The six-kingdom, two-empire classification of life. Three major lineage mergers (symbiogeneses involving cell enslavement after phagocytic engulfment) are shown as dashed lines; four additional mergers that transferred chloroplasts from green plants or chromists into different protist lineages to make novel kinds of algae (Cavalier-Smith 2007c) are omitted for clarity (figure 6). The ancestrally photosynthetic kingdoms (Bacteria, Plantae and Chromista) are in green, but in each many lineages have lost photosynthesis. Chloroplasts originated when a biciliate protozoan internally enslaved a cyanobacterium bounded by two membranes to become the first plant. Chloroplasts are in the cytosol in Plantae, but inside two extra membranes in most Chromista: the ex-plasma membrane of the enslaved red alga, plus an RER membrane. Photosynthetic chromists include brown seaweeds, diatoms, haptophytes and cryptomonads. To portray early evolution in more detail, one must expand the two ancestral kingdoms by subdividing them more finely, as in figures 3 and 4 for Bacteria and figures 4 and 6 for Protozoa. But showing such basal groups in a phylogenetic tree as a single paraphyletic taxon, as here, is perfectly permissible and better focuses on the major steps in progressive evolution that generated the kingdoms than would excessive subdivision into a forest of ancient branches.

Contrasts between paraphyletic (ancestral) and polyphyletic groups. (a) The special case used by Hennig (1974) to claim that there is no cladistic difference between them because both have the same common ancestor (A) and an identical ancestral branching pattern. (b) A more realistic case where the three black-circle taxa do not have the same last common ancestor as the white-circle group, but have a different last common ancestor (B) which also has a different phenotype (black square) from A and from themselves. Case (b) shows that Hennig's claim for cladistic equivalence between paraphyletic and polyphyletic taxa lacks generality and rested on a cunningly chosen exceptional example. A paraphyletic group includes its last common ancestor and a polyphyletic one does not, a key fact partially concealed by Hennig misleadingly putting the same-sized box around both groups; to have correctly represented paraphyly the lower box should have included A, as it does in (b), where the obvious monophyly (single origin) of the paraphyletic white-circle taxon is much clearer than in Hennig's tendentious figure. The figure on the right also more strongly makes the point that the difference between polyphyly and paraphyly lies in the shared defining character (white circle) of the paraphyletic group having had a single origin, whereas the shared defining character of the polyphyletic group had three separate origins, i.e. a strongly contrasting phylogenetic history. Moreover, in (b) taxa 1 and 2 evolved black circleness in parallel from separate but phenotypically similar white-circle ancestors, whereas taxon 3 evolved it convergently from a cladistically and phenotypically more distinct black-square ancestor. It should be obvious that classifying white-circle taxa together is phylogenetically sound, i.e. they have a shared white-circle history, whereas classifying the black-circle ones together is unsound—being strongly contradicted by the lack of shared black-circle history. Unlike (a), (b) is a proper phylogeny with all ancestors and phenotypes shown; ignoring ancestral phenotypes makes nonsense of phylogeny. Cladistic aversion to paraphyletic groups, and lumping of paraphyly and polyphyly as ‘non-monophyly’, are logically flawed and anti-evolutionary (see also Cavalier-Smith (1998) which explains that clades, grades and taxa are all useful but non-equivalent kinds of group and that all taxa need not be clades and all clades need not be taxa).

The four ages of life. The six geological eras (black capitals) are demarcated especially by their fossils, which are absent in the Hadean, extremely sparse and problematic in the Archaean, numerous after about 2.2 Gy but all microscopic in the Proterozoic, and of every size and abundant in the Phanerozoic. In recognizing four ages of life (lower case colour on the right), I group the Late Proterozoic and Phanerozoic eras as the age of eukaryotes, because the origin of eukaryotic and archaebacterial cells that immediately followed the neomuran revolution is much more fundamental than the origin of bilaterian animals (around 550 Myr ago; Martin et al. 2000) that arguably initiated the Cambrian explosion (approx. 535–525 Myr ago) at the base of the Phanerozoic. On this view, increased acritarch fossil complexity at the transition from mid- to late Proterozoic was directly caused by the origin of the eukaryote cell. The Archaean/Proterozoic boundary essentially corresponds with the origin of photosystem II and oxygenic photosynthesis, shortly before the divergence of cyanobacteria (which are holophyletic, ignoring their being chloroplast ancestors, and thus not directly ancestral to other photosynthetic glycobacteria; figure 4). The early to mid-Proterozoic boundary is the most difficult to connect to a specific biological innovation. It may correspond with the origin of the posibacterial cell by a massive thickening of the murein wall and consequent loss of the OM, which may have stimulated the colonization of primitive cyanobacteria-dominated soils by Posibacteria (Cavalier-Smith 2006a); identification of the most complex mid-Proterozoic fossils as fungi (Butterfield 2005) is not compelling (earlier suggestions of eukaryotic algae were even less convincing). Possibly they are pseudosporangia and hyphae of Actinobacteria (Cavalier-Smith 2006a). The first convincing eukaryotic fossils are Melanocyrillium testate amoebae (Porter & Knoll 2000), though I do not accept their overconfident assignment to extant protozoan phyla (Porter et al. 2003); more likely they are an extinct group of early eukaryotes (Cavalier-Smith 2009a). Except for the final Vendian Period, bearing arguably stem animal fossils not confidently assignable to extant phyla, the Neoproterozoic was an era of only protists (unicellular eukaryotes; prior to the origin of plastids, perhaps little over 600 Myr ago, probably mainly Eozoa (figure 6) and Amoebozoa) and bacteria; phagotrophs diversified and underwent symbiogenesis to make various eukaryotic algae.

The eukaryote evolutionary tree, showing the messiness of real phylogeny. Compared with figure 1, the ancestral kingdoms Protozoa (all taxa inside the orange box) and Plantae are expanded to show their deepest branches and the reticulation caused by symbiogenetic cell enslavement. Apusozoa are gliding zooflagellates (Apusomonadida, Planomonadida; Cavalier-Smith et al. 2008) deeply divergent from other main groups. The large red arrow indicates the enslavement of a phagocytosed red alga over 530 Myr ago by a biciliate protozoan to form the chimaeric common ancestor of kingdom Chromista. Previously, Alveolata (i.e. Ciliophora and Myzozoa) were treated as protozoa, but are now included within Chromista (Cavalier-Smith submitted b); Ciliophora and most Myzozoa (subphyla Dinozoa, Apicomplexa) have lost photosynthesis (though many heterotrophic Myzozoa retain colourless plastids for lipid synthesis). Likewise, Rhizaria (Cercozoa, Foraminifera, Radiozoa) and centrohelid Heliozoa, both formerly treated as Protozoa, appear to be major chromist lineages that independently lost the ancestral red algal chloroplast and are now placed within Chromista not Protozoa (Cavalier-Smith submitted b). One small lineage of dinoflagellates (Dinozoa) replaced its ancestral chloroplast symbiogenetically by another from an undigested eaten haptophyte chromist (Patron et al. 2006). Independently, another small dinoflagellate lineage replaced its plastid by one from a green alga (Viridiplantae; dashed green arrow 1). Green algal chloroplasts were similarly independently implanted into Cercozoa (to make chlorarachnean algae; arrow 2) and into Euglenozoa (to make euglenoid algae; arrow 3). Euglenozoa, a phylum of ancestrally gliding zooflagellates (euglenoids; kinetoplastids, e.g. Trypanosoma and Bodo; postgaardiids; and diplonemids), differ so greatly from all other eukaryotes, and retain primitive bacteria-like features of mitochondrial protein-targeting and nuclear DNA pre-replication implying that they are the earliest diverging eukaryotic branch (Cavalier-Smith submitted b). Excavates comprise three entirely heterotrophic phyla: the putatively ancestral largely aerobic phylum Loukozoa (jakobids, which retain the most bacteria-like mitochondrial DNA, and Malawimonas), the largely aerobic derived phylum Percolozoa, and the secondarily anaerobic phylum Metamonada (e.g. Giardia and Trichomonas) that converted its mitochondria into hydrogenosomes or mitosomes and lost their genomes. Similar anaerobic relics of mitochondria evolved independently in Fungi, Amoebozoa, Percolozoa, Euglenozoa and Chromista. Contrary to earlier ideas, there are no primitively amitochondrial or primitively non-ciliate eukaryotes; earliest eukaryotes were aerobic flagellates, some of which evolved pseudopodia and became amoeboflagellates or eventually just amoebae. Animals and fungi both evolved from the same protozoan phylum, Choanozoa, but from different subgroups, being sisters of choanoflagellates and nucleariids, respectively (Shalchian-Tabrizi et al. 2008; Cavalier-Smith 2009a). Corticates and Eozoa are grouped as ‘bikonts’; formerly, the root of the eukaryote tree was postulated to be between unikonts and bikonts, not between Euglenozoa and excavates as shown here and justified in detail elsewhere (Cavalier-Smith submitted b)—a reassessment needing extensive testing.

The tree of life, emphasizing major evolutionary changes in membrane topology and chemistry. The most basic distinction is between ancestral Negibacteria, with a cell envelope of two distinct lipid bilayer membranes, and derived unimembrana, with but one surface membrane. Negibacteria were ancestrally photosynthetic (green), while unimembrana were ancestrally heterotrophs. A photosystem duplication enabled oxygenic photosynthesis (approx. 2.5 Gy ago: Kopp et al. 2005; Kirschvink & Kopp 2008) roughly when the outer membrane (OM) dating from the first cell acquired novel impermeable lipopolysaccharide and transport machinery. The late date of the neomuran revolution involving 20 major novelties is based on morphological fossils of eukaryotes and the argument that archaebacteria cannot be substantially older than their eukaryote sisters (Cavalier-Smith 2006a,c). Eubacteria, characterized ancestrally by cell wall murein, is an ancestral paraphyletic group that I do not make a taxon because I rather subdivide bacteria into subkingdoms Negibacteria and Unibacteria (comprising the phyla Posibacteria and Archaebacteria; figure 6), as their differences in membrane topology are more fundamental and significant (and more rarely change) than wall chemistry. Neomura is an important named clade that I chose not to make a taxon to avoid conflict with the much more radical differences between bacteria and eukaryotes. This exemplifies the principle that taxonomists should (and generally do) choose points on the continuous phylogenetic tree of maximal phenotypic disparity for artificially cutting it into taxa—NOT points of greatest cladistic depth irrespective of phenotype. Taxa have an initial capital; grades and clades that are not taxa have lower-case initials. Previously, hydrocarbon biomarkers were misinterpreted to give much earlier dates for eukaryotes and cyanobacteria, but these are invalidated by isotopic proof of hydrocarbon mobility from much younger strata (Rasmussen et al. 2008). Justification for the topology of this tree and its being correctly rooted and thus historically correct is elsewhere (Cavalier-Smith 2006a,c; Valas & Bourne 2009). A widespread contrary view that the root is between eubacteria and neomura stems from protein paralogue trees with long-branch topological artifacts and ignoring palaeontological evidence that negibacteria are immensely older than eukaryotes. For simplicity, the fact that the nucleus (N) has a double envelope that is part of a pervasive endomembrane system is not shown. The ancestral eukaryote is shown with a single cilium and centriole, but both had probably doubled in number prior to the earliest divergence among extant eukaryotes (Cavalier-Smith submitted b).

The tree of life, emphasizing the deepest branches. Ancestral groups of figures 1 and 3 are subdivided. Protozoa are resolved into two subkingdoms highlighted in yellow: the basal Eozoa (i.e. Euglenozoa plus Excavata), ancestrally characterized by a rigid cell pellicle supported by microtubules and the absence of pseudopodia (Cavalier-Smith submitted b) and the derived Sarcomastigota, ancestrally amoeboflagellates—probably with pointed pseudopodia, which gave rise to animals and fungi. Posibacteria comprise two subphyla: Endobacteria (putatively holophyletic) and Actinobacteria, which are probably the ancestors of neomura, having phosphatidylinositol lipids and proteasomes that both played key roles in eukaryogenesis (Cavalier-Smith 2009b). Glycobacteria are split into six phyla: three holophyletic, three paraphyletic (Cyanobacteria being ancestors of chloroplasts and thus partially of all Plantae, Chromista and those euglenoid eozoan Protozoa that secondarily acquired a plastid from green plants (figure 6); Proteobacteria being ancestors of mitochondria and thus in part of all eukaryotes; ancestral to Posibacteria are Eurybacteria). Eurybacteria include Thermotogales, Aquificales (now; see Bousseau et al. 2008), Heliobacteria and endospore-forming heterotrophs; they are often unwisely lumped with Endobacteria as ‘Firmicutes’ merely because they group on sequence trees, despite being structurally negibacteria. Ancestrally photosynthetic groups are in green. The ancestral (paraphyletic) Eobacteria are split into two putatively holophyletic phyla: Chlorobacteria (often photosynthetic, i.e. non-sulphur green filamentous bacteria like Chloroflexus) and the heterotrophic Hadobacteria (e.g. Thermus, Deinococcus). Bacteria ancestrally lacked flagella; soon after eubacterial rotary flagella evolved, one lineage relocated them to the periplasmic space to become spirochaetes (thumbnail sketch). Many lineages lost flagella, e.g. most Sphingobacteria and ancestors of neomura: archaebacteria re-evolved flagella and eukaryotes cilia, both entirely unrelated to eubacterial flagella. The higher proportion of holophyletic groups in figure 4 than figure 1 or 3 is bought at the expense of losing simplicity that more strikingly portrays major body-plan differences within eukaryotes (figure 1) and prokaryotes (figure 3). The extra cladistic resolution at the base of figure 4 is important for some purposes but irrelevant to others. Figures 1, 3, 4 and 5 are different ways of acceptably summarizing distinct aspects of the single true historical tree (which is reticulated and has ancestors and is thus not a cladogram or sequence tree). For more details on the 10 bacterial phyla and their relationships see Cavalier-Smith (2002a, 2006c). Oxygenic photosynthesis can have evolved no later than where shown by the upper blue arrow, immediately before the divergence of Cyanobacteria, but one reasonable non-decisive argument favours a marginally earlier origin before Hadobacteria diverged (lower blue arrow, when phospholipids arose; Cavalier-Smith 2006c). A sound hierarchical classification with ranks can simply represent both the fundamental shared similarities within ancestral groups, such as Posibacteria, Eobacteria, Bacteria and Protozoa, and the profound differences between their major subgroups.