The Single-Celled Ancestors of Animals: A History of Hypotheses

Animals, with their complex and obligate multicellularity, evolved from microbial eukaryotes that were likely obligately or facultatively unicellular. The nature of the unicellular progenitors of animals has intrigued biologists since the late 19th century, coinciding with the parallel spread of the cell theory and the theory of evolution. However, views on the ancestry of animals have been extremely varied. The huge diversity of single-celled organisms, the tremendous plasticity of animal cellular phenotypes, and the difficulties of organizing both into clear phylogenies in the pre-molecular era allowed a wide range of hypotheses to flourish, with nearly every major single-celled lineage, at one time or another, having been proposed as the precursor of animals. Most of these hypotheses never gained followers beyond their originator (such as the ideas that animals evolved directly from either bacteria, Volvox or fungi) and will not be discussed here. Three concepts, however, have been enduring and influential: (1) the amoeboid theory; (2) the flagellate theory; and the (3) the ciliate theory – to which a fourth category can now be added: (4) a mixed model, in which the ancestor was phenotypically plastic. We will discuss their origin, history, and current relevance.


I. Origin of the question: the cell theory and the concept of common descent
The question of the single-celled ancestor of animals only makes sense in the light of two concepts that are now central to biology and that emerged in parallel in the second half of the 19 th century: (1) the cell theory, which posits that all living beings are composed of cells (some of many, some of only one) (Schwann 1839;Schleiden 1839); and (2) the theory of common descent, which posits that all living speciesunicellular or multicellulardescended from a single common ancestor (Darwin and Wallace 1858).
That all living beings are made of cells is the first fact many of us learned about biology, and is so familiar that we sometimes take it for granted. But the cellular organization of all life forms was not initially obvious, and it took a full 250 years after the invention of the microscope for this idea to gain general acceptance. Two of the first people to observe microorganisms ( van Leeuwenhoek 1677;Müller 1786) indiscriminately used the terms "infusorians" or "animalcules" to describe what we now think was a mélange of unicellular Table 1
Once the cell theory was accepted, several early cell biologists (including Meyen (1839), Dujardin (1841), Barry (1843) and von Siebold (1845)) took the leap to posit that the simplest life forms might consist of only one cell (reviewed in (Leadbeater and McCready 2002)).
The theory of evolution emerged in parallel with the cell theory. The first elaborate theory of evolution, proposed in 1809 by the French biologist Jean-Baptiste de Lamarck (1744Lamarck ( -1829, assumed that life started with the spontaneous generation of "infusorians"including both protists and small animals (Lamarck 1809). Infusorians were then inferred to have gradually evolved into all other organisms through a progressive increase in size and complexity, with no individual step that would have clearly paralleled our modern concept of a transition to multicellularity. Lamarck's ideas attracted attention and criticism, but the concept of common descent did not become widely accepted until after the debate spurred by the theory of evolution through natural selection proposed by Charles Darwin (1809Darwin ( -1882 and Alfred Russel Wallace   (Darwin 1859;Darwin and Wallace 1858). Their theory was the first to propose a plausible mechanism for descent with modification, and thus brought new credibility to the concept of evolution.
By the end of the 19 th century, the scientific stage was set for considering the origin of animals: both evolution and cell theory had gained widespread acceptance, and three of the most abundant and charismatic groups of single-celled organisms had been identifiedflagellates, ciliates and amoebae (Table 1). Quickly, all three were considered as potential ancestors of animals.

II. Haeckel's hypothesis: amoebae as ancestors
The first researcher to attempt to reconstruct the unicellular progenitor of animals was the German biologist Ernst Haeckel (1834-1919), arguably one of Darwin's most high-profile supporters in continental Europe (Richards 2008). While Haeckel's name is most often mentioned today in the context of his now-obsolete theory of recapitulation (according to which development directly recapitulated evolution (Gould 1977)) or for his controversial drawings of vertebrate embryos (Pennisi 1997;Richards 2009), his contributions to biology were much broader, and one can get an idea of their scope by considering that he coined the words "ecology", "ontogeny", "phylogeny", and "gastrulation" among many others.
Haeckel had an exceptionally ambitious research program: organizing all of life's diversity into a phylogenetic framework andif that was not enoughreconstituting the extinct ancestors that occupied the most important nodes of that tree. In his attempt to reconstitute the single-celled progenitor of animals, he inferred it was an amoeba based on two independent sources of evidence: (1) his theory of recapitulation; and (2) Magosphaera planula, a mysterious organism that he considered the "missing link" between protists and animals.
Haeckel's case for an amoeboid ancestor started with embryology (Haeckel 1876(Haeckel , 1914(Haeckel , 1874. He noted that the egg cells of animals lack a flagellum but are often contractile.
Moreover, he observed that in sponges, the unfertilized eggs are bona fide crawling amoeboid cells (later confirmed by (Franzen 1988;Ereskovsky 2010)). After fertilization, the sponge zygote divided to give rise to a ball of non-ciliated cells (the morula) that only later acquired cilia and collectively formed an internal space (thus becoming a blastula). According to Haeckel, future feeding cavities then formed during gastrulation. Seen through a recapitulationist lens, these developmental facts told a compelling evolutionary story: animals had evolved from freeliving amoebae that had first formed balls of cells before acquiring ciliation, an internal cavity and then, eventually, evolving a gut (Haeckel 1874(Haeckel , 1914Fig. (Nichols, Dayel, and King 2009 For all of Haeckel's fame, his amoeboid theory never seems to have gained followers. The reasons for this are unclear, but his theory may have suffered from the rise of a worthy competitor: the flagellate theory of animal origins.   (Leadbeater 2015). (In the 20 th century, it would be discovered that collar cells are in fact widespread in the animal kingdom and not restricted to spongessee section V below).

III. Metchnikoff's hypothesis: choanoflagellates as ancestors
The flagellate hypothesis was easy to combine with Haeckel's Blastaea theory: one just had to replace Haeckel's amoeboid ancestor with a flagellate. This made the resulting theory

IV. Saville-Kent's polyphyletic hypothesis of animal origins: sponges from flagellates and bilaterians from ciliates
William Saville-Kent and Henry James-Clarktwo of the first choanoflagellate experts agreed with Metchnikoff on the evolution of sponges from choanoflagellate-like ancestors. But they disagreed (collegially) with Metchnikoff and (much more passionately) with Haeckel on the connection of sponges to animals (reviewed in (Leadbeater 2015)). This led Saville-Kent to conclude that animals had a dual origin: sponges had evolved from choanoflagellates, while all other animals had evolved from ciliates.
Haeckel initially thought of sponges as protists rather than animals (Haeckel 1876) but changed his mind after he discovered that they went through a gastrula stage (Haeckel 1872)an  (Leadbeater 2015)). Saville-Kent thought the strongest blow to Haeckel's views was the discovery of his own "missing link" (see Footnote 1)-a living species that he felt was the perfect  (Fig. 4, panel 20) found its way into textbooks and has been widely reproduced since (Buss 1987;Brusca and Brusca 2003), but he also produced many lesserknown illustrations covering the complete developmental trajectory of P. haeckelii, including a single collar cell serially dividing into 2, 4, and 8 cells, after which amoeboid cells started differentiating (Fig. 4).
Saville-Kent was a thorough and careful microscopist, with his meticulous sketches anticipating structures that have been consistently detected and verified using modern techniques in microscopy. It is therefore unlikely that he would have simply misunderstood or misobserved an isolated specimen (such as a sponge larva). Instead, his description of the life history of P.
haeckelii implies a detailed and extensive familiarity with multiple specimens, followed over an extended period of time. He was also a generally reliable observer, and his descriptions of other protists have been largely confirmed. Even though choanoflagellates have recently been shown to switch to an amoeboid form under confinement (Brunet et al. 2020), it is unlikely that Saville-Kent would have accidentally confined his samples: indeed, in the same book in which he described P. haeckelii (Kent 1882), he reported the retraction of the choanoflagellate collar complex under confinement and its regeneration after confinement release.
If an honest mistake is ruled out, then P. haeckelii might have been realand close to Saville-Kent's description. However, efforts to re-isolate P. haeckelii from the source location in Kew Gardens by one of us (T.B., together with Barry Leadbeater) have failed so far 5 . Given the personal rivalry between Saville-Kent and Haeckel, an alternative interpretation is that the description of P. haeckelii by Saville-Kent was either partly or entirely fabricated, possibly to get back at Haeckel. As with Magosphaera, the existence of P. haeckelii remains a mystery.
As significant as Saville-Kent thought P. haeckelii was, he only considered it relevant to the origin of sponges, but not of other animals. Instead, he proposed that (most) animals had evolved from ciliatesnot just once, but many times, with different ciliates giving rise to different animal lineages (Kent 1882). Saville-Kent was struck by the similarity in size, shape, and behavior between ciliates and small animals (both meiofaunal specieslike rotifers or flatwormsand planktonic larvae; Fig. 5A). His idea initially drew skepticism (Lankester 1883) but had a few early supporters (Sedgwick 1895). It would, however, make a spectacular comeback and then recede again in the 20 th century.

V. 20 th century: the rise and fall of the ciliate theory
Saville-Kent was correct on one point: the similarities between ciliates and small animals of the interstitial fauna are striking (reviewed in (Rundell and Leander 2010;Leander 2008)). At first sight, one could easily mistake Paramecium for an acoel worm (Fig. 5B,C). Both are elongated, bilaterally symmetrical, nearly half a millimeter long, and densely covered in motile cilia. Acoel worms are minute animals of extreme simplicity (long believed to be flatworms, but now known to belong to a separate bilaterian lineage (Marlétaz 2019)). They lack excretory organs, an anus, and even a proper gut. Early histological studies emphasized that simplicity, and many observers went so far as to erroneously conclude that acoels lacked separate cells (except perhaps in the epidermis) and instead represented a single large syncytium containing floating nuclei. Uncertainty around this point persisted from the 1880s to the 1960s, when electron microscopy finally demonstrated that acoels were actually fully cellular (see (Delage 1886;Pedersen 1964) for reviews).
In the meantime, however, the supposedly syncytial organization of acoels, together with their overall similarity to ciliates prompted a revival of Saville-Kent's ciliate theory of 5 As a caveat, the reisolation of even a well-studied choanoflagellate species can be challenging. For example, the laboratory model species Salpingoeca rosetta (Dayel et al. 2011) has been isolated only once and we have been unable to re-isolate it from its source location despite repeated attempts. bilaterian origins. The idea was proposed independently by Jovan Hadži  and Otto Steinböck (1893Steinböck ( -1969, and further elaborated by Earl D. Hanson (1927Hanson ( -1993 (Hadzi 1953(Hadzi , 1963Steinböck 1963;Hanson 1977Hanson , 1963. These authors identified many purported homologies among ciliates and acoels: ciliary arrays of the former were homologized to the ciliated epidermis of the latter; the contractile infraciliary lattice of ciliates was inferred to represent an antecedent of acoel musculature; the digestive vacuoles were proposed to be equivalent to the acoel digestive mass; and pulsatile vacuoles in ciliates were considered homologous to nephridia (excretory organs that are absent in acoels but found in flatworms). The fact that ciliates only have two nuclei (a micronucleus and a macronucleus) and do not display a multicellular or even syncytial organization was countered by pointing to Opalina, a protist then considered to be a ciliate which possessed many nuclei underneath its cell membrane (and which is now known to be a heterokont that only convergently resembles ciliates (Cavalier-Smith and Chao 2006)). Like Saville-Kent, supporters of the ciliate theory explained the similarity between choanoflagellates and choanocytes by hypothesizing that sponges were specialized choanoflagellates, and thus unrelated to other animals. Animals were thus assumed to have had at least two independent origins in the protistan world, and maybe even three (with cnidarians possibly descending from amoebae (Hanson 1977)).
The hypothesis of the syncytial nature of acoels was finally disproven by electron microscopy in the mid-1960s (Pedersen 1964), but the ciliate theory of animal origins had by then taken a life of its own and survived the loss of its former central argument (Hanson 1977).
As late as the 1980s-1990s, the ciliate hypothesis and the polyphyletic origin of animals were still often presented as the likeliest hypotheses of animal origins in popular texts and textbooks.
In his best-seller Wonderful Life, Stephen Jay Gould wrote: "The vernacular term animal itself probably denotes a polyphyletic group, since sponges (almost surely), and probably corals and their allies as well, arose separately from unicellular ancestorswhile all other animals of our ordinary definition belong to a third distinct group." (Gould 1989). Similar statements could be found in many contemporary zoology textbooks (Mitchell, Mulmor, and Dolphin 1988;P. Willmer 1990;Miller and Harley 1999), although a few were critical (Brusca and Brusca 1990).
Surprisingly, the ciliate theory survived the first molecular phylogenies as well: early studies included only a few genes analyzed with simple, similarity-based algorithms and often failed to recover the monophyly of the animal kingdom, thus apparently lending credence to multiple independent origins of animals from several protist groups (Field et al. 1988;Christen et al. 1991;Lake 1990). It was only with larger datasets and better models of sequence evolution that a consistent picture of monophyletic animals closely related to choanoflagellates finally emerged, with ciliates relegated to a very distant branch (Wainright et al. 1993), making the ciliate theory untenable. Unsurprisingly, the hypothesized homologies also eventually failed to withstand molecular scrutiny. For example, the infraciliary contractile lattice of Paramecium was found to be made of centrins, a family of contractile proteins unrelated to actin and myosin, the contractile proteins of animal musculature (Levy et al. 1996).
With the benefit of hindsight, many of the arguments underlying the ciliate theory appear contrived. Yet, it convinced manyif not mostexperts for nearly 30 years. We now know that its proponents were misled by an impressive suite of morphological convergences between metazoans, ciliates, and additional protists like Opalina. While the ciliate theory has now been dismissed as inconsistent with the modern eukaryotic phylogeny, it serves as a reminder of how much complexityin morphology, patterning, and behaviorcan be achieved by a single cell (Marshall 2020). The animal-like behaviors of ciliates, which fascinated scientists and philosophers at the turn of the 20 th century (Schloegel and Schmidgen 2002), are currently undergoing a renaissance as a research topic (Coyle et al. 2019;Mathijssen et al. 2019; Dexter, Prabakaran, and Gunawardena 2019; Wan and Jékely 2020), as are the mechanisms of their patterning and morphogenesis (Marshall 2020). Properly understood as an independent and unique evolutionary experiment in achieving levels of size and morphological complexity that rival those of small animals, ciliates remain as fascinating as ever.

VI. 20 th century: the collared flagellate/Choanoblastaea model
Although it had to compete with the ciliate theory for part of the 20 th century, Metchnikoff's concept of a choanoflagellate-like ancestor for all animalsand not just for spongeswas continuously supported by some authors (Hyman 1940;Rieger 1976;Salvini-Plawen 1978;Nielsen and Norrevang 1985). These researchers were each convinced about the monophyly of animals based on shared features such as sperm and eggs, epithelia, and gastrulation. This implied that all animals had evolved from a single lineage of protist, of which choanoflagellates were considered the most plausible living representative as their similarity to choanocytes was so strong. This view received further support from the discovery of choanocytelike collar cells by electron microscopy in diverse animal phyla other than sponges (Nerrevang and Wingstrand 1970;Rieger 1976;Lyons 1973;Brunet and King 2017). Claus Nielsen named this revised Blastaea modelstarting from a collared ancestor -the "Choanoblastaea" (Nielsen 2008) (Fig. 6).
While early molecular studies initially contradicted the Choanoblastaea hypothesis and suggested animal polyphyly (see section IV above), improved analyses with more data and better statistical models of sequence evolution ended up consistently supporting the monophyly of animals and their sister-group relationship to choanoflagellates (Wainright et al. 1993;N. King and Carroll 2001;Nicole King, Hittinger, and Carroll 2003;Rokas et al. 2003;Lang et al. 2002;Ruiz-Trillo et al. 2008;Nicole King et al. 2008). Unlike hypothesized homologies between ciliates and animals, the inferred homology of the collar complex in animals and choanoflagellates survived molecular and biochemical analyses, which confirmed that the collar is composed of homologous cytoskeletal filaments in both choanoflagellates, sponges, and other animals (reviewed in (Leadbeater 2015;Brunet and King 2017)). The hypothesis of the homology of the collar complexproposed on morphological grounds in the 19 th centurythus appears to have been predictive (Colgren and Nichols 2020) and is now accepted by most authors (but see ( Despite its support from the data, the Choanoblastaea model leaves some questions unresolved. One is the similarity of crawling amoeboid cells, widespread in animals, to the  Fig. 4). Note that cell division is now restricted to those inner cells.
amoeboid motility of diverse protists. While some authors explicitly ascribed that similarity to evolutionary convergence (Cavalier-Smith 2017), few directly recognized or addressed the issue.
While one solution could have been to revive Haeckel's amoeboid hypothesis, a strict interpretation of his hypothesis had clearly become incompatible with structural information that had emerged in the 20 th century showing the homology of flagella in animals and diverse protists (reviewed in (Margulis 1981)). Instead, one parsimonious way to account for all the data has been to reconstruct the progenitor of animals as a shape-shifter: sometimes flagellate, sometimes amoeba, and maybe more.

VII. 20 th century: the amoeboflagellate model and the synzoospore model
Complex life cycles in protists have been known since the 19 th century. In 1898, the British medical doctor Ronald Ross (1857-1932) described the different life stages of the unicellular parasite that causes malaria, Plasmodium falciparum (reviewed in (Cox 2002)).A year later, the Austrian biologist Franz Schardinger (1853-1920 discovered Naegleria gruberi (then named Amoeba gruberi), a free-living amoeba that had the unusual ability to transdifferentiate into a flagellate form (Schardinger 1899).
The transition between the amoeboid and the flagellate forms of Naegleria is reminiscent of the reversible transdifferentiation between the flagellated choanocytes and the amoeboid archeocytes of sponges (Fig. 7) that was already known to Saville-Kent (Kent 1882) and later confirmed by modern studies (Nakanishi, Sogabe, and Degnan 2014;Sogabe et al. 2019)).
In spite of this parallel, shape-shifting protists such as Naegleria were apparently never considered relevant to animal origins before the mid-20 th century, when the Soviet biologist Alexey Zakhvatkin (1906-1950 and the British biologist E. N. Willmer (1902Willmer ( -2001  6 Which he referred to as Vahlkampfia gruberi. 7 P. citri is a secondarily non-photosynthetic, parasitic green alga, and thus belongs to a lineage whose sequenced representatives have lost regulators of cell crawling such as SCAR/WAVE (Fritz-Laylin, Lord, and Mullins 2017) and myosin II (Sebé-Pedrós et al. 2014). While the genome of P. citri itself has not been sequenced, it is interesting to wonder how amoeboid mobility could function in this species if it also lacks those genes. N. Willmer 1971). While he did not believe that Naegleria was directly related to animals, he thought it gave an idea of what animal ancestors might have looked like.
Zakhvatkin's and Willmer's ideas seem to have gone mostly unnoticed in their time, and debates regarding animal origins remained dominated by the ciliate theory and the flagellate theory. It is only in the last decadethe 2010sthat the concept of a protist ancestor with a complex life history has undergone a revival.

VIII. 21 st century: how complex was the metazoan precursor?
In 2009, Zahkvatkin's idea were shared with a broader audience thanks to a review paper which presented his theory in English and named it the "temporal-to-spatial transition" model of animal origins (Mikhailov et al. 2009). Nearly at the same time, molecular phylogenies revealed that the previously enigmatic filasterans and ichthyosporeans (Ruiz-Trillo et al. 2008) are the closest known living relatives of choanozoans (the clade formed by choanoflagellates and animals; Fig. 2). Together, choanozoans, ichthyosporeans and filasterans form the clade Holozoa 8 . Interestingly, single-celled holozoans assume diverse cellular forms (including flagellates, amoebae, and cystic forms), and many of them have complex life histories with multiple phenotypes (as do choanoflagellates, which have sessile, swimming and colonial flagellate forms, and often spores as well (Leadbeater 2015)).
Finally, choanoflagellates themselves turned out to be able to reversibly switch to an amoeboid phenotype in response to spatial confinement (Brunet et al. 2020), thus reviving Saville-Kent's concept of amoeboid phenotypes in choanoflagellates. Overall, these data converged to suggest Interestingly, comparative genomics has revealed that many genes thought to be animalspecific are present in their single-celled relatives -but often with a patchy and mosaic distribution, indicating rampant gene loss in most lineages (Richter et al. 2018;).
This suggests that the last choanozoan common ancestor possessed a mosaic of features that is not fully realized in any of its living relatives or descendants. We think this lends credibility to the possibility of a "maximalistic ancestor." Future work will help to refine the "checklist" of ( Figure 8, continued.) cells or egg cells). The maximalistic ancestor (bottom) displays several forms of facultative multicellularity and combines several additional phenotypes known in single-celled relatives of animals, all of which have parallels among animal cell types and represent hypothetical evolutionary precursors of the latter. Spherical multicellular colonies of flagellates, similar to those of some choanoflagellates (Dayel et al. 2011), resemble the Morula stage of animal embryos. Sessile flagellated cells adhere to the substrate by a combination of filopodia and secreted extracellular matrix (ECM, green), as in modern choanoflagellates (Dayel et al. 2011) and in the filasteran Capsaspora (Parra-Acero et al. 2018. This might have prefigured the adhesion of animal epithelial cells to the basal lamina. Amoeboid cells are proposed to undergo aggregative multicellularity, similar to Capsaspora (Sebé-Pedrós et al. 2013) and to dissociated sponge cells (Dunham et al. 1983). Finally, cysts are proposed to undergo hypertrophy by nuclear proliferation without cytokinesis, resulting in a syncytium that can cleave at constant volume to revert to a uninucleated state, as in modern ichthyosporeans (Suga and Ruiz-Trillo 2013;Dudin et al. 2019) and chytrid fungi (Medina et al. 2020). This process could have been the evolutionary precursor to the cleavage of animal zygotes. (Drawing: Debbie Maizels) ancestral choanozoan features -which will not necessarily include all those we depicted in Figure 8, nor will necessarily be restricted to them.

Conclusion
The past has only left incomplete traces, and our understanding of it is inevitably simplified. There is, however, another force that often pushes us to simplification: the urge to summarize history as a linear narrative that leads to the present. In this review, we have strived to embrace the complexity of the past -both of our scientific predecessors, and of our evolutionary ancestors. We hope the winding history of our field is worth appreciating for itself and for the many small gems it contains, before trying to extract an -inevitably simplifiedglobal message.
Nonetheless, a few general themes emerge. The diversity of historical hypotheses simultaneously reflects the complexity of the problem itself, the limited information available at the time, and the personal assumptions and preferences of their authors. On the one hand, morphological data were clearly confounded by multiple events of evolutionary convergence (such as between ciliates and animals), parallelism, and rampant loss. Solving the problem from morphology only was genuinely challenging (even after the advent of electron microscopy), and involved some degree of subjective judgment. On the other hand, many authors seemed to have made the task unnecessarily more difficult by assuming that the last single-celled ancestors of animals necessarily had an exact equivalent within living protists -while this ancestor likely had its own, unique combination of features that is not necessarily represented today. This point has become increasingly salient in the past few years, and we expect it to remain central to future research. Consistently, several species of single-celled holozoans with novel phenotypes have been newly described in the past few years (Tikhonenkov, Mikhailov, et al. 2020;Hehenberger et al. 2017;Brunet et al. 2019), and metagenomic surveys have provided evidence for the existence of additional undiscovered holozoan lineages (Arroyo et al. 2020;. Further exploration of single-celled biodiversity thus holds the promise to enrich our reconstitution of animal ancestors -and eventually maybe even to clarify the mysteries of Magosphaera planula and Proterospongia haeckelii. Another point of interest is the way in which past controversies were resolved. Many debates could only be settled after the invention of new techniques; yet, technical innovations alone were rarely sufficient. The first molecular phylogenies, for example, were rather inaccurate. Consensus was only reached after commonly accepted standards of evidence were agreed upon, and once multiple independent, technically solid studies converged toward the same answer. At a time where a new wealth of molecular data (notably from single-cell techniques) is promising to bring an unprecedented quantity of evidence to bear on the study the evolution of cell phenotypes, we hope that our historical summary can be read both as a cautionary tale, and as a reason for optimism.