem lineage. eLife 7, e34226 (2018).
Article PubMed PubMed Central Google Scholar
Buchfink, B., Reuter, K. & Drost, H.-G. Sensitive protein alignments at tree-of-life scale using DIAMOND. Nat. Methods 18, 366–368 (2021).
Article CAS PubMed PubMed Central Google Scholar
Slater, G. S. C. & Birney, E. Automated generation of heuristics for biological sequence comparison. BMC Bioinform. 6, 31 (2005).
Article Google Scholar
Stanke, M. et al. AUGUSTUS: ab initio prediction of alternative transcripts. Nucleic Acids Res. 34, W435–W439 (2006).
Article CAS PubMed PubMed Central Google Scholar
Korf, I. Gene finding in novel genomes. BMC Bioinform. 5, 59 (2004).
Article Google Scholar
Simão, F. A., Waterhouse, R. M., Ioannidis, P., Kriventseva, E. V. & Zdobnov, E. M. BUSCO: assessing genome assembly and annotation completeness with single-copy orthologs. Bioinformatics 31, 3210–3212 (2015).
Article PubMed Google Scholar
Majoros, W. H., Pertea, M. & Salzberg, S. L. TigrScan and GlimmerHMM: two open source ab initio eukaryotic gene-finders. Bioinformatics 20, 2878–2879 (2004).
Article CAS PubMed Google Scholar
Haas, B. J. et al. Automated eukaryotic gene structure annotation using EVidenceModeler and the Program to Assemble Spliced Alignments. Genome Biol. 9, R7 (2008).
Article PubMed PubMed Central Google Scholar
Palmer J. M. & Stajich J. Funannotate v1.8.1: eukaryotic genome annotation (v.1.8). Zenodo https://doi.org/10.5281/zenodo.1134477 (2020).
Kuznetsov, D. et al. OrthoDB v11: annotation of orthologs in the widest sampling of organismal diversity. Nucleic Acids Res. 51, D445–D451 (2023).
Article CAS PubMed PubMed Central Google Scholar
Manni, M., Berkeley, M. R., Seppey, M., Simão, F. A. & Zdobnov, E. M. BUSCO update: novel and streamlined workflows along with broader and deeper phylogenetic coverage for scoring of eukaryotic, prokaryotic, and viral genomes. Mol. Biol. Evol. 38, 4647–4654 (2021).
Article CAS PubMed PubMed Central Google Scholar
Chan, P. P. & Lowe, T. M. in Gene Prediction, Vol. 1962 (ed. Kollmar, M.) 1–14 (Springer, 2019).
Chen, S., Zhou, Y., Chen, Y. & Gu, J. fastp: an ultra-fast all-in-one FASTQ preprocessor. Bioinformatics 34, i884–i890 (2018).
Article PubMed PubMed Central Google Scholar
Li, H. Aligning sequence reads, clone sequences and assembly contigs with BWA-MEM. Preprint at arxiv.org/abs/1303.3997 (2013).
Li, H. et al. The Sequence Alignment/Map format and SAMtools. Bioinformatics 25, 2078–2079 (2009).
Article PubMed PubMed Central Google Scholar
Van der Auwera, G. A. & O’Connor B. D. Genomics in the Cloud: Using Docker, GATK, and WDL in Terra. (O’Reilly Media, 2020).
Poplin, R. et al. Scaling accurate genetic variant discovery to tens of thousands of samples. Preprint at bioRxiv https://doi.org/10.1101/201178 (2017).
Augusto Corrêa Dos Santos, R., Goldman, G. H. & Riaño-Pachón, D. M. ploidyNGS: visually exploring ploidy with next generation sequencing data. Bioinformatics 33, 2575–2576 (2017).
Article PubMed Google Scholar
Danecek, P. et al. Twelve years of SAMtools and BCFtools. GigaScience 10, giab008 (2021).
Article PubMed PubMed Central Google Scholar
Knaus, B. J. & Grünwald, N. J. vcfr: a package to manipulate and visualize variant call format data in R. Mol. Ecol. Resour. 17, 44–53 (2017).
Article CAS PubMed Google Scholar
Nguyen, L.-T., Schmidt, H. A., Von Haeseler, A. & Minh, B. Q. IQ-TREE: a fast and effective stochastic algorithm for estimating maximum-likelihood phylogenies. Mol. Biol. Evol. 32, 268–274 (2015).
Article CAS PubMed Google Scholar
Hoang, D. T., Chernomor, O., Von Haeseler, A., Minh, B. Q. & Vinh, L. S. UFBoot2: improving the ultrafast bootstrap approximation. Mol. Biol. Evol. 35, 518–522 (2018).
Article CAS PubMed PubMed Central Google Scholar
Letunic, I. & Bork, P. Interactive Tree of Life (iTOL) v6: recent updates to the phylogenetic tree display and annotation tool. Nucleic Acids Res. 52, W78–W82 (2024).
Article PubMed PubMed Central Google Scholar
Cock, P. J. A. et al. Biopython: freely available Python tools for computational molecular biology and bioinformatics. Bioinformatics 25, 1422–1423 (2009).
Wang, D., Zhang, Y., Zhang, Z., Zhu, J. & Yu, J. KaKs_Calculator 2.0: a toolkit incorporating gamma-series methods and sliding window strategies. Genom. Proteom. Bioinform. 8, 77–80 (2010).
Zhang, Z. KaKs_Calculator 3.0: calculating selective pressure on coding and non-coding sequences. Genom. Proteom. Bioinform. 20, 536–540 (2022).
Blum, M. et al. The InterPro protein families and domains database: 20 years on. Nucleic Acids Res. 49, D344–D354 (2021).
Article CAS PubMed PubMed Central Google Scholar
Jones, P. et al. InterProScan 5: genome-scale protein function classification. Bioinformatics 30, 1236–1240 (2014).
Article CAS PubMed PubMed Central Google Scholar
Wickham, H. et al. Welcome to the Tidyverse. JOSS 4, 1686 (2019).
Article ADS Google Scholar
Estrela, S. & Brown, S. P. Metabolic and demographic feedbacks shape the emergent spatial structure and function of microbial communities. PLoS Comput. Biol. 9, e1003398 (2013).
Article ADS PubMed PubMed Central Google Scholar
R Core Team. R: A Language and Environment for Statistical Computing (R Foundation for Statistical Computing, 2021)
Ros-Rocher, N. et al. Data for ‘Clonal-aggregative multicellularity tuned by salinity in a choanoflagellate’. Zenodo https://doi.org/10.5281/zenodo.13837466 (2026).
Ros-Rocher, N. et al. Data for ‘Clonal-aggregative multicellularity tuned by salinity in a choanoflagellate’. Zenodo https://doi.org/10.5281/zenodo.13837614 (2026).
Dayel, M. J. et al. Cell differentiation and morphogenesis in the colony-forming choanoflagellate Salpingoeca rosetta. Dev. Biol. 357, 73–82 (2011).
Article CAS PubMed PubMed Central Google Scholar

Download references

Acknowledgements

We thank N. King for support during this project and for providing feedback; the personnel of the Shete Boka National Park in Curaçao for access to the field site; T. Linden for help during preliminary field experiments and feedback on the manuscript; G. Milon and N. Whiteman for advice on laboratory and field experiments, respectively; D. S. Booth for providing J.R.-R. with access to microscopes and laboratory space during revision of this work after peer review; Y. Chavhan, S. Chou, S. Kapsetaki and I. Ruiz-Trillo as well as the members of the Multicellgenome laboratory for providing feedback on the preprint version of the manuscript; all of the members of the Brunet and King laboratories for feedback and discussions, and D. de Haan for feedback on the final version of the manuscript; M. Albert, S. Rigaud and J.-Y. Tinevez (Image Analysis Hub, Institut Pasteur) for advice on image analysis; P.-H. Commere and S. Megharba (Flow Cytometry Platform, Institut Pasteur) for technical support with single-cell isolation of C. flexa; the members of the HPC Core Facility of the Institut Pasteur for help with bioinformatic analyses; T. Swale and T. Rickman for help with genome sequencing and assembly and for drafting the relevant Methods sections; and J. Ramirez and K. Dinshaw for providing technical support during preliminary experiments. N.R.-R. was supported by the European Union’s Horizon Europe research and innovation funding programme under a Marie Skłodowska-Curie Actions grant (FlexAggon, grant no. 101106415), by the Institut Pasteur and by the CNRS (UMR3691). J.R.-R. was supported by a National Science Foundation Graduate Research Fellowship Program (grant no. 1752814) and the National Science Foundation Postdoctoral Research Fellowship in Biology (grant no. 2507925). U.H. is supported by the PPU programme of the Institut Pasteur. Work in the King laboratory (T.B., J.R.-R., B.T.L. and M.C.C.) is supported by the Howard Hughes Medical Institute. J.L.S. is a Howard Hughes Medical Institute Awardee of the Life Sciences Research Foundation. T.B. and work in the Brunet laboratory are supported by the Institut Pasteur (G5 package), the ERC (EvoMorphoCell, grant no. 101040745), the Vallee Foundation, the EMBO Young Investigator Programme, the LabEx Revive (ANR-10-LABX-73) and the CNRS (UMR 3691). This research was funded in part by the European Research Council (grant no. 101040745). Views and opinions expressed are those of the author(s) only and do not necessarily reflect those of the European Union or the European Research Council. Neither the European Union nor the granting authority can be held responsible for them.

Author information

Josean Reyes-Rivera
Present address: Department of Biochemistry and Biophysics, University of California, San Francisco, San Francisco, CA, USA
Ben T. Larson
Present address: Department of Biological Sciences and Center for Biotechnology and Interdisciplinary Studies, Rensselaer Polytechnic Institute, Troy, NY, USA
Maxwell C. Coyle
Present address: Department of Molecular and Cellular Biology, Harvard University, Cambridge, MA, USA
These authors contributed equally: Núria Ros-Rocher, Josean Reyes-Rivera

Authors and Affiliations

Evolutionary Cell Biology and Evolution of Morphogenesis Unit, Institut Pasteur, Université Paris-Cité, CNRS UMR3691, Paris, France
Núria Ros-Rocher, Uzuki Horo, Chantal Combredet, Yeganeh Foroughijabbari & Thibaut Brunet
Howard Hughes Medical Institute and the Department of Molecular and Cell Biology, University of California, Berkeley, Berkeley, CA, USA
Josean Reyes-Rivera, Ben T. Larson, Maxwell C. Coyle & Jacob L. Steenwyk
Caribbean Research and Management of Biodiversity foundation (CARMABI), Willemstad, Curaçao
Erik A. T. Houtepen & Mark J. A. Vermeij
Department of Freshwater and Marine Ecology, Institute for Biodiversity and Ecosystem Dynamics, University of Amsterdam, Amsterdam, The Netherlands
Mark J. A. Vermeij

Authors

Núria Ros-Rocher
View author publications
Search author on:PubMed Google Scholar
Josean Reyes-Rivera
View author publications
Search author on:PubMed Google Scholar
Uzuki Horo
View author publications
Search author on:PubMed Google Scholar
Chantal Combredet
View author publications
Search author on:PubMed Google Scholar
Yeganeh Foroughijabbari
View author publications
Search author on:PubMed Google Scholar
Ben T. Larson
View author publications
Search author on:PubMed Google Scholar
Maxwell C. Coyle
View author publications
Search author on:PubMed Google Scholar
Erik A. T. Houtepen
View author publications
Search author on:PubMed Google Scholar
Mark J. A. Vermeij
View author publications
Search author on:PubMed Google Scholar
Jacob L. Steenwyk
View author publications
Search author on:PubMed Google Scholar
Thibaut Brunet
View author publications
Search author on:PubMed Google Scholar

Contributions

N.R.-R., J.R.-R. and T.B. conceptualized the study and coordinated the work. M.C.C., B.T.L. and T.B. performed preliminary fieldwork that identified the Shete Boka site that gave first insights in the C. flexa salinity range. N.R.-R., J.R.-R. and T.B. collected fieldwork data included in this study. N.R.-R., J.R.-R. and T.B. performed all laboratory experiments, strain isolations and bioinformatic analyses (except for those specified below). B.T.L. performed preliminary aggregation assays and aphidicolin treatments. Y.F. optimized live imaging of dual-labelling aggregation experiments and performed growth curves under the addition of salts under the supervision of N.R.-R.; J.L.S. performed genome assembly and annotation. U.H. performed behavioural assays with technical support from N.R.-R. and SNP analyses with technical guidance from J.L.S.; C.C. performed single-cell bottleneck isolations and generated the material for genome sequencing of strains 1, 2 and 3. E.A.T.H. and M.J.A.V. provided support for fieldwork studies. E.A.T.H. performed drone orthomap imagery. J.R.-R. wrote the first draft of the manuscript, which was subsequently edited by N.R.-R. and T.B;. N.R.-R. made the figures with feedback from J.R.-R. and T.B.; T.B. supervised the study. All of the authors contributed to the review of the manuscript before submission for publication and approved the final version.

Corresponding authors

Correspondence to Núria Ros-Rocher, Josean Reyes-Rivera or Thibaut Brunet.

Ethics declarations

Competing interests

J.L.S. is an advisor for ForensisGroup.

Peer review

Peer review information

Nature thanks Omaya Dudin and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Extended data figures and tables

Extended Data Fig. 1 Salinity distribution and presence of sheets across Curaçao splash pools in Exped-A and Exped-B.

(a) Representative images of the landscape in Shete Boka National Park, with a close-up image of splash pools (white arrowheads). (b) Representative images of splash pools near Boka Pistol and Boka Wandomi, showing refilling from the sea via splash or waves (white arrowheads). (c) Orthomap image of the area sampled in Exped-B acquired by a DJI Mavic 2 Enterprise drone. (d) Distribution of seawater salinity of sampled splash pools in Exped-A and Exped-B (n = 150 splash pools). Pink and turquoise colours indicate that sheets were respectively found and not found in splash pool samples. Exped-A data are from the first day of collection (not the follow-up time course). Figure related to Fig. 3, Video S14, and Table S1.

Extended Data Fig. 2 Species identification, morphology and behaviour of C. flexa sheets collected in splash pools.

(a) Representative brightfield images of sheets observed in natural splash pool samples. Note the two different sheet conformations: ‘flagella in’ and ‘flagella out’. (b) Snapshots of a brightfield timelapse movie of a medium-sized sheet inverting its curvature from a ‘flagella in’ to a ‘flagella out’ conformation in response to a light-to-dark transition. (c) Representative brightfield images of multiple sheets in the same field of view identified in natural splash pool samples. Imaging of sheets in a-c performed in N = 4 independent experiments (n = 18 sheets). (d) 18S rDNA phylogenetic tree of newly isolated C. flexa single-sheet-bottlenecked and clonal cultures (turquoise), several other choanoflagellates (orange), and other opisthokonts, including animals (red), other holozoans and fungal species. Shown is a Maximum Likelihood phylogenetic tree of 18S rDNA sequences of three single-sheet-bottlenecked cultures isolated from M44, M60, and M61 (Exped-B) and three clonal cultures isolated from M44 and M60 (Exped-B). Support values on nodes: approximate likelihood ratio test (aLRT). Figure related to Fig. 3f and Video S15.

Extended Data Fig. 3 Splash pools undergo natural cycles of evaporation and refilling.

(a, c, e) Representative images of monitored splash pools near Boka Kalki (Sp15) and Boka Wandomi (Sp50 and Sp69) during a 9-day time course as part of Exped-A (n = 15 splash pools). Some splash pools experienced evaporation (a, n = 7), complete desiccation (c, n = 4), and evaporation and refilling (e, n = 4). Dashed lines: splash pool outlines. Upper left: salinity in ppt. Upper right: time in days. (b, d, f) Salinity (top) and depth (bottom) over time. Filled circles: active sheets observed. Empty circles: no sheets observed. Grey cross: one irregular non-motile sheet observed. Grey area: salinity saturation (outside the refractometer measurement range). Grey dashed lines in d and f: dry periods during which splash pools were completely desiccated. Orange circle in f: refilling events. Figure related to Fig. 3, Extended Data Figs. 1–2, and Table S1.

Extended Data Fig. 4 Gradual evaporation triggers loss of multicellularity in sheets from natural splash pool samples.

Representative brightfield images of sheets identified in natural samples from Sp64 (Exped-A) over a 6-day gradual evaporation time course in the laboratory. After 48 h, sheets had disappeared and had been replaced by single cells appeared (black arrowheads). Time under gradual evaporation is depicted in the upper right. Salinity was not measured at 96 and 144 h. N = 2 independent replicates, two technical replicates each (n = 4). Figure related to Figs. 3 and 4.

Extended Data Fig. 5 Additon of salts induces dissociation-encystation, similar to gradual evaporation.

(a-c) DIC images of C. flexa cells over a 6-day time course experiment where salinity was progressively increased 1-fold daily by the addition of seawater salts after 3 days of growth show that sheets dissociate and cells undergo morphological changes in hypersaline media. At the end of the 6-day time course, cells grown for 6 days at 1X salinity (a, no salinity increase) remain in multicellular sheets composed of flagellated cells (i) with a collar (green pseudocolour) and a flagellum (magenta pseudocolour). Under progressive salinity increase by salts addition, sheets dissociate into unicellular cysts (b-c). Cysts often lack a collar and a flagellum, and exhibit filopodia-like protrusions (pink arrowheads) after 2 days at 3X salinity (b) and after 1 day at 4X salinity (c). Scale bars in (i-iii): 10 µm. (d) Quantification of cell number over a 6-day time course where salinity was progressively increased 1-fold daily by the addition of salts after 3 days of growth. 1X salinity corresponds to 35 ppt. (e) Growth rate of cells over a 6-day time course experiment in d. Black circles represent mean values and error bars represent standard deviation. Mean values of each biological replicate are represented as diamonds. p-values in e by the Mann-Whitney U test. N = 2 biological replicates, two technical replicates each (n = 4). Figure related to Fig. 4f–i, m.

Extended Data Fig. 6 Latrunculin B treatment causes sheet dissociation and prevents aggregation in C. flexa.

(a) Experimental design for (a-d): sheets were treated with 2 µM of latrunculin B and imaged 4 h post-treatment (hpt). (b) Stills of C. flexa sheets treated with 2 µM latrunculin B and imaged 4 hpt. Control sheets treated with drug carrier (ethanol) or water remain multicellular. (b’) Zoom-in of white squares in b. (c) Quantification of average particle area of treated and untreated cells in b. (d) 3D reconstructions of confocal z-stacks of latrunculin B-treated and control sheets fixed 4 hpt, with a membrane dye labelling the cell body and flagella (FM 4-64FX, magenta) and F-actin dye labelling the collar (Phalloidin 488, white). Note that sheets treated with latrunculin B dissociate into single cells and lose the collar (white arrowheads), whereas untreated cells remain multicellular. (e) Experimental design for (f-h): single cells were treated with 2 µM of latrunculin B and imaged 4 hpt. (f) Stills of C. flexa single cells treated with 2 µM latrunculin B and imaged 4 hpt. Controls treated with drug carrier (ethanol) or water aggregate and form sheets normally. (f’) Zoom-in on areas outlined by white squares in f. (g) Quantification of average particle area of treated and untreated cells in f. (h) 3D reconstructions of confocal z-stacks of latrunculin B-treated and control cells fixed 4 hpt, stained as in d. Note that cells treated with latrunculin B lose the collar and remain in single-cell form (white arrowheads), whereas untreated cells aggregate and form sheets normally. All experiments were performed in N = 3 biological replicates, two technical replicates per condition (n = 6). p-values in c,g by the Mann-Whitney U test (n.s., non-significant).

Extended Data Fig. 7 Aggregative sheets are comparable in morphology to control sheets.

(a) Mean sheet area at 1X (grey) and 2X (turquoise) salinities over a 2-hour aggregation experiment. (b) Cell density measured in n = 12 splash pools (SpA to SpL, Exped-C) compared to the range used in laboratory aggregation experiments (grey area). Diamonds are average estimates, and bars indicate uncertainty (ranging from minimal to maximal estimates). (c) Mean sheet area at 1X (grey) and 2X (turquoise) salinities across different cell densities. (d-g) 3D reconstructions of confocal z-stacks of aggregative (d-e) and control sheets (f-g), with a membrane dye labelling the cell body and flagella (FM 4-64FX, magenta) and F-actin dye labelling the collar (Phalloidin 488, white). (d’-g’). Mid-z cross-sections (dashed lines) in d-g. Note that cells within aggregative sheets are connected by collar-collar contacts (white arrowheads) and have aligned apico-basal axes as in control sheets. (h) Quantification of cell number per colony in d-g. (h’) Quantification of sheet area in d-g. (i) Quantification of percentage of aligned cells within colonies in d-g. (i’) Quantification of mean collar-collar angles between cells per sheet in d-g. (j) Quantification of circularity of sheets in d-g. Black circles: means. Error bars: standard deviations. Diamonds: mean values of each independent replicate. p-values in a and h-j by the Mann-Whitney U test (n.s., non-significant). All experiments shown (except in b) were performed in N = 3 biological replicates, two technical replicates each (n = 41 aggregative sheets; n = 46 control sheets).

Extended Data Fig. 8 Aggregative sheets invert in response to light.

(a) Quantification of light-to-dark inversions comparing aggregative and control sheets, showing normalized area change of five sheets per condition (n = 5) before (-20 sec, grey) and after (20 sec, blue) light-to-dark transitions. (b) Percentage of inverted control and aggregative sheets. All experiments were performed in N = 3 biological replicates, two technical replicates per condition (n = 60 sheets). p-values in a-b by the Mann-Whitney U test (n.s., non-significant). Figure related to Fig. 5k.

Extended Data Fig. 9 Aggregation is species-specific.

(a) (top) Stills of C. flexa single cells labelled with CellTrace Far Red (magenta) and S. rosetta single cells labelled with CellTrace CFSE (green) 0 h post-dissociation. (bottom) Stills of C. flexa and S. rosetta 20-22 h post-dissociation. Note that C. flexa forms sheets composed of only C. flexa cells (magenta arrowheads), excluding S. rosetta cells (green arrowheads), which form chains clonally⁹⁷. Unmixed C. flexa and S. rosetta cells were used as controls. (b) Quantification of percentage of C. flexa or S. rosetta cells in sheets in mixed conditions 20-22 h post-dissociation. N = 3 biological replicates, two technical replicates per condition (n = 211 sheets). p-value in b by the Mann-Whitney U test. Figure related to Fig. 6.

Extended Data Table 1 Estimates of cell division rate for various microorganisms in nature. Data from the literature

Full size table

Supplementary information

Supplementary Information

Supplementary Methods, Supplementary Figs. 1–11, Supplementary Table 2 and Supplementary References.

Reporting Summary

Supplementary Table 1

Zipped file containing Supplementary Table 1 (fieldwork data collection) and the full table legend.

Supplementary Data 1–6

This zipped file contains Supplementary Data files 1–6 and legends for the Supplementary Data files.

Peer Review file

Supplementary Videos 1–19

Zipped file containing Supplementary Videos 1–19 and their descriptions.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Cite this article

Ros-Rocher, N., Reyes-Rivera, J., Horo, U. et al. Clonal-aggregative multicellularity tuned by salinity in a choanoflagellate. Nature (2026). https://doi.org/10.1038/s41586-026-10137-y

Download citation

Received: 01 October 2024
Accepted: 14 January 2026
Published: 25 February 2026
Version of record: 25 February 2026
DOI: https://doi.org/10.1038/s41586-026-10137-y