Downloaded from http://cshperspectives.cshlp.org/ on October 2, 2021 - Published by Cold Spring Harbor Laboratory Press

How Natural a Kind Is “?”

W. Ford Doolittle

Department of Biochemistry and Molecular Biology, Dalhousie University, Halifax, Nova Scotia B3H 4R2, Canada Correspondence: [email protected]

Systematics balances uneasily between realism and nominalism, uncommitted as to whether biological taxa are discoveries or inventions. If the former, they might be taken as natural kinds. I briefly review some philosophers’ concepts of natural kinds and then argue that several of these apply well enough to “eukaryote.” Although there are some sticky issues around genomic chimerism and when first appeared, if we allow for degrees in the naturalness of kinds, existing eukaryotes rank highly, higher than . Most bio- logists feel this intuitively: All I attempt to do here is provide some conceptual justification.

ystematics biologists today often appear that represents nature as it is, and reflects the Sequivocal or confused about the reality of network of causal laws” (Hacking 1991, p. 111, taxa in general, about whether they are inven- emphasis mine). tions (constructs of the human mind) or dis- Natural kinds, the recognizable cuts that are coveries (actually out there in the world). At supposed to be arrayed on our plate when na- the bottom and top of the Linnaean hierarchy, ture is “carved at the joints,” have been variously however, many biologists do see taxonomizing conceived (or declared nonexistent) by philos- as a reality-driven discovery process. A group ophers starting with the ancient Greeks (Hack- of them recently publicly endorsed the view ing 1991; Griffiths 1999; Campbell et al. 2011; that—even without any agreed on definition Bird and Tobin 2012). Biologists cannot glibly of “species”—one can calculate how many propose to name and count kinds of organisms such entities there are on Earth and in the ocean at any rank without committing themselves to (8.7 + 1.3 million) (Mora et al. 2011). And, some particular view as to what conception of more directly my subject here, the debate about “natural kind” they have in mind. My goal whether living things are of two fundamental here is to provide my own admittedly incom- kinds (prokaryotes and eukaryotes) or three plete catalog of some kinds of kinds that might (bacteria, , and eukarya (Wheelis et al. be appropriate for systematists, and then to 1992) is conducted as if there were some fact show how—in the case of “eukaryote”—we of this matter, ascertainable by analyzing the might use a mix of concepts, depending on data (Pace 2006). It is as if we thought, to quote our purposes. I will argue that the naturalness Ian Hacking’s apt description, that “There is a of kinds come in degrees, and by most standards unique best in terms of natural kinds, “eukaryote” holds up well, although there are a

Editors: Patrick J. Keeling and Eugene V. Koonin Additional Perspectives on The Origin and Evolution of Eukaryotes available at www.cshperspectives.org Copyright # 2014 Cold Spring Harbor Laboratory Press; all rights reserved; doi: 10.1101/cshperspect.a015974 Cite this article as Cold Spring Harb Perspect Biol 2014;6:a015974

1 Downloaded from http://cshperspectives.cshlp.org/ on October 2, 2021 - Published by Cold Spring Harbor Laboratory Press

W.F. Doolittle

few sticky issues.The naturalnessof “” one cannot have a science of laws without them. is more problematic. Nevertheless, as I and Olga Griffiths (1999) provides an appealingly mini- Zhaxybayeva have argued elsewhere (Doolittle malist criterion based on this presumption, and Zhaxybayeva 2013), the term is not as value- which permits degrees of “naturalness.” less as Norman Pace (Pace 2006) would hold. A kind is (minimally) natural if is possible to make better than chance predictions about the properties of its instances. Surprisingly, this ut- NATURAL KINDS BASED ON SHARED terly minimal conception of a natural kind is not PROPERTIES toothless. It does not license the conclusion that Outside of biology, instances of a single natu- any way of classifying nature is as good as any ral kind are quite typically considered to share a other. Natural kinds are ways of classifying the world that correspond to some structure inher- unique trait or set of traits, common to all and ent in the subject matter being classified (Grif- only members of that kind. This essence (or fiths 1999, p. 216). essences) might be material or immaterial, as There is no “unique best taxonomy” with kinds in Platonic ideals. Paradigm cases include the of this sort, and it is possible that an individual chemical elements, such as gold, whose essence can belong to several such kinds. But, Griffiths is its atomic structure. Not only does this struc- explains: ture unambiguously identify instances of gold wherever and whenever they might be found, Even if different categories are valuable for dif- but many of gold’s other properties follow ferent purposes, it is still true that some are better from it. Gold found on Mars or in another uni- for a particular purpose than others and that some have no foreseeable use at all. Embodying verse would still be gold, and still be good for these ideas in the language of natural kinds links crowning human teeth. it to a broadly realist perspective in which the An alternative to such essentialism, requir- predictive and explanatory value of categories ing no single universally shared property but a is taken to be prima facie evidence that they cap- cluster of defining features, would be Richard ture part of the structure of the world (Griffiths Boyd’s homeostatic property cluster (HPC) the- 1999, p. 217). ory of natural kinds, seemingly quite appropri- After an accounting of historical natural kinds, ate to biology and the “species problem” in par- which most biologists after Darwin consider all ticular (Boyd 1999). HPC kinds share traits, but taxa minimally to be, I will present some argu- no single trait (no single property of the cluster) ments in favor of “eukaryote” as a natural kind is necessary in all cases. Most dogs have four for many “particular purposes.” legs and a tail, but three-legged tail-less individ- uals might still be dogs if they bark and chase NATURAL KINDS BASED ON SHARED cats. Although no single property is necessary HISTORY or sufficient to distinguish members of an HPC kind, an underlying “homeostatic causal Many have claimed that preDarwinian classifi- mechanism(s)” ensures cohesiveness, continui- cation was essentialistic, most famously Ernst ty, and, generally, considerable overlap in prop- Mayr (1982), who called such thinking typolog- erties between individuals. For dogs the homeo- ical, linked it to Platonism and deemed it anti- static mechanism(s) would be a shared gene thetical to the acceptance of evolution, which he pool, environmental constraints, and devel- argued entails “population thinking.” Similarly, opmental program. Although this may sound the philosopher David Hull (1965a,b) blamed suspiciously like an essence, the emphasis is “two thousand years of stasis,” which in his view on processes not properties, and causal mecha- still stultifies systematics, on Aristotelian essen- nisms and the HPC kinds they maintain can, tialism. Although there is considerable contem- like species, evolve. porary scholarly disagreement about just how Natural kinds are what laws of nature are enthusiastically wedded pre-Darwinian system- supposed to be about and it is often held that atists were to ancient Greek philosophy (Winsor

2 Cite this article as Cold Spring Harb Perspect Biol 2014;6:a015974 Downloaded from http://cshperspectives.cshlp.org/ on October 2, 2021 - Published by Cold Spring Harbor Laboratory Press

How Natural a Kind Is “Eukaryote?”

2006; McOuat 2009), there is no doubt that The ... students cannot characterize the theory of Origin of Species changed the game, whatever it evolution. Admittedly, some will give a (usually had been. Darwin wrote: erroneous) account of the theory of Natural Se- lection, which they will associate (correctly) with ...that the characters which naturalists consider the name of Charles Darwin and, much less of- as showing true affinity between any two or more ten, also that of Alfred Russel Walllace. Natural species, are those which have been inherited selection is one component of evolutionary the- from a common parent, all true classification ory as proposed by Darwin and Wallace, but the being genealogical;—that community of descent other, for which selection is merely a hypothesis is the hidden bond which naturalists have been of mechanism, is the theory that evolution has unconsciously seeking, and not some unknown occurred. But that theory must have been pro- plan of creation, or the enunciation of general posed to explain some body of data and/or lower propositions, and the mere putting together and level theories. What I wanted to hear from my separating objects more or less alike (Darwin students was what that corpus of knowledge was. 1859, p. 420). The answer, as we have seen, is that the theory of evolution states that the apparent relationships In Darwinian biology, it is only common de- of organisms in a systematic classification are real scent that comprises the essence uniting organ- relationships, because “relationship” in such a isms of the same kind. Essences do not inhere context is not a metaphor but is actually to be in individuals or taxa themselves but in the pat- ascribed to community of descent. Thus the the- ory of evolution was proposed by Darwin and tern of (cladistic) relationships between them, Wallace to explain the pattern of relationships and are historical rather than material or even in what we may now term “Natural Classifica- ideal (in Plato’s sense) in character. Any living tion” (Panchen 1992, p. 3). creature born of dogs would be a dog even Such historical or genealogical essentialism un- if it has none of the cluster of properties by fortunately tells us nothing about taxonomic which dogs might be recognized as an HPC rank, however, only relationships. Darwin rec- species. Conversely, any cat surgically modified ognized the arbitrariness of distinctions be- to show all those properties would still not be tween strains, species, and genera, whereas a dog. higher taxa were understood to be man-made However, because progeny without surgi- categories by Linnaeus and others even before cal intervention generally resemble parents, 1859 (Ereshefsky 1999). So, whereas shared phenetic classifications (based only on shared properties might identify Lepidoptera as a nat- properties) are often good proxies for cladistic ural kind sharing a common developmental branching relationships. In an important sense, program or some more abstract moth–butterfly Darwin’s first profound and revolutionary con- essence, historical essentialism remains uncom- tribution to biology was thus to legitimize phe- mitted about the boundaries and reality of such netic classification by providing a theoretical higher groups. basis for its frequent success. His theory liber- ates taxonomy from creationist essentialism THE SPECIAL CASES OF SPECIES (“some unknown plan of creation”) and ele- AND DOMAINS vates it above nominalism (“the mere putting together and separating objects more or less Mayr’s biological species concept (BSC) can be alike”) because a tree-like pattern of lineage seen as an attempt to overcome the ranking splittings underwrites the tree-like classification problem, for that taxonomic level (Mayr 1982). of “groups subordinate to groups.” Species are the most inclusive grouping of Panchen (1992) nicely emphasizes the cen- genealogically related individuals capable of trality of this linkage of phenetics and phylo- forming an interbreeding population, according genetics in his Classification, Evolution and to the BSC. At the level of gene genealogy this the Nature of Biology. He recounts his begin- would normally entail that phylogenetic trees ning students’ inability to answer the question for different genes sampled from within a spe- “What is the theory of evolution?” cies would be different, while trees for such

Cite this article as Cold Spring Harb Perspect Biol 2014;6:a015974 3 Downloaded from http://cshperspectives.cshlp.org/ on October 2, 2021 - Published by Cold Spring Harbor Laboratory Press

W.F. Doolittle

genes sampled from different species would be han 2002). So in this formulation we might take the same—a criterion that has been used to ex- individual taxa as real, but they are still not tend the BSC into prokaryotic systematics, property-based, gold-like, natural kinds. Mar- where “interbreeding” has quite a different tians, even if they looked exactly like us or our character, but recombination can be rampant dogs, would not be H. sapiens or C. familiaris (Dykhuizen and Green 1991). unless descended from human colonists and The “species” category or kind defined in their companion animals. this way seems to me a natural enough one, But, again, neither the BSC nor treating taxa certainly by Griffiths’ criteria and maybe even as individuals will help us to delimit taxa higher for essentialists, the essence being just that sort than species. The component members of high- of interbreeding population genetic behavior. er taxa do not interbreed nor can they be con- One might find species on Mars or in another strued as reproductively competing parts in the universe. But that does not mean that particular same way. Only phenetic criteria and some pre- species taxa (Canis familiaris or Homo sapiens, sumptions about their relative fundamentalness for instance) have distinct and permanent in- can be used to cut horizontally across the tree- dividual essences that make them distinct en- like pattern of descent relating species to delimit during natural kinds (like gold), as would be genera, families, orders and on up. It is widely consistent with “typological thinking.” Indeed, understood that such rankings are endlessly ar- it means that they do not: C. familiaris and guable and not comparable across groups, and H. sapiens are just particular instantiations or there is no causal story invoking a genus or individuals of the BSC natural kind, and we higher taxon that is not in fact a story about would not expect to find them on Mars. Nor some or all of its constituent species. Devitt, does it mean that all organisms, many of which using something like Griffiths’minimal concep- seldom if ever “interbreed,” must belong to a tion, concludes that “although the higher cate- species. So BSC-defined species cannot play a gories may have some practical value, they are reliably general role as the fundamental unit of doing no explanatory work: they are not natural systematics. kinds” (Devitt 2011, p. 167). Hull (1978) and Michael Ghiselin (1974) An exception could be made for the very have made much of the notion that particular highest taxa, but in a different way than in the species are, indeed, spatiotemporally bounded case of species. If the Tree of Life were an unre- individuals, like you or me. Ghiselin wrote: ticulated pattern of bifurcations all the way down to a single last common ancestor, then Traditionally, species (like other taxa) have been its deepest division (its earliest bifurcation) treated as classes (universals). In fact they may be would unambiguously separate all life into considered individuals (particular things). The logical term “individual” has been confused with two Darwinian (historical) natural kinds (“do- a biological synonym for “organism.” If species mains” if you will), an indisputable fact how- are individuals, then: (1) their names are proper, ever we might chose to rank the subsequent (2) there cannot be instances of them, (3) they do bifurcations along either of the two branches. not have defining properties (intensions), and (4) their constituent organisms are parts, not members. “Species” may be defined as the most THREE DOMAINS VERSUS TWO extensive units in the natural economy such that reproductive competition occurs among their This unavoidable logic notwithstanding, the parts (Ghiselin 1974, p. 536). 1977 publication by Woese and Fox argued for the view that Life comprises three basic types of Even asexual organisms outside the scope of the organisms—bacteria, archaea, and eukarya— BSC can form species such that “reproductive which they then called “urkingdoms.” Although competition occurs among the parts”—indeed at that time admitting the possibility that “One something like this has been used to define spe- of the three may represent a far earlier bifurca- cies for prokaryotes that do not recombine (Co- tion than the other two, making there in effect

4 Cite this article as Cold Spring Harb Perspect Biol 2014;6:a015974 Downloaded from http://cshperspectives.cshlp.org/ on October 2, 2021 - Published by Cold Spring Harbor Laboratory Press

How Natural a Kind Is “Eukaryote?”

only two urkingdoms” (Woese and Fox 1977, p. ganism must be one or the other can be traced, 5089, emphasis mine), these authors soon de- understood the dichotomy to be based on veloped the notion that the last common an- shared properties. Most famously, the former cestor was not a single cell or species, but a wrote in a magisterial textbook, that: population of entities quite different from any contemporary cell, called collectively “the pro- ... this basic divergence in cellular structure, genote” (Woese 1982). Thus Woese (1998) which separates the bacteria and blue-green algae from all other cellular organisms [eukaryotes], would write some 20 years later that: represents the greatest single evolutionary dis- continuity to be found in the present day world Organismal lineages, and so organisms as we (Stanier et al. 1963, p. 85). know them, did not exist at these early stages. The universal phylogenetic tree, therefore, is It was in fact mostly (and problematically) the not an organismal tree at its base but gradu- ally becomes one as its peripheral branchings shared lack by bacteria and blue-green algae emerge. The universal ancestor is not a discrete of the properties thought to be common to all entity. It is, rather, a diverse community of cells and every cell of the eukaryotic natural kind that survives and evolves as a biological unit. that defined prokaryotes for Stanier and van This communal ancestor has a physical history Niel, although now we might also include sev- but not a genealogical one. Over time, this an- eral more positive discriminators (Martin and cestor refined into a smaller number of increas- Koonin 2006). Whether Stanier and van Niel ingly complex cell types with the ancestors of the three primary groupings of organisms arising as understood the “evolutionary discontinuity” a result. to be phylogenetic as well as phenetic is unclear and unlikely, although this is what Pace (and This model of common ancestry without a Woese before him) charges them with. Most common ancestor seems to license the claim likely Stanier and van Niel actually thought, as for a tripartite living world, but does so by did most of their contemporaries, that more stepping outside of the Darwinian systematic complex cells of the eukaryote kind arose some- framework (“all true classification being genea- how from some particular lineage(s) of prokary- logical”). It is also strikingly inconsistent with otes, making the latter paraphyletic, but not the usual presentation of the universal tree by polyphyletic. The deepest phylogenetic gap Woese and colleagues, which does indeed dis- would thus have separated prokaryotes into two play bacteria as an “earlier branching than the prokaryotic clades, as indeed Pace’s three do- other two.” main tree actually shows (Pace 2006). Such ambivalence undermines Pace’s (2006) The conflation of phenetic, or property- arguments against the use of the term “pro- based, and phylogenetic, history-based, con- karyote.” On the one hand, Pace argues in an ceptions of natural kinds was made very clear influential Nature opinion piece, prokaryotes in a noisy debate between Woese (1998) and are paraphyletic (eukaryotes having arisen Mayr (1998) played out in the pages of the Pro- from within them) and thus not a legitimate ceedings of the National Academy of Sciences. taxon. On the other, he holds that eukaryotes The latter, while accepting the former’s three- did not arise from prokaryotes, because their domain tree in which archaea and eukarya are last common ancestor was neither, but rather sisters, nevertheless insisted that those “basic “the progenote.” divergences” in cell structure that separated eu- Much of what is at issue here is a conflict karyotes from either bacteria or archaea were between recognizing natural kinds based on many and important, whereas those separat- shared properties (phenetics) versus natural ing bacteria and archaea were (relatively) fewer kinds based on shared history (phylogenetics). and less consequential. So a “natural classifica- Without doubt, Stanier and van Niel, to whose tion” taking into account “degree of difference” 1962 paper the widespread use of “prokaryotes” as well as evolutionary branching pattern would and “eukaryotes” and the belief that any or- retain the prokaryote/eukaryote dichotomy, he

Cite this article as Cold Spring Harb Perspect Biol 2014;6:a015974 5 Downloaded from http://cshperspectives.cshlp.org/ on October 2, 2021 - Published by Cold Spring Harbor Laboratory Press

W.F. Doolittle

held. Mayr called such a classification “evolu- as that between prokaryotes and eukaryotes and tionary” or “Darwinian” because relationships rule for the three-domain perspective on such within it remain strictly genealogical (polyphy- strictly phenetic grounds. Although historical letic groups are not allowed); but naming and kinds are de rigueur in post-Darwinian practice, ranking of taxa may take phenetic similarity and there is no principle by which they can claim cohesion into account. exclusive representation of “nature as it is.” But Although scholars may disagree on this (Pa- jury bias would be difficult to eliminate; some dian 1999; Wilkins 2009), the following passage might insist that components of the translation from The Origin seems to indicate that Darwin machinery should receive extra weight in phe- would indeed concur with such an approach: netic analyses, some might favor cellular ultra- structure, and still others would argue that an I believe that the arrangement of the groups with- “all-traits-are-equal” approach is more fair- in each class, in due subordination and relation to each other, must be strictly genealogical in minded. Disputes in systematics are arguments order to be natural; but that the amount of dif- about what should be done masquerading as ference in the several branches or groups, though competing claims about what is. allied in the same degree in blood to their com- mon progenitor, may differ greatly, being due to the different degrees of modification which they THE EUKARYOTE KIND have undergone; and this is expressed by the forms being ranked under different genera, fam- Still, what was not in dispute between Mayr and ilies, sections, or orders (Darwin 1859, p. 369). Woeseand Pace was the monophyly and pheno- typic cohesion of extant eukaryotes, a natural Woese (1998), in vigorous rebuttal of Mayr kind on several counts. We should consider (1998), asserted that the latter gave too little these before discussing the problematics. attention to the degree of phenetic difference Woese’s writings often made him sound between bacteria and archaea. Moreover, al- like a born-again pre-Darwinian “typological though admitting that “a phylogenetic tree thinker” of the sort of which Mayr wished to based upon the molecular data shows that the rid the discipline. newly recognized domain, the archaea, is, if anything, more closely related to eukaryotes Twenty-first century biology will concern itself with the great “nonreductionist” 19th century than to the familiar bacteria” (Woese 1998, p. biological problems that molecular biology left 1104), Woesewould not accept the cladistic im- untouched. All of these problems are different plication of this finding, that there are “in effect aspects of one of the great problems in all of only two urkingdoms,” one bacterial and the science, namely, the nature of (complex) organi- other archaeal plus eukaryotic. zation. Evolution represents its dynamic, gener- Thus, this Battle of the Titans was about (1) ative aspect; morphology and morphogenesis degrees of “degrees of difference,” a subjective represent its emergent, material aspect. One can already see the problem of the evolution of matter, and (2) the practice of systematics, a cellular organization coming to the fore. And sociophilosophical one. Or put another way, because of both its pressing practical and its fun- it was about different kinds of natural kinds, damental nature, the problem of the basic struc- and which is best. One way of carving nature ture of the biosphere is doing so as well. ...My at the joints, based on properties (cell struc- own career is one of the links between biology’s ture), arguably gives us prokaryotes and eukary- reductionist molecular past and its holistic fu- otes. Another, historical/genealogical in focus, ture (Woese 2004, p. 176). yields bacteria and a second domain comprising Most certainly, the Woese school’s focus on ri- archaea plus eukarya. Neither unambiguously bosomal RNA and secondarily on other com- gives us three domains. ponents of the translational and transcriptional It may be that some unbiased jury could machinery as telling the true evolutionary story decide that the phenetic distinction between has an essentialist ring. The accepted position of bacteria and archaea is at least as “significant” any species on the Tree of Life is that dictated by

6 Cite this article as Cold Spring Harb Perspect Biol 2014;6:a015974 Downloaded from http://cshperspectives.cshlp.org/ on October 2, 2021 - Published by Cold Spring Harbor Laboratory Press

How Natural a Kind Is “Eukaryote?”

these components, even if the majority of its and systems plus a differentiated cytoskeleton to genome would put it elsewhere. For many mi- coordinate location and function. Heterochro- crobiologists, natural classification (to the ex- matin in some form could also support life-cycle / tent systematics concerns them) is rRNA phylo- and or environmental cue-dependent coor- dinate gene expression. LECA also supported genetics. meiosis. It seems unsurprising that Woese and early adapters of his methodology (myself included) Some extant lineages lack some of these fea- were for the most part trained as molecular tures, for instance, “fully functional mitochon- biologists, predisposed to see genes involved in dria,” but this is fully consonant with being an replication, transcription, and translation as an HPC kind. And the interactions of eukaryotic organism’s most fundamental and important. mobility, compartmentalization, and mem- It may well be that many of these genes are less brane trafficking systems seem sufficient to es- frequently transferred than others, but the rea- tablish a “homeostatic causal mechanism.” sons are more likely to be high rates of expres- Some investigators, by attributing to LECA sion and “connectivity” (Cohen et al. 2011; Park all those properties present in some substantial and Zhang 2012) than their fundamental roles. fraction of its surviving descendants, are forced And it is true that no species defined as eukary- to infer an ancestor with substantially more ca- otic by rRNA sequence has been considered pabilities than all or most of these descendants, prokaryotic on other grounds. so that the history of post-LECA eukaryotic lin- Something more like HPC natural kind sta- eages is one of specialization and differential tus has also been invoked for eukaryotes, at least loss. For example, Harish et al. (2013) describe since Stanier and van Niel’s 1962 manifesto and a non-Woesian three-domain tree (bacteria likely before Sapp (2009). Nowadays, properties and archaea being sisters) in which the last com- thought to characterize all or most eukaryotes mon ancestral eukaryote and prokaryote boast and no or only a few prokaryotes are numerous. more protein superfamilies (SFs) than their Indeed, one of the fruits of comparative geno- typical descendant lineages, and the “most re- mics is the discovery (or at least the commonly cent” (i.e., “last”) universal common ancestor held belief ) that many of the complex cellular (MRUCA or LUCA) carried even more still. structures of contemporary eukaryotes were These researchers elaborate a remarkably Veli- likely present in LECA (the last eukaryotic com- kovskian scenario around this notion. mon ancestor). Koumandou et al. (2013), for instance, recently conclude that: MRUCA, identified as the root of the modern crown of global phylogeny, features an unexpect- It is now clear that the LECAwas both a flagellate edly complex genome that defies characteriza- and capable of movement by actin-based pseu- tion as a primitive ur-cell such as the progenote dopodia and possessed a sophisticated cytoskel- of Woese. Thus, it seems inconceivable to us that eton, including large families of kinesin and dy- an ancestral proteome containing three fourths nein motors. It possessed a complex and likely of all SFs in the modern crown can be identified very flexible, metabolism and a fully functional with the proteome of the ur-ancestor, i.e. the first mitochondrion. Endomembrane compartments cell and the root of Earth’s first phylogenetic tree. would have been essentially indistinguishable Accordingly, we suggest that the modern crown from modern cells, and included the endoplas- is a re-diversified tree rooted in complex survi- mic reticulum, the Golgi complex, endosomes, vors of mass extinction events that occurred autophagosomes and others. ...The LECA was some time before the Cambrian radiation. ... also capable of both conventional endocytosis In other words, the modern crown is the most and phagocytosis. The nucleus was fully differ- recent in a series of two or more global radiations entiated with nuclear pore complexes and a so- that produced diverse biota from bottlenecked phisticated system for organization and regula- populations that arose in the wake of mass ex- tion of chromatin. A high energy burden is tinctions ...MRUCA[,] the seed of the modern clearly implied by this architecture and required crown in this interpretation[,] would be the sur- to construct and maintain these compartments vivor of a mass extinction that laid waste and

Cite this article as Cold Spring Harb Perspect Biol 2014;6:a015974 7 Downloaded from http://cshperspectives.cshlp.org/ on October 2, 2021 - Published by Cold Spring Harbor Laboratory Press

W.F. Doolittle

bottlenecked an earlier crown population. In this symbiosis affecting only part of the eukaryotes; scenario MRUCA is a unicellular or a multicel- see below) might. lular survivor of a global collapse of the environ- Thus, eukaryotes appear to be natural kinds ment due to extreme climate change (Harish on the basis of shared properties and shared et al. 2013, pp. 1602–1603). history. Moreover, by Griffiths’ looser criteria, The more prosaic alternative, in my view much that the “predictive and explanatory value of more consistent with the principle of uniformi- categories is taken to be prima facie evidence tarianism beloved of Darwin, is to recognize that they capture part of the structure of the that lateral gene transfer (LGT) allows descen- world,” eukaryotes pass the test. Even with the dant lineages of a single ancestral entity (cell most prokaryote-like eukaryotes of which I am or species) to harbor many genes not present aware, the microsporidia (highly derived fungi), in that ancestor. This ancestor (which need in the safest assumption for any unknown proper- principle have borne no genes directly ancestral ty would be that it is “eukaryotic” in character to any in circulation today) would not be rec- (Corradi and Selman 2013). Prokaryotes, if the ognized as unusual in the size or complexity of rooted three-domain tree is accepted and para- its genome, were it still alive. This scenario is phyly eschewed, are natural only in the first way uniformitarian in that it assumes that cells and (shared properties, or, worse, shared lack of eu- the evolutionary forces affecting them (in par- karyotic characteristics), and arguably designat- ticular, LGT) were not unlike what we observe ing a species as prokaryotic has weaker “predic- now, as far back in time as we can know. The tive and explanatory value,” although such pools of genes shared through LGT by prokary- value is hard to quantify or compare. otes and (at lower frequency) unicellular eu- karyotes, occasionally overlapping but generally STICKY ISSUES kept distinct by barriers to gene expression (e.g., incompatible transcriptional and translational I see two flies in the eukaryote-as-natural-kind signals and introns), would be important “ho- ointment. The first is genomic chimerism—not meostatic causal mechanisms.” that which some have argued gave rise to the This complication aside, there seems little very first cell we might want to consider eukary- expressed doubt in the literature that LECA otic (called the “FECA” by some) (Wilson and was a single species that would now be consid- Dawson 2011)—but that which was created by ered eukaryotic on phenetic grounds (as above) organellar endosymbiosis and EGT, as well as and few (although some, quite idiosyncratic, LGT from subsequently disappeared endosym- like Harish et al. 2013) suggestions that any spe- bionts and mere prokaryotic food. An early de- cies we now consider prokaryotic derives from cision by Woese and colleagues to consider nu- it. In particular, if we take possession of mito- clear-encoded 18S, not organellar 16S rRNA, as chondria with genomes reduced by loss and properly representing eukaryotes in the univer- EGT as a sure sign of eukaryoteness, and the sal Tree of Life was not at the time questioned. fact that all extant mitochondria and mitochon- Woese and Fox (1977) wrote: drion-derived organelles likely derive from If there had been an “engulfing species” in rela- a single last common mitochondrial ancestor tion to which all the other organisms were endo- whose genome was already reduced by 90% symbionts, then it seems likely that 18S rRNA (Gray et al. 1999), LECA was already a eukary- represents that species. This hypothetical group ote. That is to say, post-LECA eukaryotes are of organisms, in one sense the major ancestors of monophyletic and “holophyletic” (not para- eukaryotic cells, might appropriately be called phyletic) (see Ashlock 1974). That they might urkaryotes. Detailed study of anaerobic amoebae and the like, which seem not to contain mito- be initially chimeric (the product of some bac- chondria and in general are cytologically simpler terial–archaeal “fusion”) would not alter such a than customary examples of eukaryotes, might claim for monophyly, although separate infu- help to resolve this question (Woese and Fox sions of foreign genes (as in the plastid endo- 1977, p. 5089).

8 Cite this article as Cold Spring Harb Perspect Biol 2014;6:a015974 Downloaded from http://cshperspectives.cshlp.org/ on October 2, 2021 - Published by Cold Spring Harbor Laboratory Press

How Natural a Kind Is “Eukaryote?”

Assumed here are (1) that it is the host and not organellar origin; the point is that they are not the hosted whose history should be taken as archaeal. that of the eukaryotic lineage; (2) that “urkary- Cotton and McInerney (2010), while again otic” organisms that never had mitochondria finding that bacteria-similar genes outnumber but that we would want to call eukaryotes on archaea-similar genes by more than four to one other grounds (Cavalier-Smith’s 1983 archezoa) (in yeast), nevertheless note that the: once existed and might still be around to serve ... archaebacterium-derived genes are signifi- as support for this notion; and (3) implicitly cantly more likely to be essential to yeast viabil- but surely in the back of all our minds in the ity, are more highly expressed, and are signifi- late 1970s, that the genetic contribution of cantly more highly connected and more central mitochondria or plastids—that fraction of any in the yeast protein interaction network ... eukaryote’s genome comprising genes of organ- and thus argue that: ellar origin (either still organellar or now nucle- ... genes of archaebacterial origin are in some ar)—was small. We now know assumptions 2 senses more important to yeast metabolism than and 3 to be problematic if not clearly false, but genes of eubacterial origin. This importance re- taking 18S rRNA (and associated translational flects these genes’ origin as the ancestral nuclear machinery) as the proper trackers of eukaryote component of the eukaryotic genome (Cotton phylogeny remains unquestioned. and McInerney 2010, p. 17252). The apparent falseness of 2 was a surprise. This would no doubt be the common feeling, In a history many times recounted, Cavalier- and the accepted structure of the three-domain Smith’s notion that certain amitochondriate tree remains the most popular. My point is that anaerobic eukaryotes (archezoa) that fortu- whether or not we root eukaryotes with archaea itously branched low in relatively unsophisti- or in proteobacteria is a judgment call, inti- cated phylogenetic analyses were primitively mately tied up with prevailing theories about without organelles held great appeal (Cava- the origins of eukaryotes, and certain prejudices lier-Smith 1983). It was consonant with what about the importance of hosts over symbionts. was then the standard version of the endosym- It is not a discoverable fact! biont hypothesis, which imagined that phago- In any case, if mitochondria were already cytosis and the associated complexities of eu- established in LECA, where we root eukaryotes karyotic cell structure preceded the acquisition does not impinge on their claim to monophyly. of mitochondria. But now, all examined eu- What might impinge is the subsequent incur- karyotes prove to harbor mitochondria or sion into only part of the eukaryote clade, of “mitochondrion-related organelles,”sometimes plastids. An earlier paper from McInerney’s highly derived (e.g., lacking DNA) and often group (Pisani et al. 2007) concludes that: engaged in activities other than respiration (Gray 2012). We may never find any living “ar- ...there are three distinct phylogenetic signals in eukaryotic genomes. In order of strength, these chezoa,” and one could at least imagine con- link eukaryotes with the Cyanobacteria, the Pro- structing a tree including all known eukaryotes teobacteria, and the , an from various mitochondrion-derived gene se- archaebacterial (euryarchaeotes) group. These quences rather than nuclear genes. Such a tree signals correspond to distinct symbiotic partners might be rooted in the proteobacteria, not as involved in eukaryote evolution: plastids, mito- sister to archaea. Support for this approach chondria, and the elusive host lineage ...(Pisani could come from the fact that, contrary to as- et al. 2007, p. 1752). sumption 3 above, the majority of eukaryotic It would be surprising if a preponderance of genes that can be shown to have prokaryotic cyanobacterial genes characterized all eukary- homologs find these homologs in the bacteria, otes, even those with no plastids in their history not archaea, the presumed prokaryotic sisters (indeed, see Koonin 2010). But quite possibly (or ancestors) of the “host” lineage (e.g., Pisani a Tree of Life displaying eukaryotes according et al. 2007). Such genes are not necessarily of to the majority of their genes, not just those

Cite this article as Cold Spring Harb Perspect Biol 2014;6:a015974 9 Downloaded from http://cshperspectives.cshlp.org/ on October 2, 2021 - Published by Cold Spring Harbor Laboratory Press

W.F. Doolittle

deemed “most important,” would show them to cally with whichever lineage of genes we think be polyphyletic, some grouping with proteo- tracks that history. Extant eukaryotes may be a bacteria and some with cyanobacteria. Again, good enough natural kind (as I will conclude that we do not make trees in this way is a choice momentarily), but we do not really have a prin- based on theories about origins and notions of cipled way of deciding when the kind first ap- relative importance, contingencies of the histo- peared on Earth. To be sure, molecular and cel- ry of our discipline and not unarguable facts of lular paleontologists make various claims about nature. when eukaryotic cells or eukaryotic metabolites The second fly in the ointment is built into first appear in the fossil record, and those brave the notion of eukaryogenesis: that eukaryotes enough to put dates on gene trees also have such must have arisen from some thing or things opinions. But they are using different diagnos- that were not eukaryotes. There are at least tics and definitions of “eukaryoteness” between four ways of thinking about this, and they can which there can be no principled reconciliation. be variously parsed (O’Malley 2010; Forterre 2011; Koumandou et al. 2013). First, one might “EUKARYOTE” IS A QUITE GOOD NATURAL assume that the many eukaryotic peculiarities KIND, “PROKARYOTE” LESS SO arose slowly and sequentially, perhaps driven by the advantages of phagocytosis, so that most Existing eukaryotes nevertheless might be taken were in place before the acquisition of mito- as natural kinds of any of the several general chondria. Unfortunately, either by bad luck or sorts sketched at the beginning of this essay. If because of mass extinctions and bottlenecks, one must have essences, the translational ma- none of the intermediate stages have left sur- chinery seems to provide that. Indeed, it was the vivors (the hapless “archezoa”). Second, one clear distinction between eukaryotic and pro- might suppose that mitochondriawere acquired karyotic ribosomal RNAs that first convinced first (prokaryotes with endosymbiotic pro- biologists that the endosymbiont theory for karyotes are now known) (Husnik et al. 2013), organelle origins was correct. As “homeostatic and all the complexities of eukaryoteness were property clusters,” the collection of eukaryote- then rapidly developed secondary accommoda- specific molecular machinery that phyloge- tions to this great evolutionary advance (Lane nomic reconstruction assigns to LECA is exten- and Martin 2010). Third, although a bacterial/ sive and bespeaks a community of integrated archaeal syntrophy is, indeed, the basis of eu- cellular processes serving a limited range of en- karyogenesis, integration into a single cellular ergy metabolisms. As historical kinds they ap- entity/lineage was mutualistic, and not as imag- pear monophyletic, at least as their history ined in traditional endosymbiosis theory. The is now traced. Most importantly, the causal cou- “Hydrogen Hypothesis” is an example (Martin pling of phenetic similarity and phylogenetic and Mu¨ller 1998). Fourth, the eukaryotic line- relatedness that is foundational for Darwin’s age is actually the oldest. Thus, the fact that theory of evolution holds up well. That is, as many eukaryotic genes are very dissimilar to Panchen put it: “apparent relationships of or- homologous prokaryotic genes does not reflect ganisms in a systematic classification are real rapid evolution leading up to or following the relationships.”And, because of this, “eukaryote” development of phagocytosis and the acquisi- readily meets Griffiths’ criteria as a way of “clas- tion of mitochondria, but rather ancient diver- sifying the world that correspond[s] to some gence (before the bacterial/archaeal split) (e.g., structure inherent in the subject matter being Sogin 1991). classified.” Why eukaryogenesis is a problem is that Pace (2006) is right that “prokaryote” de- whichever of these four scenarios we adopt, scribes a less satisfactory natural kind. Even if there will be an uncoupling, more or less sud- considered monophyletic, prokaryotes are not den, between “eukaryote” as defined by shared holophyletic. If the translational machinery is properties and “eukaryote” as defined histori- taken as essence-defining, they fall apart: Ar-

10 Cite this article as Cold Spring Harb Perspect Biol 2014;6:a015974 Downloaded from http://cshperspectives.cshlp.org/ on October 2, 2021 - Published by Cold Spring Harbor Laboratory Press

How Natural a Kind Is “Eukaryote?”

chaea are indeed closer to eukaryotes. As ho- and science (topics in contemporary philosophy) (ed. meostatic property clusters, they would still be Campbell JK, et al.), pp. 155–173. MIT Press, Cam- bridge, MA. recognized more for the numerous eukaryote- Doolittle WF, Zhaxybayeva O. 2013. What is a prokaryote? specific features they all lack than the prokary- In The prokaryotes—Prokaryotic biology and symbiotic as- otic-specific features they all share. However, the sociations (ed. Rosenberg E, et al.), pp. 41–58. Springer, prokaryote/eukaryote distinction was drawn Berlin. Dykhuizen DE, Green L. 1991. Recombination in Escheri- with cell structure and likely genomic simplic- chia coli and the definition of biological species. J Bacter- ity, not phylogenetic relationships, in mind. The iol 173: 7257–7268. kind “prokaryote” remains of use in that regard Ereshefsky M. 1999. Species and the Linnaean hierarchy. In (Doolittle and Zhaxybayeva 2013). In many ar- Species: New interdisciplinary essays (ed. Wilson RA), pp. 285–306. MIT Press, Cambridge, MA. eas of cell and especially molecular biology, it is Forterre P. 2011. A new fusion hypothesis for the origin of still “possible to make better than chance pre- Eukarya: Better than previous ones, but probably also dictions about the properties of its instances,” wrong. Res Microbiol 162: 77–91. Griffiths’ minimal criterion. There is much to Ghiselin MT. 1974. A radical solution to the species prob- lem. Systematic Biol 23: 536–544. be gained in biology, I think, by allowing natu- Gray MW.2012. Mitochondrial evolution. Cold Spring Harb ralness of kinds to come in degrees. Perspect Biol 4: a011403. Gray MW, Burger G, Lang BF. 1999. Mitochondrial evolu- tion. Science 283: 1476–1481. ACKNOWLEDGMENTS Griffiths PE. 1999. Squaring the circle: Natural kinds with I thank Maureen O’Malley for critical com- historical essences. In Species: New interdisciplinary essays (ed. Wilson RA), pp. 208–228. MIT Press, Cambridge ments on this essay. MA. Hacking I. 1991. A tradition of natural kinds. Philos Stud 61: 109–126. REFERENCES Harish A, TunlidA, Kurland GC. 2013. Rooted phylogeny of the three superkingdoms. Biochimie 95: 1593–1604. Ashlock PD. 1974. The uses of cladistics. Ann Rev Ecol Sys- tematics 5: 81–99. Hull D. 1965a. The effect of essentialism on taxonomy: Two thousand years of stasis (I). Br J Philos Sci 15: 314–326. Bird A, Tobin E. 2012. Natural kinds. In The Stanford encyclopedia of philosophy, Winter 2012 ed. (ed. Zalta Hull D. 1965b. The effect of essentialism on taxonomy: Two EN). Stanford University Press, Stanford, CA. thousand years of stasis (II). Br J Philos Sci 16: 1–18. Boyd R. 1999. Homeostasis, species, and higher taxa. In Hull D. 1978. A matter of individuality. Philos Sci 45: 335– Species: New interdisciplinary essays (ed. Wilson RA), 360. pp. 141–185. MIT Press, Cambridge, MA. Husnik F, Nikoh N, Koga R, Ross L, Duncan RP, Fujie M, Campbell JK, O’Rourke M, Slater M. 2011. Carving nature Tanaka M, Satoh N, Bachtrog D, Wilson AC, et al. 2013. at its joints: Natural kinds in metaphysics and science (top- Horizontal gene transfer from diverse bacteria to an in- ics in contemporary philosophy). MIT Press, Cambridge, sect genome enables a tripartite nested mealybug sym- MA. biosis. Cell 153: 1567–1578. Cavalier-Smith T. 1983. A revised six-kingdom system of Koonin EV.2010. The origin and early evolution of eukary- life. Biol Rev Camb Philos Soc 73: 203–266. otes in the light of phylogenomics. Genome Biol 11: 209. Cohan FM. 2002. What are bacterial species? Annu Rev Mi- Koumandou VL, Wickstead B, Ginger M, van der Giezen M, crobiol 56: 457–487. Dacks JB, Fiekd MC. 2013. Molecular paleontology and complexity in the last eukaryotic common ancestor. Crit Cohen O, Gophna U, Pupko T. 2011. The complexity hy- Rev Biochem Mol Biol 48: 373–396. pothesis revisited: Connectivity rather than function constitutes a barrier to horizontal gene transfer. Mol Lane N, Martin W. 2010. The energetics of genome com- Biol Evol 28: 1481–1489. plexity. Nature 467: 929–934. Corradi N, Selman M. 2013. Latest progress in microsporo- Martin W, Koonin EV. 2006. A positive definition of pro- dian genome research. J Eukaryot Microbiol 60: 309–312. karyotes. Nature 442: 868. Cotton JA, McInerney JO. 2010. Eukaryotic genes of arch- Martin W,Mu¨ller M. 1998. The hydrogen hypothesis for the aebacterial origin are more important than the more nu- first eukaryote. Nature 392: 37–41. merous eubacterial genes, irrespective of function. Proc Mayr E. 1982. The growth of biological thought. Harvard Natl Acad Sci 107: 17252–17255. University Press, Cambridge, MA. Darwin CR. 1859. On the origin of species by means of natural Mayr E. 1998. Two empires or three? Proc Natl Acad Sci 95: selection, or the preservation of favoured races in the strug- 9720–9723. gle for life, 1st ed. J. Murray, London. McOuat G. 2009. The origins of “natural kinds”: Keeping Devitt M. 2011. Natural kinds and biological realisms. In “essentialism” at bay in the Age of Reform. Intellectual Carving Nature at its joints: Natural kinds in metaphysics Hist Rev 19: 211–230.

Cite this article as Cold Spring Harb Perspect Biol 2014;6:a015974 11 Downloaded from http://cshperspectives.cshlp.org/ on October 2, 2021 - Published by Cold Spring Harbor Laboratory Press

W.F. Doolittle

Mora C, Tittensor DP, Adl S, Simpson AG, Worm B. 2011 Stanier RY, Doudoroff M, Adelberg E. 1963. The microbial How many species are there on Earth and in the ocean? world, 2nd ed. Prentice-Hall, Englewood Cliffs, NJ. PLoS Biol 9: e1001127. Wheelis ML, Kandler O, Woese CR. 1992. On the nature of O’Malley MA. 2010. The first eukaryotic cell: An unfinished global classification. Proc Natl Acad Sci 89: 2930–2934. history of contestation. Stud Hist Philo Biol Biomed Sci Wilkins JS. 2009. Species: A history of the idea. University of 41: 212–224. California Press, Berkeley, CA. Pace NR. 2006. Time for a change. Nature 441: 289. Wilson KL, Dawson SC. 2011. Functional evolution of nu- Padian K. 1999. Charles Darwin’s view of classification in clear structure. J Cell Biol 195: 171–181. theory and practice. Syst Biol 48: 352–364. Winsor MP. 2006. Linnaeus’s biology was not essentialist. Panchen AL. 1992. Classification, evolution and the nature Ann Mo Bot Gard 93: 2–7. of biology. Cambridge University Press, Cambridge, UK. Woese CR. 1982. Archabacteria and cellular origins: An overview. In Archaebacteria: Proceedings of the 1st Inter- Park C, Zhang J. 2012. High expression hampers horizontal national Workshop on Archaebacteria (ed. Kandler O), gene transfer. Genome Biol Evol 4: 523–532. pp. 3–17. Gustav Fischer, Stuttgart, Germany. Pisani D, Cotton JA, McInerney JO. 2007. Supertrees disen- Woese C. 1998a. The universal ancestor. Proc Natl Acad Sci tangle the chimerical origin of eukaryotic genomes. Mol 95: 6854–6859. Biol Evol 24: 1752–1760. Woese CR. 1998b. Default taxonomy: Ernst Mayr’s view of Sapp J. 2009. The new foundations of evolution. Oxford the microbial world. Proc Natl Acad Sci 95: 11043–11046. University Press, Oxford. WoeseCR. 2004. A new biology for a new century. Microbiol Sogin ML. 1991. Early evolution and the origin of eukary- Mol Biol Rev 68: 173–186. otes. Curr Opin Genet Dev 1: 457–463. Woese CR, Fox GE. 1977. Phylogenetic structure of the Stanier RY, van Niel CB. 1962. The concept of a bacterium. prokaryotic domains: The primary kingdoms. Proc Natl Arch Microbiol 42: 17–35. Acad Sci 74: 5088–5090.

12 Cite this article as Cold Spring Harb Perspect Biol 2014;6:a015974 Downloaded from http://cshperspectives.cshlp.org/ on October 2, 2021 - Published by Cold Spring Harbor Laboratory Press

How Natural a Kind Is ''Eukaryote?''

W. Ford Doolittle

Cold Spring Harb Perspect Biol 2014; doi: 10.1101/cshperspect.a015974

Subject Collection The Origin and Evolution of Eukaryotes

The Persistent Contributions of RNA to Eukaryotic Origins: How and When Was the Eukaryotic Gen(om)e Architecture and Cellular Mitochondrion Acquired? Function Anthony M. Poole and Simonetta Gribaldo Jürgen Brosius Green Algae and the Origins of Multicellularity in Bacterial Influences on Animal Origins the Plant Kingdom Rosanna A. Alegado and Nicole King James G. Umen The Archaeal Legacy of Eukaryotes: A Missing Pieces of an Ancient Puzzle: Evolution of Phylogenomic Perspective the Eukaryotic Membrane-Trafficking System Lionel Guy, Jimmy H. Saw and Thijs J.G. Ettema Alexander Schlacht, Emily K. Herman, Mary J. Klute, et al. Origin and Evolution of the Self-Organizing The Neomuran Revolution and Phagotrophic Cytoskeleton in the Network of Eukaryotic Origin of Eukaryotes and Cilia in the Light of Organelles Intracellular Coevolution and a Revised Tree of Gáspár Jékely Life Thomas Cavalier-Smith On the Age of Eukaryotes: Evaluating Evidence Protein Targeting and Transport as a Necessary from Fossils and Molecular Clocks Consequence of Increased Cellular Complexity Laura Eme, Susan C. Sharpe, Matthew W. Brown, Maik S. Sommer and Enrico Schleiff et al. Origin of Spliceosomal Introns and Alternative How Natural a Kind Is ''Eukaryote?'' Splicing W. Ford Doolittle Manuel Irimia and Scott William Roy Protein and DNA Modifications: Evolutionary What Was the Real Contribution of Imprints of Bacterial Biochemical Diversification Endosymbionts to the Eukaryotic Nucleus? and Geochemistry on the Provenance of Insights from Photosynthetic Eukaryotes Eukaryotic Epigenetics David Moreira and Philippe Deschamps L. Aravind, A. Maxwell Burroughs, Dapeng Zhang, et al.

For additional articles in this collection, see http://cshperspectives.cshlp.org/cgi/collection/

Copyright © 2014 Cold Spring Harbor Laboratory Press; all rights reserved Downloaded from http://cshperspectives.cshlp.org/ on October 2, 2021 - Published by Cold Spring Harbor Laboratory Press

The Eukaryotic Tree of Life from a Global Bioenergetic Constraints on the Evolution of Phylogenomic Perspective Complex Life Fabien Burki Nick Lane

For additional articles in this collection, see http://cshperspectives.cshlp.org/cgi/collection/

Copyright © 2014 Cold Spring Harbor Laboratory Press; all rights reserved