Multigene phylogeny and cell evolution of chromist infrakingdom Rhizaria: contrasting cell organisation of sister phyla Cercozoa and Retaria

Protoplasma, Apr 2018

Infrakingdom Rhizaria is one of four major subgroups with distinct cell body plans that comprise eukaryotic kingdom Chromista. Unlike other chromists, Rhizaria are mostly heterotrophic flagellates, amoebae or amoeboflagellates, commonly with reticulose (net-like) or filose (thread-like) feeding pseudopodia; uniquely for eukaryotes, cilia have proximal ciliary transition-zone hub-lattices. They comprise predominantly flagellate phylum Cercozoa and reticulopodial phylum Retaria, whose exact phylogenetic relationship has been uncertain. Given even less clear relationships amongst cercozoan classes, we sequenced partial transcriptomes of seven Cercozoa representing five classes and endomyxan retarian Filoreta marina to establish 187-gene multiprotein phylogenies. Ectoreta (retarian infraphyla Foraminifera, Radiozoa) branch within classical Cercozoa as sister to reticulose Endomyxa. This supports recent transfer of subphylum Endomyxa from Cercozoa to Retaria alongside subphylum Ectoreta which embraces classical retarians where capsules or tests subdivide cells into organelle-containing endoplasm and anastomosing pseudopodial net-like ectoplasm. Cercozoa are more homogeneously filose, often with filose pseudopodia and/or posterior ciliary gliding motility: zooflagellate Helkesimastix and amoeboid Guttulinopsis form a strongly supported clade, order Helkesida. Cercomonads are polyphyletic (Cercomonadida sister to glissomonads; Paracercomonadida deeper). Thecofilosea are a clade, whereas Imbricatea may not be; Sarcomonadea may be paraphyletic. Helkesea and Metromonadea are successively deeper outgroups within cercozoan subphylum Monadofilosa; subphylum Reticulofilosa (paraphyletic on site-heterogeneous trees) branches earliest, Granofilosea before Chlorarachnea. Our multiprotein trees confirm that Rhizaria are sisters of infrakingdom Halvaria (Alveolata, Heterokonta) within chromist subkingdom Harosa (= SAR); they further support holophyly of chromist subkingdom Hacrobia, and are consistent with holophyly of Chromista as sister of kingdom Plantae. Site-heterogeneous rDNA trees group Kraken with environmental DNA clade ‘eSarcomonad’, not Paracercomonadida. Ectoretan fossil dates evidence ultrarapid episodic stem sequence evolution. We discuss early rhizarian cell evolution and multigene tree coevolutionary patterns, gene-paralogue evidence for chromist monophyly, and integrate this with fossil evidence for the age of Rhizaria and eukaryote cells, and revise rhizarian classification.

A PDF file should load here. If you do not see its contents the file may be temporarily unavailable at the journal website or you do not have a PDF plug-in installed and enabled in your browser.

Alternatively, you can download the file locally and open with any standalone PDF reader:

Multigene phylogeny and cell evolution of chromist infrakingdom Rhizaria: contrasting cell organisation of sister phyla Cercozoa and Retaria

Multigene phylogeny and cell evolution of chromist infrakingdom Rhizaria: contrasting cell organisation of sister phyla Cercozoa and Retaria Thomas Cavalier-Smith 0 1 Ema E. Chao 0 1 Rhodri Lewis 0 1 Handling Editor: Peter Nick 0 Department of Zoology, University of Oxford , South Parks Road, Oxford OX1 3PS , UK 1 Thomas Cavalier-Smith Infrakingdom Rhizaria is one of four major subgroups with distinct cell body plans that comprise eukaryotic kingdom Chromista. Unlike other chromists, Rhizaria are mostly heterotrophic flagellates, amoebae or amoeboflagellates, commonly with reticulose (net-like) or filose (thread-like) feeding pseudopodia; uniquely for eukaryotes, cilia have proximal ciliary transition-zone hublattices. They comprise predominantly flagellate phylum Cercozoa and reticulopodial phylum Retaria, whose exact phylogenetic relationship has been uncertain. Given even less clear relationships amongst cercozoan classes, we sequenced partial transcriptomes of seven Cercozoa representing five classes and endomyxan retarian Filoreta marina to establish 187-gene multiprotein phylogenies. Ectoreta (retarian infraphyla Foraminifera, Radiozoa) branch within classical Cercozoa as sister to reticulose Endomyxa. This supports recent transfer of subphylum Endomyxa from Cercozoa to Retaria alongside subphylum Ectoreta which embraces classical retarians where capsules or tests subdivide cells into organelle-containing endoplasm and anastomosing pseudopodial net-like ectoplasm. Cercozoa are more homogeneously filose, often with filose pseudopodia and/or posterior ciliary gliding motility: zooflagellate Helkesimastix and amoeboid Guttulinopsis form a strongly supported clade, order Helkesida. Cercomonads are polyphyletic (Cercomonadida sister to glissomonads; Paracercomonadida deeper). Thecofilosea are a clade, whereas Imbricatea may not be; Sarcomonadea may be paraphyletic. Helkesea and Metromonadea are successively deeper outgroups within cercozoan subphylum Monadofilosa; subphylum Reticulofilosa (paraphyletic on site-heterogeneous trees) branches earliest, Granofilosea before Chlorarachnea. Our multiprotein trees confirm that Rhizaria are sisters of infrakingdom Halvaria (Alveolata, Heterokonta) within chromist subkingdom Harosa (= SAR); they further support holophyly of chromist subkingdom Hacrobia, and are consistent with holophyly of Chromista as sister of kingdom Plantae. Siteheterogeneous rDNA trees group Kraken with environmental DNA clade 'eSarcomonad', not Paracercomonadida. Ectoretan fossil dates evidence ultrarapid episodic stem sequence evolution. We discuss early rhizarian cell evolution and multigene tree coevolutionary patterns, gene-paralogue evidence for chromist monophyly, and integrate this with fossil evidence for the age of Rhizaria and eukaryote cells, and revise rhizarian classification. Cell evolution; Chromista; Harosa; Rhizarian phylogeny; Cercozoa; Retaria Introduction Eukaryotes are classified in five kingdoms: unicellular, largely phagotrophic Protozoa are ancestors of four biologically very distinct clades—osmotrophic Fungi, epithelial phagotrophic Animalia, photosynthetic Plantae and Chromista (CavalierSmith 1998a, 2018; Ruggiero et al. 2015) . Animalia and Fungi are all heterotrophs. Almost all Plantae are photosynthetic. Their arguably sister kingdom Chromista includes all chromophyte algae (e.g. diatoms, brown algae, dinoflagellates, haptophytes), as well as some major heterotrophic protist groups, notably ciliates, sporozoa, heterotrophic heterokonts, heliozoans, and the largely non-photosynthetic infrakingdom Rhizaria, the subject of this paper. The predatory lifestyle of Rhizaria and details of motility, cell ultrastructure, and body form differ radically from those familiar to cell biologists focusing on animal, fungal or plant cells. Rhizaria are characterised by (1) having an ultrastructurally unique ciliary transition zone [Cavalier-Smith and Karpov 2012b; earlier Cavalier-Smith et al. (2008, 2009) incorrectly thought the transitional proximal hub-lattice and distal hubspoke structures were restricted to classical Cercozoa] and (2) the widespread presence of reticulose (net-like) or filose (thread-like) feeding pseudopodia (Bass et al. 2009a) . It is argued that the ciliary transition zone hub-lattice, hub-spoke structure, and reticulose pseudopodia are three rhizarian synapomorphies (Cavalier-Smith and Karpov 2012b) , though it is sometimes wrongly said they lack well-defined synapomorphies (e.g. Krabberød et al. 2017) . Distinctiveness of rhizarian cell biology is emphasised by novel features of actin, myosin and tubulin evolution with unique protein paralogues absent in other eukaryotes (Krabberød et al. 2017). As delimited by Cavalier-Smith (2003b), Rhizaria comprise phyla Cercozoa (Cavalier-Smith 1998a, b) and Retaria (Foraminifera plus Radiozoa: Cavalier-Smith 1999) . A major re-evaluation of kingdom Chromista adjusted the boundary between Cercozoa and Retaria by transferring former cercozoan subphylum Endomyxa to Retaria and establishing a new retarian subphylum Ectoreta to embrace Radiozoa and Foraminifera (Cavalier-Smith 2018) ; in doing so, Cercozoa and Retaria became sister clades with contrasting cellular phenotypes. That transfer was made for three reasons: (1) it emphasised a primary divergence between often filose amoeboflagellates (Cercozoa) and non-flagellates with reticulose pseudopodia (Retaria); (2) 187-protein trees (Cavalier-Smith et al. 2015a) strongly showed Endomyxa plus Ectoreta as a clade (revised Retaria); and (3) it made Cercozoa phenotypically more uniform (ancestrally flagellates that move by ciliary gliding on surfaces rather than swimming, a property never seen in Retaria). More recent 255-protein trees also robustly show Endomyxa, Ectoreta, revised Retaria and revised Cercozoa as clades (Krabberød et al. 2017), and thus fully support the transfer of Endomyxa to Retaria and formal restriction of Cercozoa to what used to be called ‘core Cercozoa’ (Nikolaev et al. 2004; Pawlowski 2008) or Filosa (Cavalier-Smith 2003b, 2018) . Many earlier sequence trees now appear to have been misleading with respect to relationships of Endomyxa, Cercozoa and Ectoreta owing to insufficient gene and taxon sampling. Thus some had placed either Foraminifera alone or Ectoreta as a whole within Endomyxa (Burki et al. 2010, 2013; Sierra et al. 2013, 36 gene analysis; Roy et al. 2014, 27 protein analysis) , and other single- or multigene trees had shown Ectoreta and classical Cercozoa (i.e. including Endomyxa) as sister clades (e.g. Cavalier Smith et al. 2004; Burki et al. 2010; Sierra et al. 2013, 109gene analysis). However, improved rhizarian gene and taxon sampling for multiprotein trees and use of evolutionarily more realistic site-heterogeneous algorithms (Cavalier-Smith et al. 2015a; Krabberød et al. 2017) yielded a consensus that Endomyxa and revised Retaria and Cercozoa (CavalierSmith 2018) are all clades. Multiprotein trees robustly support monophyly of Rhizaria, always a strongly supported clade (Burki et al. 2007, 2008, 2009, 2010, 2012; Cavalier-Smith et al. 2014, 2015a; Krabberød et al. 2017) . Cercozoa are now recognised as ancestrally biciliate heterotrophic flagellates that typically glide on the posterior cilium and have a marked propensity to form filose pseudopodia (filopodia) for help in catching or ingesting prey (e.g. thaumatomonads); several times, some became filose amoebae by losing both cilia [often forming shell-like tests into which they can withdraw pseudopodia for protection, e.g. Euglyphida, Lecythium, Penardeugenia (Dumack et al. 2016a, 2017b; Heger et al. 2010; Wylezich et al. 2002) ] or more rarely became planktonic flagellates by giving up the benthic gliding lifestyle [e.g. heterotrophic biciliate Katabia (Karpov et al. 2003); photosynthetic uniciliate Bigelowiella]. Cercozoa now comprise two subphyla (Cavalier-Smith 2018) : early diverging Reticulofilosa (classes Chlorarachnea, Granofilosea, Skiomonadea) and later evolving Monadofilosa (classes Metromonadea, Helkesea, Sarcomonadea, Imbricatea, Thecofilosea). Unlike Retaria, reticulopodia are very rare in Cercozoa, apart from the strongly net-like granofilosean Reticulamoeba (Bass et al. 2012) , recently discovered scaly Kraken (Dumack et al. 2016b, 2017a) and atypical netforming meropodial Chlorarachnion. Filopodia predominate in five of the eight cercozoan classes, but are undetected in Metromonadea or Skiomonadea, and pseudopodia in amoeboid guttulinopsid Helkesea are mostly lobose (as they can be rarely in sarcomonads or Thecofilosea). Apart from the strongly supported deepest branching of the skiomonad Tremula on 18S rRNA trees (Howe et al. 2011a) , the basal branching order of Cercozoa has always been poorly resolved on single-gene trees (Cavalier-Smith and Chao 2003a; Howe et al. 2011a) ; demarcation between Sarcomonadea and Imbricatea is particularly unclear (Scoble and Cavalier-Smith 2014) . Retaria ancestrally were trophically not flagellates but large amoeboid cells, usually with reticulose pseudopods (reticulopodia) rather than filopodia; they form swimming biciliate or uniciliate stages (never gliding) only transiently for dispersal or sexual reproduction and display a strong tendency to cellular gigantism and evolution of mineral skeletons (Cavalier-Smith 2018) . These skeletons, notably in foraminifera (Payne et al. 2013) and polycystine radiolaria (Biard et al. 2015) , provide billions of fossils important for palaeontology and economic geology. Two endomyxan lineages became reduced parasites of photosynthetic chromists or plants (class Phytomyxea) or of animals (class Ascetosporea: much more diverse than once thought: Ward et al. 2018) , of economic significance respectively for agriculture and fisheries. Except for the planktonic axopodial swimmer Sticholonche, reticulose pseudopodia are universal in Ectoreta and were the basis for defining phylum Retaria (Cavalier-Smith 1999) , but since then turned out to be widespread also in free-living members of subphylum Endomyxa. Endomyxa were initially limited to the non-reticulose, parasitically simplified Ascetosporea and Phytomyxea (Cavalier-Smith 2002a) but now also include two phylogenetically distinct free-living classes which can be filose or reticulose—marine Gromiidea (Gromia, Filoreta) and the often soil-dwelling and eukaryovorous vampyrellid amoebae (Ruggiero et al. 2015). The position of Rhizaria within Chromista has also been debated, as Chromista are a huge mixture of morp h o l o g i c a l l y e x t r e m e l y d i v e r s e p h o t o s y n t h e t i c , phagotrophic and osmotrophic organisms, whose exceptional disparity in gross body form long delayed recognition of their phylogenetic unity (Cavalier-Smith 1981a, 1986, 1989, 2010a, 2013a, 2018; Cavalier-Smith et al. 2015a) . Chromista include (1) all organisms with chlorophyll-c containing chloroplasts of secondary red algal origin (i.e. chromophyte algae) that have a fundamentally different membrane topology and chloroplast proteintargeting machinery from Plantae (Cavalier-Smith 2000a, 2003a, 2013a, 2018) ; (2) all organisms with tubular ciliary hairs; (3) all protists with axopodia; and those with any combination of these characters that clearly distinguish them from the usually heterotrophic ancestral kingdom Protozoa. Cavalier-Smith et al. (2015a) argued that the first two characters arose once in the ancestral chromist, whereas axopodia evolved polyphyletically is several distinct lineages within Rhizaria, Heterokonta (= stramenopiles), and Haptista. A cytoskeletal synapomorphy for all four chromist groups has now been identified that arguably predisposes chromists to evolve axopodia and clarifies the origin of distinctive cercozoan ventral centriolar roots (Cavalier-Smith 2018) ; evolution of periplastid protein import was reassessed and arguments against monophyly of Chromista refuted (Cavalier-Smith 2018) . Some lineages that secondarily lost some or all uniquely chromist characters were historically confused with Protozoa, where many heterotrophic chromists were long wrongly classified. Rhizaria were recognised as a distinct clade (Cavalier-Smith 2002a, 2003b; Nikolaev et al. 2004) decades after kingdom Chromista was established (Cavalier-Smith 1981a, 1986) ; because no Rhizaria have chlorophyll-c containing chloroplasts or tubular ciliary hairs, the defining chromistan characters, Rhizaria were originally wrongly excluded from kingdom Chromista and left in Protozoa (Cavalier-Smith 1981a) . Burki et al. (2007) showed that multigene trees strongly group Rhizaria with Halvaria (Heterokonta and Alveolata) as a major eukaryote clade provisionally labelled SAR. Accordingly, Cavalier-Smith (2010a) formally grouped Rhizaria (as an infrakingdom) with Halvaria as new chromist subkingdom Harosa, which is universally accepted as a robust clade including Rhizaria. The other chromist subkingdom Hacrobia (comprising Haptista and Cryptista: Okamoto et al. 2009; Cavalier-Smith 2010a) remains somewhat controversial because some multigene trees split Hacrobia, group Haptista and the divergent cryptist Telonema with Harosa as a reduced chromist clade, and place other Cryptista with Plantae (Burki et al. 2013, 2016; Krabberød et al. 2017, though this analysis omitted the important hacrobian clade Corbihelia) . Some trees show both Hacrobia and Chromista as clades, with Chromista sister to Plantae and Hacrobia sister to Harosa (Burki et al. 2009; Roy et al. 2014) ; but others showing Hacrobia as a clade group them with Plantae not Harosa (Burki et al. 2012; Cavalier-Smith et al. 2014, 2015b) . Thus multigene trees may show Hacrobia and/or Chromista as holophyletic or paraphyletic. Cavalier-Smith et al. (2015a) clarified reasons for these apparent conflicts, some related to the fundamentally chimaeric nature of chromist cells and genomes (CavalierSmith 2013a), concluding that Hacrobia and Haptista (haptophytes, centrohelids) are probably both clades and that overall evidence best fits chromist holophyly also. That and refutation of other criticisms of chromist monophyly (Cavalier-Smith 2018) make it unnecessary to exclude Cryptista from Chromista, a possible revision formerly considered in the light of poorly resolved 18S rDNA trees (Cavalier-Smith et al. 1994; Cavalier-Smith 1995) that the 255-protein PhyloBayes (but not ML) tree of Krabberød et al. (2017) might be supposed to favour. A siteheterogeneous analysis using 478 highly conserved genes found maximum support for both Harosa and Hacrobia being monophyletic and for Chromista plus Plantae being a clade (superkingdom Corticata) but showed Hacrobia as sister to Plantae (Ren et al. 2016; contradicting Krabberød et al. 2017) , but their study undersampled Hacrobia, omitting Corbihelia studied by Cavalier-Smith et al. (2015a). Halvaria and Hacrobia each include two phyla with chlorophyll-c containing chloroplasts of red algal origin (Cavalier-Smith et al. 2015a) , but Rhizaria are almost all hete r o t r o p h i c p h a g o t r o p h s — o n l y c e r c o z o a n o r d e r Chlorarachnida has true chloroplasts acquired together with a miniaturised secondary nucleus (the nucleomorph) by secondary symbiogenetic enslavement of a green alga (CavalierSmith 2013a; Keeling 2013) , and the cercozoan testate amoeba Paulinella has an enslaved cyanobacterium functioning as a chromatophore unrelated to chloroplasts (Cavalier-Smith 2013a) . The first site-homogeneous multigene trees (123 genes, 49 species), though robustly grouping Rhizaria with Halvaria, incorrectly put them as sisters to heterokonts only (Burki et al. 2007). Later, as evolutionarily more realistic siteheterogeneous amino acid substitution models were used, and taxon and gene sampling incrementally improved, Rhizaria more and more strongly were established as sisters of holophyletic Halvaria, not of heterokonts alone (Burki et al. 2008, 2009, 2012, 2016; Cavalier-Smith et al. 2015a) , though there can still be conflict between site-homogeneous and siteheterogeneous trees on this point (Krabberød et al. 2017). Rhizaria are megadiverse: Ruggiero et al. (2015) recognised 17 classes and 63 orders, although that classification erroneously included Pseudosporida and Rotosphaerida and omitted Axomonadida and Mikrocytida (Hartikainen et al. 2014) , and Cavalier-Smith (2018) established a new order Helkesida and new class Helkesea for some distinctive flagellates and amoebae, as well as new order Minorisida for the closest heterotrophic relatives of chlorarachnids. Class and ordinal relationships within both rhizarian phyla are mostly uncertain because of contradictory or poorly resolved singlegene trees and serious taxonomic undersampling of multigene trees. To clarify internal phylogeny of Cercozoa, and to test the monophyly of Cercozoa and Retaria more thoroughly and strengthen the basis for accurately placing Rhizaria within Harosa, we carried out partial transcriptome sequencing for eight Rhizaria: the net-like endomyxan Filoreta marina and seven Cercozoa in five non-chlorarachnean classes. Seven of the 8 cercozoan classes (including Imbricatea and Sarcomonadea whose monophyly has been uncertain) are now represented by 14 species as are 5 endomyxan species in our 162-taxon concatenated 187-gene alignment of 50,964 amino acids. For the first time, our multiprotein trees in conjunction with those of Sierra et al. (2016) and Krabberød et al. (2017) establish the relative branching order of all rhizarian classes, but suggest that Imbricatea may be paraphyletic (if scales evolved once) or polyphyletic (if scales arose twice). For comparison, we also show the first site-heterogeneous 18S rDNA trees to include both Helkesida and Ventricleftida as well as the recently described scaly, branching, filose amoeba Kraken (Dumack et al. 2016b, 2017a) : Kraken is apparently related not to Paracercomonadida as earlier supposed but to environmental DNA clade eSarcomonad (Scoble and Cavalier-Smith 2014) and possibly also to imbricate subclass P l a c o n u d a . T h e n o n - a m o e b o i d g l i d i n g c e r c o z o a n zooflagellate Helkesimastix because of its aberrantly long branch was not credibly placed on rDNA trees (CavalierSmith et al. 2009); it groups strongly on our multiprotein trees with the non-ciliate aggregative amoeba Guttulinopsis (Brown et al. 2012)—this clade, recently seen by 18S rDNA (Bass et al. 2016) , representing the recently established class Helkesea (Cavalier-Smith 2018), does not belong in Sarcomonadea or Thecofilosea. Our phylogenetic analyses support holophyly of revised phylum Retaria, and both its subphyla Ectoreta and Endomyxa, and revised Cercozoa, and show Reticulofilosa (represented by cercozoan classes Chlorarachnea and Granofilosea) branching robustly earlier than clade Monadofilosa. Within Monadofilosa, class Sarcomonadea is probably ancestral to class Imbricatea. We show that Granofilosea branch more deeply than Chlorarachnea, c o n f i r m i n g t h a t R e t i c u l o f i l o s a a r e a n c e s t r a l t o monadofilosans, and that Monadofilosa are a clade with Metromonadea the earliest branch—as some 18S rDNA trees weakly hinted (Howe et al. 2011a) . We discuss cell evolutionary implications of these improved phylogenies and explain how our multigene trees illuminate processes of protein sequence evolution, especially the pervasive but neglected importance of molecular coevolution, and the relative timing of key rhizarian cellular innovations. We list 25 key conclusions at the end of the paper. Materials and methods Details of culturing, RNA extraction and c-DNA library construction are in the electronic supplementary material for the reticulose non-ciliate endomyxan Filoreta marina Bass and Cavalier-Smith, 2009 (CCAP 1921/2) and seven separately studied Cercozoa: Minimassisteria diva Arndt and CavalierSmith in Howe et al., 2011a (CCAP 1947/1); Micrometopion nutans Howe et al., 2011a (CCAP 1904/1); Neocercomonas (= Cercomonas) clavideferens Vickerman in Bass et al., 2009 (strain 17-12-D (Bass et al. 2009b) , identical in rRNA ITS1 and 2 to type strain ATCC 50319; originally a Cercomonas but moved to Neocercomonas in Cavalier-Smith and Karpov, 2012b); Sandona ubiquita Howe et al. 2009 (strain C19 from Chile, closely related to the type strain W36; Nies-2427; see Howe 2008); Nudifila producta Howe and Cavalier-Smith in Howe et al., 2011a (CCAP 1911/1); Thaumatomonas oxoniensis Bass and Cavalier-Smith, 2011 in Howe et al., 2011a (CCAP 1903/2); and Helkesimastix marina CavalierSmith in Cavalier-Smith et al. 2009 (ATCC 50328). Filoreta was sequenced by standard Sanger sequencing by Agencourt Bioscience (Beverly MA, USA) and the seven Cercozoa by multiplex 454 pyrosequencing as previously described (Cavalier-Smith et al. 2014) . Supplementary Table S1 summarises the total sequences obtained. Assembly and editing of the 187-gene multiple alignment were described previously (Cavalier-Smith et al. 2014, 2015a,b) . Table S2 shows how some composite pseudotaxa were made by combining sequences from two species in the same genus to increase gene representation and is a key to labelling included sequences. Table S3 shows gene representation for each sequenced taxon and Table S4 lists the full names for the analysed genes. The Paulinella chromatophora Lauterborn, 1895 sequences are from the partial transcriptome of Nakayama and Ishida (2009) , who kindly provided them before public availability and agreed to their inclusion. As Micrometopion was fed on the bodonid flagellate Procryptobia sorokini (supplied by Mylnikov, presumably the type strain from a Sicilian saline lake: Frolov et al. 2001) , the transcriptome for this culture included genes from both species which were unambiguously separated by examining all 187 single-gene trees: of the 2 sequences present for most genes, 1 invariably grouped within bodonids, and was therefore from Procryptobia (confirmed by being extremely close but not identical to the sequences for another strain of Procryptobia sorokini: Deschamps and Moreira 2011, labelled Procryptobia sorokini M in the figures) and the other (Micrometopion itself) invariably grouped within Cercozoa and Rhizaria. As explained previously (Cavalier-Smith and Chao 2012a; Cavalier-Smith et al. 2015a; Howe et al. 2011a) , the Minimassisteria culture was also mixed, with the also axopodial naked heliozoan Oxnerella micra being included in a joint transcriptome. Though these two organisms are less evolutionarily divergent than are Micrometopion and Procryptobia, because they belong to different chromist subkingdoms (haptist Oxnerella in Hacrobia and cercozoan Minimassisteria in Harosa), we were able to decide which was which using the 187 single-gene trees. Multigene phylogenetic analyses used 50,964 amino acids and the best available site-heterogeneous amino-acid substitution model (PhyloBayes-MPI v.1b GTR-CAT-Γ 4 rates: Lartillot and Philippe 2004; Lartillot et al. 2013) using two separate chains. To verify that both components of the two mixtures were correctly identified, we ran eukaryote-wide analyses for 158 and 159 eukaryotes but excluded Plantae to avoid possible distortion by unrecognised red algal paralogues in chromists (see Deschamps and Moreira 2009; CavalierSmith et al. 2015a) . To avoid perturbation of rhizarian branching by distant outgroups and make convergence of PhyloBayes trees easier, we also ran trees restricted to chromists (72 taxa). For these and others restricted to Rhizaria only, we also ran maximum likelihood (ML) trees using the best available amino acid substitution model (LG) by RAxML-MPI v.7.2.8 PROTGAMMALGF (Stamatakis et al. 2005) and 100 fast bootstrap resamplings (Stamatakis et al. 2008) , even though this model is site-homogeneous and thus less evolutionarily realistic than GTR-CAT-Γ (henceforth called CAT for brevity) (Lartillot and Philippe 2004) . The acantharian Amphilonche elongata included by Sierra et al. (2013) but not Burki et al. (2010) was omitted from these trees as relatively few sequences are available and it is closely related to included Phyllostaurus sicula, both belonging to acantharian 18S rDNA clade F (Decelle et al. 2012) , but was included in trees restricted to rhizarians only. In addition to transcriptome sequencing, we attempted to partially sequence the genome from a few cells of a uniprotist culture of the thecofilosean amoeba Rhogostoma minus whose DNA was amplified using a whole genome amplification protocol ( see s upplementary material) i nstead of its transcriptome. The amplified DNA was prepared for and used in multiplex 454 DNA pyrosequencing together with the trancriptomes as Cavalier-Smith et al. (2014) described in detail. Unsurprisingly, we obtained relatively few sequences for our target genes and were only able to include 897 R h o g o s t o m a a m i n o a c i d s . We t h e r e f o r e e x l u d e d Rhogostoma from the main analyses, but included it in separate analyses of 74 and 75 chromist and 33 rhizarian taxa only. New 454 sequences are available from GenBank under B i o P r o j e c t n u m b e r s S R P 0 4 8 6 5 3 - 6 , S R P 0 4 8 6 5 8 , SRP048662 and SRP053325. Filoreta marina ESTs are available under numbers JZ972228–JZ976734. We excluded sequences of Mikrocytos mackini from alignments because of its excessively long branch on multigene trees (Burki et al. 2013) , so attempted to obtain sequences from the haplosporidian, Minchinia chitonis, as a trans c r i p t o m e u n d e r t h a t n a m e i s p u b l i c l y a v a i l a b l e (MMETSP0186; Slamovits in Sierra et al. 2016, collected from a beach in Devon, UK in 2010) ; we downloaded it from h t t p : / / c a m e r a . c a l i t 2 . n e t / m m e t s p / d e t a i l s . p h p ? i d = MMETSP0186 in the hope of including data for a representative of the parasitic cercozoan class Ascetosporea; blasting this against our alignment revealed 57 potential gene matches. As our quality-control single-gene trees showed that some ‘Minchinia’ sequences grouped strongly with diverse protist phyla other than Cercozoa (mostly diatoms or Micromonas), we excluded all these and any obviously animal sequences from our alignment as probable contaminants and concatenated the rest as putative ‘Minchinia’ genes. Unfortunately, our multigene trees for 74 or 75 chromist taxa showed that this concatenated sequence also did not group with Cercozoa but with the cryptist Palpitomonas, implying that these genes were either from a single cryptist contaminant genuinely related to Palpitomonas or else a confusing mixture of one or more non-ascetosporean contaminants and genuine Minchinia genes that collectively grouped with Palpitomonas as an artefactual compromise. Therefore, these supposed ‘Minchinia’ sequences were excluded from our main trees (e.g. Figs. 1 and 2) as irrelevant to the position of haplosporidia and potentially distorting were they a mixture not one contaminant. To verify that they grouped with cryptists and not some non-chromist group, we ran 162taxon eukaryote-wide trees with Plantae represented only by the short-branch glaucophytes. As a further check against tree distortion by more distant outgroups and of reproducibility amongst taxon samples, we also ran Rhizaria-only trees with (32 taxa) and without Rhogostoma (31 taxa), and additionally without Limnofila (30 taxa) as Bass et al. (2009a) suggested it might have been contaminated with other protists and potentially misleading. These three alignments also included the third acantharian Amphilonche elongata, excluded from our other trees because only 2621 amino acids are available; Amphilonche grouped with Phyllostaurus with 100% ML and 0.99 CAT support (in agreement with Sierra et al. 2013, 2016) and did not change topology elsewhere in the tree. We tried to obtain fast bootstrap RAxML trees for 158, 159, and 162 taxa, but calculating the final optimal ML tree exceeded the memory capacity of the 256 processors used: runs terminated without outputting an optimal tree. This is not a serious problem as it is generally accepted that siteheterogeneous trees are evolutionarily more accurate than RAxML LG as they better fit eukaryote multigene data (e.g. Brown et al. 2013; Cavalier-Smith et al. 2014, 2015a, b) , so some studies now use only CAT (Derelle et al. 2015). For comparison with the gene-rich multiprotein trees, we conducted site-heterogeneous 18S rDNA phylogenetic analyses, using PhyloBayes CAT-GTR-Γ with 4 rate classes, for the taxonomically most comprehensive rhizarian alignment to date (464–467 species, a balanced selection of all major lineages of Rhizaria), and for 316/7 species of Monadofilosa only to check the position of Kraken and Cyranomonas and see whether exclusion of more distant Rhizaria improved tree topology. Results Separation of mixed protist cultures The eukaryote-wide 159-taxon tree (Fig. 1) shows that each c o m p o n e n t o f b o t h m i x e d t w o - p r o t i s t c u l t u r e s ( M i n i m a s s i s t e r i a / O x n e re l l a a n d M i c ro m e t o p i o n / Procryptobia) goes to its right phylum, indicating that their sequences were correctly assigned using the single-gene trees. Minimassisteria is within Cercozoa with maximal support by CAT and sister to Limnofila borokensis, the only other granofilosean, with slightly weaker support (0.93). The separation of Minimassisteria from the co-cultured centrohelid heliozoan Oxnerella micra (red arrow, Fig. 1) is clear; the latter branches within subphylum Hacrobia as sister to Polyplacocystis (the only other centrohelid on the tree) with maximal support, this clade being robustly sister to haptophytes (0.97) in agreement with Cavalier-Smith et al. (2015a). Minimassisteria would have to cross five maximally supported nodes (and three less well-supported nodes) to adopt a sister relationship with Oxnerella. Micrometopion branches within Cercozoa as the most divergent member of Monadofilosa, as 18S rDNA previously suggested (Howe et al. 2011a) . By contrast, its food Procryptobia sorokini is the maximally supported sister to another culture of Procryptobia sorokini (labelled M) isolated by Deschamps et al. (2011) from deep Marmara Sea sediment. The P. sorokini we used to feed Micrometopion (presumably the type strain from a saline lake: Frolov et al. 2001) branched closely (Fig. 1 Fig. 1 PhyloBayes GTR-CAT-Γ tree of 159 eukaryote-wide taxa, exclud-„ ing Plantae, using 187 proteins (50,964 amino acid positions). Black arrows show new rhizarian sequences and red arrows the two nonrhizarian sequences from two mixed cultures that we separated phylogenetically. Numbers after species names show how many amino acids were included for each. Most bipartitions had maximal support (1); posterior probabilities are only shown if they did not—in red for the only three for which both chains did not show this topology (max.diff. 1; 5735 trees summed for two chains after removing 1674 as burnin). On all figures, Rhizaria subgroup names reflect the revised Table 1 classification red arrow) with the deep sea Deschamps et al. strain that is sufficiently different to be an unrecognised sibling species. Rhizarian phylogeny Rhizaria are always maximally supported as a clade and as sister to harosan infrakingdom Halvaria. Taxa are named here in accord with the revised classification in Table 1. Cercozoa and retarian infraphylum Foraminifera are invariably maximally or 99% supported as clades. The 159-taxon consensus tree (Fig. 1) weakly (0.69) shows Endomyxa as sister to Ectoreta (classical Retaria), chain 2 showing this with maximal support; but chain 1 contradicted that by grouping Endomyxa with Cercozoa (0.63), thus excluding Ectoreta from classical Cercozoa. Within Ectoreta, infraphylum Radiozoa is a maximally supported clade in Fig. 1 on both chains, but for a separate 158-taxon tree differing only by omitting Microheliella (Supplementary Fig. S1), one chain showed this with maximal support while on the other Radiozoa appear paraphyletic with Polycystinea sister to Foraminifera also with maximal support. Both chains of the 158-taxon tree excluded Ectoreta from Endomyxa with reasonably good support (0.77, 0.99; consensus 0.88), but were contradictory over whether Ectoreta are sisters to Endomyxa or to Cercozoa plus Endomyxa, which together weakly formed a classical Cercozoa clade on the consensus tree (0.56). For analyses restricted to the 72 chromists from the 158taxon trees (Microheliella absent), both chains converged satisfactorily (Fig. 2), and we got a corresponding ML tree (Fig. S2). Both methods strongly showed Ectoreta as sister to Endomyxa (1, 96%), forming a clade corresponding to expanded Retaria, with Endomyxa weakly (0.58, 45%) a clade. Both methods strongly showed Radiozoa as a clade (1, 99%). For Cercozoa, the (Fig. 2) CAT branching order was identical to the 158-taxon tree (Fig. S1, both chains) and to chain 1 of the 159-taxon tree (part of Fig. 1 analysis), with Granofilosea the deepest branch followed by Bigelowiella (Fig. 2). This is contradictory to Fig. 1 where Bigelowiella was weakly sister to Granofilosea (0.67 as shown in the consensus tree (Fig. 1 itself) and 0.87 in chain 2 alone). In eukaryote-wide trees, basal branching for Cercozoa generally was more strongly supported for 158 taxa (Fig. S1 Multigene phylogeny and cell evolution of chromist infrakingdom Rhizaria: contrasting cell organisation of... Fig. 2 PhyloBayes GTR-CAT-Γ consensus tree of 72 chromists using 187 proteins (50, 964 amino acid positions). Black arrows show new rhizarian sequences; the red arrow highlights Oxnerella, whose sequences were phylogenetically separated from Minimassisteria. Numbers after species names show how many amino acids were included for each. Most bipartitions had maximal support by both CAT and ML (1/100); support values are only shown for those that did not (posterior probabilities left; ML 100 fast bootstraps right). Dashes indicate bipartitions not found on the corresponding ML tree (Fig. S2). The two chains converged satisfactorily (maxdiff 0.244572; 25,345 trees summed after removing 14,399 as burnin) summing 25,433 trees, slightly more than for Fig. 2) than for 159 taxa (Fig. 1 summing only 5735). Monadofilosa is a maximally supported clade on 72- and 158-taxon trees and strongly so (0.93) on Fig. 1. All three trees had identical m o n a d o f i l o s a n t o p o l o g y w i t h M i c r o m e t o p i o n (Metromonadea) strongly supported as the deepest branch (six independent chains agreed, i.e. both chains for the 159, 158 and 72 taxa analyses) (Figs. 1, S1 and 2). Next deepest is a clade comprising the sainouroid flagellate Helkesimastix and aggregative amoeba Guttulinopsis (order Helkesida, maximally supported as a clade on all trees by both methods: Figs. 1 and 2). Helkesida are sister to a large, moderately to strongly supported (0.92 Fig. 1; 0.97 Fig. S1; 0.75 Fig. 2) clade composed of the closely related Thecofilosea, Sarcomonadea and Imbricatea, all three of which are revised in Table 1. Overall, our chromist-only CAT trees strongly show Granofilosea as the deepest branching cercozoan lineage, in contrast to many extremely weakly supported but taxonomically much richer site-homogeneous 18S rDNA trees that tend to show Bigelowiella (Chlorarachnea) branching more deeply than Granofilosea (e.g. Cavalier-Smith and Chao 2003a; Bass et al. 2009a; Howe et al. 2011a; Yabuki and Ishida 2011; Shiratori et al. 2012; and our present ML trees, e.g. Fig. S9) . By contrast, chromist-only ML trees grouped Bigelowiella and Minimassisteria with trivial support and misplaced Limnofila with similarly insignificant support; its position is more unstable than for Minimassisteria (see discussion regarding the conflicting positions of Limnofila). Sarcomonad phylogeny Within Sarcomonadea Neocercomonas is strongly sister to the glissomonad Sandona (maximally on Figs. 1 and 4, S1; 0.99 by CAT, 99% by ML in Fig. 2; support dropped to 0.75/81 on Fig. 3 that included the genically grossly undersampled Rhogostoma), corresponding to new sarcomonad subclass Pediglissa (Table 1), whereas Paracercomonas always branches substantially more deeply and is now put in a separate subclass in keeping with its distinctive cytoskeleton (see discussion). With CAT, the amoeboflagellate Nudifila and photosynthetic euglyphid amoeba Paulinella are sisters with moderate to strong support (0.75 Fig. 1; 0.88 Fig. S1; 0.85 Fig. 2; 0.83 Fig. 3; 0.77 Fig. 4), and this clade strongly supported (0.99, 0.99, 0.92, 0.67, 1) as sister to Pediglissa. In Fig. S2 (ML equivalent of Fig. 2), long-branch and genically sparsely represented Paulinella only is extremely weakly (29%) sister to Pediglissa; Nudifila is their immediate outgroup; this one node lower position, arguably a long-branch/missing data artefact, was similarly weakly supported on other ML trees (Figs. S3 and S4). This close relation between the imbricate Nudifila/Paulinella subclade and Pediglissa on all these trees shows that Sarcomonadea are probably paraphyletic, and cercomonads are polyphyletic, as some ML rDNA trees and ultrastructure previously tentatively suggested (Cavalier-Smith and Karpov 2012b) . All (Figs. 1, 2 and 3) CAT trees put the other scaly imbricate flagellate Thaumatomonas (sole representative of imbricate subclass Placoperla) as sister to the Pediglissa/Nudifila/Paulinella c l a d e , t h u s c l o s e r t o N e o c e r c o m o n a s t h a n i s Paracercomonas, reinforcing cercomonad polyphyly. No trees group Thaumatomonas with Paulinella/Nudifila as would be expected were Imbricatea a clade. The grouping of Pediglissa instead with Paulinella and Nudifila is sufficiently strong by CAT in Figs. 1 and 2 that Imbricatea are most likely paraphyletic. Imbricate subclasses Placonuda and Placoperla each has a mixture of naked species and others secreting siliceous surface structures—scales in Placonuda, scales or perles in Placoperla. In addition, we establish here the new imbricate subclass Krakenia (see below) to include the unique scaly reticulose amoeba Kraken (Dumack et al. 2016b, 2017a) . The joint Sarcomonadea/Imbricatea clade is strongly supported on Figs. 1 and 2 CAT trees (weakly on Fig. 3 in the presence of grossly undersampled Rhogostoma). However, except for the strong clade Pediglissa, no branches within that joint clade are significantly supported by ML (showing that CAT is superior for these very close branches), though ML internal branching order of Sarcomonadea is the same except for Nudifila/Paulinella not being a clade. However by ML (Fig. S2), Micrometopion moved into sarcomonads to become sister to Thaumatomonas (trivial 25%; seen also on one CAT tree: Fig. 4) and Limnofila also wrongly moved into sarcomonads to be sister to Paracercomonas (insignificant 42%) and Minimassisteria became sister to Bigelowiella (26%). This further shows that ML LG is often less accurate than CAT, but there are no statistically significant contradictions by ML to the stabler, well-supported CAT-based topology. Incorrect attraction of Limnofila and Paracercomonas might be an artefact of weak gene sampling or else stem from contamination of the sequenced culture by a paracercomonad, as our discussion explains. If the Limnofila transcriptome has contaminating paracercomonad genes, they might also a r t e f a c t u a l l y p u l l P a r a c e r c o m o n a s a w a y f r o m Neocercomonas giving a spurious position intermediate between Micrometopion and Neocercomonas. To test whether Paracercomonas is misplaced in Figs. 1, 2, 3 and 4 by the possible presence of paracercomonad genes in a contaminated Limnofila culture, we also ran trees for Rhizaria only including Limnofila (31 taxa Fig. S5) and excluding it (30 taxa Fig. S7). Figure S7 shows that Paracercomonas still does not group with Neocercomonas even when Limnofila is excluded, thus the poly/para-phyly of cercomonads is robust and not distorted by the sparsely sampled and likely contaminated Limnofila transcriptome. Revised classification of chromist infrakingdom Rhizaria and its 2 phyla, 4 subphyla, 18 classes, and 65 orders As there has been no recent comprehensive summary of cercozoan classification, families are shown for all Cercozoa except Phaeodaria and Ebriida (neither with recent changes), but only for vampyrellids in Retaria a Taxa represented on our multiprotein trees b Chantangsi and Leander (2010) used these names for clades but did not validly publish them as families c Discomonadida were originally placed incertae sedis within Ventrifilosa and not assigned to a class (Scoble and Cavalier-Smith 2014) ; our more broadly sampled PhyloBayes 18S rDNA trees (e.g. Figs. S8, S12) make it even less likely that Discomonas are thecofiloseans. The Rhizaria-wide Figs S8–11 put them as weakly/insignificantly sister to Pediglissa plus Paracercomonadida and the probably more reliable Monadofilosa-only trees (Figs S12–15) put them within Imbricatea, grouping them with Placonuda; following the latter we place Discomonadida in Placonuda, consistent with their little known morphology d The five rhizarian orders with no rDNA or other sequence evidence for their taxonomic position e Without sequence data for this genus, we cannot be sure that the other three listed genera belong in this family f Probably paraphyletic g Amphitremida are Labyrinthulea (Gomaa et al. 2013; Takahashi et al. 2014) not Cercozoa (Cavalier-Smith and Chao 2012) Phylogeny of Thecofilosea Mataza and the phaeodarian Aulacantha are consistently a clade (class Thecofilosea) with moderate to strong CAT support on Figs. 1, 2, 3 and 4 (0.95; 0.94; 0.75; 0.99) but weak ML support (49% Fig. S2; 33% Fig. S3) probably because of low gene representation for Aulacantha. The Mataza sp. sequences (MMETSP0087) are from a previously unidentified cultured marine protist from California (downloaded from http://; strain D1 Palenik unpub.; link no longer available— sample/view/1715 is a replacement), which our cercozoan 18S rDNA tree (Supplementary Fig. S8) shows is so closely related to Mataza hastifera (Yabuki and Ishida 2011) that it must belong to that genus, but sufficiently different that it must represent a second undescribed species. Partial genome sequencing from the thecofilosean amoeba, Rhogostoma minus, be longing to a different order (Cryomonadida) from the other thecofiloseans, Mataza and Aulacantha, provided too few sequences for our 187 genes (only 897 amino acids) to be included in Figs. 1 or 2 without risking distortions because the vast majority of genes are missing. As multigene trees can sometimes place incomplete short sequences correctly if trees include a fairly close relative with good gene sampling, we also ran chromist-only trees after adding Rhogostoma and sequences ostensibly from the ascetosporean parasite ‘Minchinia’: an analysis of 75 taxa (Fig. 3) including all 73 chromists from Fig. 1, and another of 74 taxa including all 72 chromists from Fig. S1 [as CAT did not fully converge (max. diff. 1), Fig. S4 shows only the ML analysis]. These 75- and 74-taxon trees were generally similar to the 72-taxon tree (Fig. 2) but Rhogostoma was contradictorily placed: one (Fig. 3) CAT chain put it in the maximally supported Thecofilosea clade as sister to Aulacantha (0.99) in agreement with rDNA (Howe et al. 2011a) , the other probably wrongly as sister to Micrometopion (1). In Fig. 3, the two chains were contradictory in some places (e.g. one put Ectoreta as sister to Endomyxa as in Fig. 2 and the other placed it within Endomyxa as sister to Gromiidea as shown weakly on Fig. 3 consensus tree) probably because they were run for much less time than for Fig. 2. Therefore, we ran a third 75-taxon chain somewhat longer and summed 2194 trees; this also showed Ectoreta as sister to Endomyxa, being excluded from them with 0.61 support (also by ML: Fig. S3, 52% support). A separate 32-taxon multigene tree restricted to Rhizaria (supplementary Fig. S6; maxdiff 0.278303) grouped Rhogostoma correctly but with trivial support (0.36) within Thecofilosea (as sister to the phaeodarian Aulacantha rather than to Mataza as with rDNA, all three a weak (0.44) thecofilosean clade). Thus in the absence of non-rhizarian outgroups, CAT can place Rhogostoma with only 897 amino acids almost as expected from taxon-rich 18S rDNA trees. Despite adding Rhogostoma and ostensible ‘Minchinia’ (Fig. 3), internal phylogeny of both rhizarian phyla by CAT is identical to Figs. 1 and 2. Thus Rhogostoma’s sequence shortness makes little or no difference to tree topology but prevents its reliable placement because so few amino acids cause high random errors. That is consistent with 18S rDNA trees where all three Thecofilosea mutually diverge close to the base of Thecofilosea, with Aulacantha deepest (Fig. S8; Howe et al. 2011a; Yabuki and Ishida 2011where the Rhogostoma sequence was incorrectly called Lecythium sp. because of earlier misidentification) . Endomyxan phylogeny Minchinia is a haplosporidian parasite, which rDNA trees established as an endomyxan cercozoan (Cavalier-Smith 2002a; Bass et al. 2009a) . The fact that supposed ‘Minchinia’ did not group with Endomyxa (as it always does by rDNA (Figs. S8–S11) or even within Rhizaria or Harosa, but strongly (1, 84%) within Hacrobia (Fig. 3) shows that most sequences included from this transcriptome are not from Minchinia but unrelated contaminants from one or more deepbranching hacrobians (as methods explain, we removed obvious diatom and Micromonas contaminants, but our singlegene trees could not have detected all chromist contaminants as they included no certain ascetosporan proteins). As one cannot cultivate ascetosporans, their spores have to be extracted from wild marine invertebrates, making contamination by unrelated marine protists likely. In the 74-taxon tree, ‘Minchinia’ is weakly (0.5, 42% Fig. S4) sister to the deepbranching marine cryptist Palpitomonas. As ‘Minchinia’ groups within Cryptista, we also ran 75-taxon trees including the pseudoheliozoan Microheliella recently shown to branch nearby (Cavalier-Smith et al. 2015a) in case this would clarify its position. This confirmed that ‘Minchinia’ branches as sister to Palpitomonas (0.73, 45%) and that Microheliella is sister to Corbistoma and does not branch with ‘Minchinia’ (Fig. 3). This tree did not converge, chains differing in three ways: (1) one showed Endomyxa as a clade, the other made it paraphyletic with Ectoreta sister to Gromiidea only (and on consensus Fig. 3); (2) one correctly grouped Rhogostoma with Aulacantha (0.51) (but wrongly put the three-species Thecofilosea clade as sister to Thaumatomonas not Sarcomonadea as in Fig. 2), whereas the other wrongly put Rhogostoma with Micrometopion (1.0); (3) Limnofila was correctly with Minimassisteria on one chain, wrongly with Paracercomonas on the other. For CAT, cercozoan topology is otherwise identical to Figs. 1 and 2; ML topology was identical for the 72, 74, 75 taxon trees. This novel inclusion of ‘Minchinia’ within Cryptista did not alter its topology from that recently established (Cavalier-Smith 2015) . To rule out the possibility that ‘Minchinia’ belongs to a non-chromist group excluded from Fig. 3, we ran a 162taxon CAT tree including it (and Microheliella) and two glaucophytes as short-branch Plantae to rule out the possibility that ‘Minchinia’ is closer to plants than to chromists. In this broader analysis, ‘Minchinia’ was not sister to Pa lpitomonas but instead to either Rollomonadia (Cryptomonada plus Leucocrypta) in chain 2 (Fig. 4) or to core Cryptista (Palpitomonas plus Rollomonadia) on chain 1 and the consensus tree. Although the two chains did not fully converge, their topology was identical in all except five respects: (1) they showed exactly the same conflict in the position of the two metamonad clades as did the 158-taxon analyses; (2) the exact position of Retaria differed—strongly sister to Endomyxa in chain 2 (Fig. 4) and the consensus but weakly (0.54) to Gromiidea only in chain 1; (3) Microheliella was sister to ‘Minchinia’ plus core Cryptista in chain 2 but to Corbistoma in chain 1 and the consensus tree forming a Corbihelia clade as in Fig. 3 and previously (Cavalier-Smith et al. 2015a) ; (4) Micrometopion is wrongly sister to Thaumatomonas in chain 2 but in the same position as on Figs. 1, 2 and 3 in chain 1 and the consensus tree. (5) Chromista were holophyletic with maximal support (with Glaucophyta maximally supported as their sister) on chain 2 (Fig. 4), but chain 1 put glaucophytes weakly (0.71) within them as weak sister to core Cryptista plus ‘Minchinia’. This shows that supposed ‘Minchinia’ is not a harosan, rhizarian or cercozoan, and not closer to any non-chromist group. We ran the analysis long enough to establish that, but given the agreement between Figs. 1, 2, 3 and 4 on the internal phylogeny of Cercozoa, it was not worth while to run them longer in the hope of reducing these conflicts, especially as these sequences are apparently predominantly from a cryptist contaminant [a core cryptist (Fig. 4) or one related to Palpitomonas (Fig. 3)] or from one or more contaminants mixed with some genuine Minchinia genes (see ‘Discussion’ section). Figures 1, 2, 3 and 4 CAT trees show Radiozoa as a clade with maximal support, as did one chain of the 158-taxon analysis. Radiozoa are also a clade on corrresponding ML trees with strong support: 99% for 72/75-taxon trees (Figs. 2/S2; 3/ S3) and 100% support for 74 taxa (Fig. S4). Within Endomyxa, our extensive new sequences for Filoreta marina group with maximal support with the previously sequenced but sparsely sampled Filoreta tenera. However, two contradictory phylogenies were found for Endomyxa. In four CAT analyses, Filoreta and Gromia are sisters in all eight chains, a clade corresponding to revised class Gromiidea (Table 1); support was weak in chromistwide samples that did not converge (Fig. 3 0.55; Fig. 4 (0.60 in chain 2 shown; 0.68 in consensus, chains not converged), but strong in smaller, better converged Rhizaria-only alignments—in a 31-taxon tree that excluded Rhogostoma (Fig. S5 max. diff. 0.290668; 0.97) and a 32-taxon one with Rhogostoma (Fig. S6, max. diff. 0.278303; 0.95). The 31taxon ML tree had a strong (96%) Gromiidea clade. In four other CAT trees, Gromiidea was weakly paraphyletic with Gromia sister to plasmodiophorids instead (Figs. 1, 2 and S1). ML trees corresponding to Figs. 2 and 3 had weakly paraphyletic Gromiidea (Figs. S2–4: 57, 63 and 55%). Endomyxa are most often a clade, moderately supported (0.71) on Fig. 1, weakly (0.58, 45%) on Fig. 2. However, in contrast to all our other trees, our converged 31-taxon Rhizaria-only tree (Fig. S5) placed Ectoreta between Phytomyxea and Gromiidea; if rooted as in Fig. 4, they would be sisters to Gromiidea with 0.99 and 65% support; the same contradiction is evident on Fig. S6 (that includes Fig. 3 187-protein PhyloBayes GTR-CAT-Γ tree for 75 chromists; those on Fig. 2 plus Rhogostoma, Microheliella and ‘Minchinia’ (blue arrow). Black arrows show new rhizarian sequences; the red arrow highlights Oxnerella, whose sequences were phylogenetically separated from Minimassisteria. Numbers after species names show how many amino acids were included for each (maximum possible 50,964). Most bipartitions had maximal support by both CAT and ML; support values are only shown for those that did not (posterior probabilities left; ML 100 fast bootstraps right); in red for those where the two summed chains gave conflicting topology (max. diff. 1; 3499 trees summed after removing 1016 as burn-in; as the text explains, a third chain agreed with this consensus topology for Cercozoa, with much stronger support (except for the incorrect position of Rhogostoma), but placed Ectoreta as sister to Endomyxa in agreement with Fig. 2). The corresponding ML tree is Fig. S3. P. p. pro parte (in part) Rhogostoma), and thus is a consistent feature of Rhizaria-only trees—it might result from long-branch attraction of the very long Foraminifera/Radiozoa branch towards the gromiid branch that is longer than the phytomyxan one. In Figs. 1, 2, 3 and 4 and S1–4, the presence of non-rhizarian outgroups that break the stem between Cercozoa and Retaria might largely or entirely prevent that by enabling algorithms to reconstruct ancestral states more accurately. Site-heterogeneous rhizarian 18S rDNA trees Our comprehensive site-heterogeneous Bayesian 18S rDNA trees with 467 Rhizaria (Fig. S8) agree (as do trees with fewer or no Helkesida—Figs. S10/11) with the taxonomically far sparser multiprotein trees in placing Granofilosea more deeply than Chlorarachnea. 18S rDNA shows Gromiidea as monophyletic (CAT and ML: paraphyletic because Ascetosporea are robustly sister to Gromiida only). Because in some respects these trees differ within Monadofilosa from the previous most comprehensive CAT analyses for Monadofilosa (Scoble and Cavalier-Smith 2014) , and because these analyses were done before the remarkable aciliate filose amoeba Kraken was discovered [shown to have single-tier (?silica) scales and flat cristae, a previously unknown structural combination] (Dumack et al. 2016b, 2017a) , we also did analyses restricted to 315 or 316 Monadofilosa (using 1790 amino acid positions: Figs. S12–S15) including Kraken and several recently discovered Thecofilosea and imbricates, with one short-branch granofilosean as an outgroup. New sequence addition enabled slight improvements in alignment in the most divergent regions (mainly in Thecofilosea and imbricates) so we could include the whole molecule (only terminal PCR primer regions and rare inserts were excluded). In one notable respect, the Monadofilosa-only 18S rDNA trees (Figs. S12–15) disagreed with the Rhizaria-wide 18S rDNA trees but agreed with the CAT multiprotein trees (Figs. 1, 2, 3 and 4): Cercomonadida are not sisters of Paracercomonadida. Yet when distant rhizarian outgroups are included in rDNA trees (Figs. S8, S10 and S11), paracercomonads are sister to cercomonads with moderate support—one of our ML tree groups them (Fig. S9) with insignificant support but the others do not; published sitehomogeneous trees may group them or not depending on taxon sampling. Supplementary Material (after Fig. S15) discusses other phylogenetically important aspects of the new cercozoan-wide rDNA analyses. These Monadofilosa-only CAT analyses (Fig. S12 including Cholamonas, S13 omitting long-branch helkesid Cholamonas) were more similar in topology to the Scoble and Cavalier-Smith trees, suggesting that adding more distant, often longer branch, reticulofilosan and retarian outgroups in Rhizaria-wide trees slightly distorts branching order in the hard-to-resolve ventrifilosan basal radiation. CAT trees (summarised for the more comprehensive analysis in Fig. 5) differed from that of Scoble and Cavalier-Smith only in two Fig. 4 187-protein PhyloBayes GTR-CAT-Γ tree of 162 eukaryote-wide„ taxa, including short-branch glaucophytes to represent Plantae and ‘Minchinia’ (blue arrow). Black arrows show new rhizarian sequences and red arrows the two non-rhizarian sequences from the two mixed cultures that we separated phylogenetically. Numbers after species names show how many amino acids were included for each. Most bipartitions had maximal support on both chains; posterior probabilities are only shown if they did not—in red for those for which both chains did not show this topology (max. diff. 1); the tree shown is for chain 2 (3571 trees summed after removing 3148 as burn-in); in chain 1, Cryptista were rearranged, as in Cavalier-Smith et al. 2015a, with Microheliella forming a Corbihelia clade with Telonema and Picomonas (0.84 support) and Glaucophyta were weakly within Hacrobia (0.64, 0.56 support) phylogenetically important (but statistically insignificant) respects. First, Spongomonas is sister to Thaumatomonadida in agreement with some ML trees and with Rhizaria-wide analyses, and did not branch within them as sister to the deepest (Peregriniidae/Esquamulidae) clade as it did (probably incorrectly) in Scoble and Cavalier-Smith (2014) . Second, Discomonas was not sister to classical imbricates but deeply within them as sister to Placonuda, as in an earlier Rhizariawide analysis that led to their classification in Imbricatea (Howe et al. 2011a) , but in disagreement with our present Rhizaria-wide analyses that weakly put Discomonas as sister to Pediglissa plus paracercomonads. We also did 316/7 taxon ML trees (Figs. S14 and S15); all four Monadofilosa-only trees failed to group Kraken with paracercomonads as taxonomically much sparser sitehomogeneous trees originally found (weakly by 18S alone, strongly with both rDNAs: Dumack et al. 2016b, 2017a) . By CAT Kraken was sister instead to environmental DNA clade eSarcomonad (on Figs. S8, S10–11 within classical imbricates, but as in Fig. S9 weakly sister to glissomonads as originally: Scoble and Cavalier-Smith 2014) , not included by Dumack et al. (2016b), with marginally stronger support ( 0 . 6 4 : F i g s . 5 a n d S 1 2 ) t h a n t h e y f o u n d f o r t h e p a r a c e r c o m o n a d g r o u p i n g . O n S 1 2 , t h e h e l k e s i d C h o l a m o n a s w a s i n s i g n i f i c a n t l y ( 0 . 4 3 ) s i s t e r t o Cercomonadida and Paracercomonadia were the deepest branching sarcomonad clade. Thus as in Dumack et al. (2016b), Cercomonadida and Paracercomonadida were not sisters, but when Cholamonas was omitted they became sisters (Fig. S13 0.84, but only 0.5 in a 318 taxon analysis with an extra deep-branching imbricate). Omitting Cholamonas made eSarcomonad unstable (Fig. S13): one chain showed the same topology as Figs. 5 and S12 but in the other only the two original short-branch eSarcomonad sequences remained significantly (0.67) sisters to Kraken, the two longer ones that branched within them moved into pansomonads within Glissomonadida. As their intrusion into pansomonads reduced support for pansomonads, glissomonads and subclass Krakenia (established in Table 1 for Kraken plus eSarcomonad) on the consensus tree (Fig. S13), it is likely a long-branch attraction artefact. This confusing attraction may Multigene phylogeny and cell evolution of chromist infrakingdom Rhizaria: contrasting cell organisation of... be specifically related to aberrant long-branch environmental DNA AY620290 which by ML grouped within eSarcomonad not within pansomonads (see additional supplementary discussion). One consistent difference from the Rhizaria-wide trees (Figs. S8–11) is that Ventricleftida did not group with or near Helkesida or become two separate clades. Ventricleftida were a moderately supported clade sister to other Thecofilosea as in Howe et al. (2011a) not to classical imbricates as in Scoble and Cavalier-Smith (2014) . Our trees confirm that environmental DNA AB695519 from Antarctic lake moss is Kraken and also identify a soil sequence from a North American aspen forest (EF024287) as from Kraken. With ML Kraken did not group with eSarcomonad or any other specific clade but was the third deepest branch after metromonads and discomonads without significant support. eSarcomonad (probably wrongly including AY620290) was insignificantly (9%) sister to glissomonads. All Monadofilosa-only trees rooted solely by a shortbranch granofilosan showed Metromonadea as a clade (up to 98% support), indicating that including long-branch Chlorarachnea and more distant long-branch outgroups in the Rhizaria-wide (and similar published) trees probably distorts monadofilosan topology by making them wrongly appear paraphyletic. Thus our multiprotein trees including only Micrometopion likely accurately represent the position of the whole class as the deepest branching Monadofilosa. By CAT (not ML), rDNA Metromonadea was the deepest-branching monadofilosan clade. Our trees also more clearly than before e s t a b l i s h b r a n c h i n g o r d e r w i t h i n M a r i m o n a d i d a : Cyranomonas is more closely related to the Pseudopirsonia/ Auranticordis clade than to Abollifer (Figs. S12–S15). We establish new marimonad families to reflect their ultrastructural diversity and substantial branching depth on the trees (Table 1), and a new family for Quadricilia (rDNA sequence unavailable when we ran our trees), which goups with Clautriavia and Nudifila (weakly by ML but strongly on a site-homogeneous Bayesian tree) as the first non-gliding variglissid (Yabuki and Ishida 2018) . Their ML tree weakly grouped Kraken with Ventrifussura as sister to Imbricatea, and like ours did not group them with paracercomonads as in Dumack et al. (2016b) and thus agrees with our placing Kraken in Imbricatea (Table 1). Figures 5, S12 and S13 trees weakly show superorder Nudisarca as a clade, as in Rhizariawide trees (insignificantly paraphyletic in Figs. S14–15). The overall eukaryote multiprotein tree In the absence of Plantae, chromist subkingdoms Harosa and Hacrobia are sister clades, both maximally supported (Figs. 1 and S1). The maximally supported chromist clade is sister to the maximally supported scotokaryote clade (i.e. podiates plus Malawimonas and Metamonada). Halvaria (comprising Alveolata and Heterokonta) is a maximally supported clade Fig. 5 Summary of ordinal and subordinal relationships amongst Monadofilosa as shown by a site-heterogeneous tree for 317 18S rDNA sequences (complete tree is Fig. S12). Support values are posterior probabilities from Fig. S12 and on the right also for ML (Fig. S14) for the same alignment when exactly the same clades were present. Support values for terminal clades are to the right of their names. The position of Phaeodaria (excluded from these analyses in case their long branches artefactually attracted Cholamonas) is from the Rhizaria-wide analyses (Figs. S8–S11) where they were sister to Fiscullina irrespective of Cholamonas presence or absence by CAT in Figs. 1, 2, 3, 4, and S1, although not always by ML (95% in 2/S2, 82% in 3/S3). Alveolata is always a maximally supported clade by both methods (Figs. 1, 2, 3, 4 and S1). Heterokonta are maximally supported by CAT and usually maximally by ML (except Figs. 3 and S3). However, in the presence of glaucophytes to represent Plantae, support for Harosa reduced and Telonema (only) was excluded from the Hacrobia clade but Chromista remained a maximally supported clade that was maximally supported as sister to Plantae (i.e. superkingdom Corticata is a clade). Whether movement of Telonema was caused by adding glaucophytes or the ‘Minchinia’ sequence that might in fact be a mixture of hacrobian and harosan sequences is unclear, but does not affect Fig. 4’s clear evidence for holophyly of Chromista. As noted above, CAT trees for the exceptionally large Fig. 1 alignment did not converge because of contradictions within Rhizaria. Topology was identical for both chains for all nonrhizarian branches, including those that gave contradictory results in published trees near the base of scotokaryotes and Hacrobia (e.g. Cavalier-Smith et al. 2014, 2015a) . The hacrobian sister phyla Cryptista and Haptista were each clades in Figs. 1, 2 and 3 by both methods: Haptista were always strongly supported as a clade by both methods (maximally or nearly so by CAT), whereas CAT support for Cryptista was maximal in Figs. 1 and 3 but weak in Fig. 2. Presence of the ‘Minchinia’ putative cryptist/haplosporidian mixture disrupted Cryptista on eukaryote-wide (Fig. 4) but not on the chromist-only (Fig. 3) where cryptist subphylum Corbihelia was a clade. Topology of Eozoa was maximally supported throughout and identical to all our previous multigene papers. Within scotokaryotes, Metamonada are strongly a clade on Fig. 1 (0.96% support) and the long-branch Trichozoa did not separate from the short-branch Anaeromonadea as on Fig. S1 and some earlier trees (Cavalier-Smith et al. 2015a, b) . Figure 1 also provides stronger evidence than before that Metamonada are holophyletic. Metamonada grouped with Malawimonas as a moderately supported (0.79) clade, in contrast to several earlier CAT trees (Cavalier-Smith et al. 2015a, b, 2016) . H o w e v e r, e a r l i e r C AT t r e e s i n t h e M e t a m o n a d a / Malawimonas region tended not to converge or were poorly converged. Figure 1 had maximal support for planomonads being the deepest podiate clade followed by a clade comprising Mantamonas and Collodictyon (0.99, much stronger support than the 0.51 found when this clade was first seen: Cavalier-Smith et al. 2014) . Within obazoa, successive branching of breviates, apusomonads and opisthokonts and their internal branching was exactly as before and (except for one apusomonad node) maximally supported. However, a 158-taxon tree excluding Microheliella that was run much longer still did not converge (Fig. S1 max. diff. 1) and showed discrepancies within Rhizaria and in two other regions, suggesting that non-rhizarian inter-chain agreement with 159 taxa was fortuitous. Topology was identical for both chains on the 158-taxon trees except in three local regions with very close deep branches, all having shown similar contradictions amongst previously published trees. The first concerns basal branching within scotokaryotes (unlike Fig. 1 that for the first time in any CAT tree shows a strongly supported metamonad clade); its two deeply divergent subclades do not group together and are contradictorily placed. In chain 1, unlike Fig. 1, only anaeromonads were sister to Malawimonas (0.62; 0.47 in the consensus tree); the long-branch Trichozoa were one node higher as sister to all other scotokaryotes (0.61). Contradictorily in chain 2, anaeromonads were one node deeper (below Malawimonas; 0.49) and Trichozoa one node higher (above Planomonadida 0.46; 0.40 in the consensus tree). Secondly, Picomonas and Telonema were sister on chain 2 (0.89; 0.59 on consensus tree) as in Fig. 1, with this Corbistoma clade (Cavalier-Smith et al. 2015a) being wrongly sister to all other Hacrobia, not just to other Cryptista as correctly in Fig. 1. However, chain 1 put Telonema one node lower as sister to Harosa (0.51; incorrect, see Cavalier-Smith et al. 2015a) and Picomonas higher, as (correctly) sister to other cryptists (0.46). This less good topology for Corbihelia probably stems partly because, as previously found (Cavalier-Smith 2015) , including Microheliella (Fig. 1 with thus three corbihelians) stabilises trees compared with having only two corbihelians [Fig. S4; in Burki et al. (2016) without Microheliella Corbistoma was not a clade]. Thirdly, amongst basal Rhizaria, Ectoreta were sisters to a moderately supported (0.90) classical Cercozoa clade and strongly excluded from Endomyxa (0.99) on chain 2 based on 11,912 trees but are weakly (0.74) sister to Endomyxa on chain 1 based on 13,480 trees (also excluded from Endomyxa quite weakly: 0.77). Within Ectoreta, for Radiozoa, the chains were highly contradictory, one showing it as a clade, the other as paraphyletic both with maximal support. Discussion Sequence information now available places 60 of the 65 highly diverse rhizarian orders (Table 1) with reasonable confidence on the evolutionary tree. Many aspects of their cell evolution are clearer, though three orders still need sequence evidence to verify that they are truly cercozoan: helioflagellate Axomonadida; the silica-scaled filose amoeboid order Zoelucasida; and Perlofilida, filose non-flagellates with silica perles. Two retarian orders also lack sequence evidence, but morphology makes it virtually certain that Acanthoplegmida are Acantharea and likely that Claustrosporida are Ascetosporea. We here not only discuss our results but use them to clarify many aspects of rhizarian phylogeny and cell evolution. Distinctness of Gromiidea and Foraminifera The idea that the highly reticulose Foraminifera might be related to the predominantly filose testate marine amoeba Gromia predates sequencing studies; Gromia was initially considered a foraminiferan despite its pseudopodia being non-granular and filose (with rare anastomoses) not granular and highly reticulose as in Foraminifera. This misconception stemmed largely from confusion with the superficially similar foraminiferan Allogromia (Arnold 1972) and the presence in both of a complex sexual life cycle (Arnold 1966, 1972) — though Gromia gametes are uniciliate not biciliate as in Foraminifera. Later, Gromia was placed in Testaceafilosea (de Saedeleer 1934) a now abandoned polyphyletic group that included all testate amoebae later included in Cercozoa (Cavalier-Smith 1998a; Cavalier-Smith and Chao 2003a) . The first 18S rDNA trees for Gromia strongly grouped it with Cercozoa and showed Cercozoa and Acantharea (one of the three classes of Radiozoa) as sisters (Burki et al. 2002), but were contradictory, one weakly grouping it with Phytomyxea, the other as sister to Phytomyxea plus Filosa, neither with significant support; Foraminifera were excluded from the tree because of their long branches. The idea of a Gromia/foraminiferan relationship was revived when single-gene trees including both sometimes grouped them together or nearby (18S rDNA, Berney and Pawlowski 2003; RNA polymerase II, Longet et al. 2003, 2004; Longet and Pawlowski 2007; actin, Longet et al. 2004; Flakowski et al. 2005, 20 06) . Tekle et al. (200 7) fou nd the long- branc h Foraminifera and Ascetosporea (parasites of invertebrates) grouping together by rDNA on site-homogeneous Bayesian trees (although not with ML, wrongly stated to be congruent), but on actin trees Foraminifera were sisters to Filoreta (= Corallomyxa) tenera (an ATCC culture earlier misidentified as Corallomyxa, later named Filoreta: Bass et al. 2009a) not to Gromia. More comprehensive cercozoan 18S rDNA trees showed Gromia as sister to the marine Ascetosporea rather than to Phytomyxea (Bass et al. 2012; Cavalier-Smith et al. 2008, 2009) and also showed a clade Endomyxa with weak support (Bass et al. 2009a) and a distinct classical Retaria clade (Acantharea, Polycystinea, Foraminifera) separate from Endomyxa and other Cercozoa. Our present ML and siteheterogeneous CAT rDNA trees where Ectoreta is represented only by the shorter branch Radiozoa (long-branch Foraminifera excluded to prevent artefacts) all grouped Gromia and Filoreta with Ascetosporea, never with Ectoreta or Cercozoa. Neither our nor any other gene-rich multiprotein trees group Gromia with foraminifera; all robustly exclude it from Ectoreta and place it nearer Filoreta, usually as sister. The idea that Gromia is specifically related to Foraminifera is firmly refuted. Demarcation and relationship between phyla Cercozoa and Retaria Multigene analyses of Rhizaria have also been contradictory. The 167-gene tree (36,735 amino acids) of Burki et al. (2010) excluded Ectoreta (2 Foraminifera, 2 Acantharea) from classical Cercozoa with maximal support by ML LG and MrBayes WAG analyses, and showed Endomyxa (just Gromia and two Phytomyxea) as a clade with 99% and 1.0 support. Contradictorily, their PhyloBayes CAT (not the evolutionarily better, computationally more onerous GTR-CAT used here) placed Ectoreta as sisters to Gromia with 0.92 support, but support for the Endomyxa/Retaria clade was neglible (0.51). When Reticulomyxa, the ectoretan with the highest gene representation, was omitted (foraminifera represented only by Quinqueloculina with less than 32 genes or omitted altogether), ML and CAT both put Ectoreta (or Acantharea alone) as strongly sister to Gromia and support for an Acantharea/Endomyxa clade also increased. A 36gene analysis of Harosa (with only haptophytes as outgroup) with 4 Endomyxa including 12 genes from Filoreta (under the wrong name Corallomyxa) and 22 Rhizaria (Sierra et al. 2013) placed Ectoreta as sister to Filoreta plus Gromia, with moderate support ML LGF (78%) and 0.99 by CAT-Poisson (chains did not converge; it is unclear whether their statement applied to both CAT-Poisson and CAT-GTR analyses). However, sampling was low (5 Cercozoa, 4 Endomyxa) and the presumably unconverged CAT 36-gene tree peculiar in that plasmodiophorids were sister to all other Rhizaria (not found on the ML tree not shown, or on their CAT-Poisson 109-gene Fig. S1 tree, that was more sensible for Cercozoa). This oddity probably stems mainly from having only nine genes for one and seven for the other plasmodiophorids (only three shared by both). Brown et al. (2012) using 159 proteins (43,615 amino acids), 11 Rhizaria but only 3 Endomyxa also found Radiozoa as sisters of Gromia with maximal Bayesian (CAT Poisson) and 52% ML LG support. Retaria were also sisters of Gromia on the 238-gene (50,293 amino acids) tree of Burki et al. (2012) but this included only 5 Rhizaria, all except Bigelowiella with 68–81% of amino acids missing. A recent analysis (Sierra et al. 2016) with improved endomyxan gene and taxon sampling removes some of these problems and put Ectoreta as sister to the Ascetosporea/ Gromiidea clade, agreeing with our conclusion that they are not specifically related to Gromiidea. Rhizarian multigene phylogeny has been plagued by gene sampling for Endomyxa and Ectoreta being generally low with much missing data, as well as by long branches of Retaria on sequence trees. Except for that study and a 119-protein tree (23,162 amino acids) including Mikrocytos (Burki et al. 2013), whose branch is so immensely long that one cannot rule out the possibility of tree distortion by attraction towards other long ones like Foraminifera, our study is the first to include an endomyxan with a fairly high gene representation (Filoreta marina: 26,479 amino acids). This may have stabilised some trees and helped to exclude Ectoreta; for Ectoreta, only Reticulomyxa has high gene sampling (35,136 amino acids). Having a second filosan, Mataza, with high gene representation (39,490 amino acids, albeit less than the 45,750 of Bigelowiella) will have helped stabilise our trees, as will now having 14 Cercozoa (only 3–5 on previous rhizarian multigene trees). Though a minority of our CAT trees had conflicting topologies, we now conclude that Ectoreta are most likely sister to Endomyxa as a whole, not to Gromia only or Filoreta only, nor to Gromia plus Filoreta, as some previous trees suggested—though Sierra et al. (2016) found paraphyletic Endomyxa by four site-homogeneous methods, with site-heterogeneous PhyloBayes CAT-GTR (seemingly not the better CAT-GTR-GAMMA we used) support for that was insignificant 0.5; thus, that tree does not significantly contradict our conclusion. Krabberød et al. (2017) have now added two important ectoretans to a 255-protein tree, a nassellarian radiozoan and Sticholonche that has a unique cell organisation, and increased gene sampling for foraminifera. Their analysis, though much less well sampled than ours for monadofilosan Cercozoa and Endomyxa, has maximal support for Rhizaria, Ectoreta, Endomyxa, Retaria and Cercozoa Auranticordis suspected to harbour cyanobacterial endosymbionts (Chantangsi et al. 2008) needs further study to see if they are integrated or dispensible. As Cavalier-Smith (2018) explained, origin of ciliary gliding by Cercozoa may have favoured the conversion of the ancestral bypassing microtubule band of Chromista into ventral posterior ‘root’ vpr2, which is probably not really a centriolar root as it bypasses both centrioles and is probably not nucleated by either. It would have helped strengthen the ventral groove alongside the gliding posterior cilium which has to support the cell body during gliding. As Cavalier-Smith (2018) also explained, axopodia of desmothoracid Cercozoa and of Radiozoa and Sticholonche may have evolved from the ancestral chromist bypassing band that is present relatively unchanged in many Halvaria. If that band is not nucleated by centrioles, it could have been preadapted for nuclear nucleation as in Radiozoa and Sticholonche, thus the ancestral condition for Ectoreta. Non-reliance on centrioles for axopodial nucleation is compatible with the complete loss of centrioles and cilia in vegetative cells of Retaria and thus enables an aciliate axopodial lifestyle. Position of Rhizaria within Harosa and Chromista Our trees strongly support Halvaria being a clade, and thus establish a robust deep phylogeny for chromist subkingdom Harosa: Rhizaria are always sister to Halvaria. Burki et al. (2016) using 250 proteins and 55,554 amino acids (excluding all positions missing in more than 20% of taxa) and 148 or 150 eukaryote-wide taxa likewise found maximal support for this topology on site-heterogeneous trees. Their ML trees showed the same topology with weak support using 55,554 positions. When only the most conservative positions were used (36,507 amino acids), Rhizaria were sisters of heterokonts only with strong support by both methods (a contradiction not mentioned in their text). Oddly, Sierra et al. (2016) using 229 proteins (64,107) amino acids, but only 56 taxa contradictorily put Rhizaria instead as sister to Alveolata on ML trees. Neither of these alternative topologies for Harosa was found in any of our other multigene papers (Cavalier-Smith et al. 2014, 2015a, b) . A likely reason for these major discrepancies is the extremely high fraction of missing data in all Ectoreta (noted above) that dominated the relatively small taxon sample of Sierra et al. (2016) and high missing data in most Rhizaria. No Ectoreta had less than 56% missing positions, and all but 2 has at least 70% missing; Bigelowiella and Plasmodiophora were the only rhizarians with less than 20% missing and only 19 of the 56 taxa met the criterion of < 20% used by Burki et al. to exclude weakly sampled taxa. Even some species with complete genomes were surprisingly undersampled (Arabidopsis 43% missing; Toxoplasma 25% missing, Blastocystis 27% missing, Plasmodium 24% missing). Sierra et al. (2016) and Burki et al. (2016) did not include the foraminifera Sorites or Quinqueloculina, two of the three genetically most richly sampled ectoretans in our analyses. As nine of the ten new rhizarian sequences added here are genetically more richly sampled than the vast majority of Rhizaria in Sierra et al. (2016) , it is unsurprising that we found the same topology as did the most sequence-rich trees of Burki et al. (2016) , who carefully excluded grossly undersampled taxa which can cause serious systematic errors in tree reconstruction (Roure et al. 2013) . A second problem with the Sierra et al. (2016) analysis is arbitrary and suboptimal outgroup choice: using only Viridiplantae as outgroup for Harosa was unwise, as they are not the closest outgroup (Hacrobia are, as used in our Figs. 2 and 3) and also have much longer branches than Glaucophyta that we therefore used instead to represent Plantae in our eukaryote-wide trees. In general, it is safest to use a balanced spread of diverse outgroups for rooting trees, as a single arbitrary one (far too often used) could by chance have features that pull it to the wrong position in the ingroup, especially if it is distant or rapidly or oddly evolving, and especially if the ingroup is genetically undersampled. That clearly happened in Sierra et al. (2016) if compared with other recent studies using eukaryote-wide outgroups that all found monophyletic Halvaria by both CAT and ML (Burki et al. 2012, 2013; Cavalier-Smith et al. 2014, 2015a, b, 2016) . Sierra et al. (2016) also used PhyloBayes CAT, less prone to such artefacts; that also strongly gave the same discordant phylogeny when Mikrocytos was excluded but did not significantly support it in the presence of Mikrocytos. Grant and Katz (2014) using 232 eukaryotes and the 150 most evenly sampled genes across these taxa also found the Halvaria clade by ML though with under 80% support. However, they found the same discordant Alveolata/Rhizaria topology when using 207 genes with a mere 17,220 positions for eukaryotes plus prokaryotes, but the basal branching of eukaryotes in this 3-domain tree is obviously grossly distorted by long-branch artefacts; using ML alone rather than CAT for 3domain trees is technically inadequate (Lasek-Nesselquist and Gogarten 2013; Raymann et al. 2015) and cannot be considered reliable. Nonetheless, their eukaryote-only ML tree that agrees with ours and Burki et al. (2012, 2013, and most trees in 2016) for harosan deep phylogeny confirms the importance of good gene sampling and a broad and sensibly chosen set of outgroups and the reproducibility of clade Halvaria in the best analyses. Grouping Rhizaria with Alveolata alone by Sierra et al. (2016) is therefore no more correct than the first ML-only multigene tree that clearly showed the harosan clade but incorrectly put Rhizaria as sister to Heterokonta (Burki et al. 2007) . A multiprotein phylogenetic study of the imbricate cercozoan filose amoeba Euglypha found maximal support for clade Halvaria using site-heterogeneous trees for 147 proteins (He et al. 2016) in agreement with all our analyses; but unlike ours, their ML tree contradictorily grouped Rhizaria with Alveolata— as by both methods did genically sparse trees restricted to 34 or only 27 proteins. They claimed that their sparse 34- and 27-gene trees were correct and that their genically well-represented 147gene site-heterogeneous tree was wrong and distorted by a long branch artefact, despite recognising that in theory it should be more accurate than the corresponding ML tree. This belief was based on the assumption that their 27-protein set had selectively removed genes most likely to cause artefactual grouping of Halvaria. However, the method aimed at that goal evidently failed to achieve it, as their 27-protein tree (their Fig. 1) shows Rhizaria and alveolates both with systematically substantially longer branches than any other groups; therefore, this tree itself may suffer from a long-branch attraction artefact directly between these two groups; we therefore suggest that their unusual grouping was indeed a long-branch artefact and that trees using more genes [like ours and even better protein sampling by Krabberød et al. (2017) ] are in general more reliable. Their 27-protein tree was also based on a very different protein set from their 147-protein tree and our 187-protein trees and most other published multiprotein trees. Of the 27 included proteins, 19 were from genes of eubacterial origin (He et al. 2014) , and thus were not genes from the eukaryotic host that enslaved mitochondria as are our 187 proteins—their 34 protein trees included 14 eubacterial-origin proteins. As their selection method certainly did not remove proteins whose tree branches are systematically longer in alveolates and rhizaria than in heterokonts, it obviously failed to remove proteins with systematically above average evolutionary rates in some but not all harosan groups as they had erroneously expected. It may even have introduced a long-branch problem by eliminating numerous phylogenetically informative genes, and retaining a small residue of systematically biased ones. The protein selection method used by He et al. (2016) should not have been expected to eliminate such biases because to obtain the 27-protein set it selectively removed proteins for which within one of the three harosan groups at least three species had branches at least twice as long as average for that subgroup on single-protein trees. Thus they did not remove proteins with a systematic bias in branch lengths between groups (as would have been necessary to reduce the supposed problem of biases between groups) but proteins that had more variable rates within just one subgroup, which is irrelevant to the question of systematic bias amongst groups. Many such genes could have had a lower average rate of evolution than genes not removed, but would have been wrongly removed just because some non-representative lineages had fast rates. Thus the claim that the 27-protein trees were technically superior was based on a fallacy and a confusion between systematic cross-group variation in evolutionary rates and random within group variation. Conversely, their method would have retained genes for which all genes within a group were systematically biased by extra rapid evolution, which potentially could cause long-branch attraction, but would have eliminated many proteins where the majority within a group were well-behaved short branches just because some branches were longer—both the exact opposite of what the method was supposed to achieve. Fast site removal by Krabberød et al. (2017) , a more reliable way of detecting long-branch problems, showed the exact opposite of what He et al. (2016) concluded, i.e. that the ML trees are more affected by rapid evolutionary rates than CAT trees and thus less, not more, reliable. As explained above, we found that much of He et al. (2016) raw data for Rhizaria included contaminants. He et al. (2016) recognised that, but instead of just removing protein sequences for particular species recognisable as reasonably unambiguous contaminants, as we did, they generated a 34-protein alignment by removing all proteins from the alignment where more than one species did not group with other Rhizaria. That was unwise as it must have removed large numbers of genes from the analysis for both rhizarian and non-rhizarian taxa for which individually there was no reason whatever to suspect contamination. In our experience, it is often the case that single-gene trees do not show monophyly of groups that from other evidence are known to be monophyletic, even in samples where possibility of contamination can be excluded; that apparent polyphyly is probably just because many genes (e.g. ribosomal protein genes) are too short or erratically variable in evolutionary rates to retain enough information to place all branches accurately. Their excessively loose criterion for identifying ‘contamination’ would inevitably have excluded vast numbers of sequences that were not contaminants and left some that were; the authors appear to have had far too much faith in the accuracy of single-protein trees, and as a result discarded almost all the probably useful data and in all likelihood severely degraded the phylogenetic signal. It was illogical to remove all the genes from other taxa just because two were considered to be misplaced; they ought to have removed just the individually suspect genes. It was equally illogical to remove genes where two rhizarian lineages failed to group appropriately but to retain those where only one sequence was misplaced, because contamination affects samples individually; therefore, one might expect it to be more likely that a single outlier will be a contaminant than a group of two, which might reflect a different kind of error. If He et al. (2016) really believed that all proteins thus (mis)identified as contaminants really did include potentially misleading contaminants, they ought to have excluded them from the 27-protein set, but curiously did not. The 27 proteins included 14 (a majority) that they claimed to have been contaminants and ought to have excluded; that would have left only 13 proteins for analysis, so few to have been useless. In sum, the authors’ claim to have identified and eliminated a long-branch artefact affecting harosan deep phylogeny is seriously flawed. To support their entirely unwarranted conclusion, He et al. (2016) asserted that the numerous previous studies that contradicted it ‘mostly either lack substantial data for Rhizaria ... and/or have only extremely long rhizarian branches to work with’. Neither criticism is applicable to our present study, which adds massive data for nine rhizarians (eight cercozoan), every one of which has shorter branches than any Ectoreta; the majority have shorter branches than in alveolates and several have shorter branches than in many heterokonts. In number of genes and taxa, the shorter-branch Cercozoa and Endomyxa now dominate our trees compared with the more sparsely sampled longer branch Ectoreta. Our trees for 56 Harosa using 187 proteins are therefore much more convincing than theirs for just 26 Harosa using only 27 or 34 proteins selected for entirely erroneous reasons. It is now clear that neither cercozoan nor rhizarian proteins are systematically fast-evolving compared with those of Halvaria; of the four rhizarian subphyla, only Ectoreta has systematically longer branches than harosans generally, which can cause longbranch artefacts on sparsely sampled trees such as those of He et al. (2016) that are deficient in cercozoan data. Those sparse trees included only nine rhizarian branches and only three for Cercozoa, as many taxa were lumped as in silico composites, and therefore told us nothing more than had 18S rDNA trees about the position of Euglypha, the initial target of their study; they merely confirmed that Euglypha is closer to Paracercomonas and Aulacantha than to Bigelowiella, which our trees with 14 Cercozoa sensu stricto show much more decisively using Paulinella as the euglyphid representative. As rDNA trees had already indicated that Euglypha is probably phylogenetically closer to Aulacantha than to Paracercomonas (Scoble and Cavalier-Smith 2014) , lumping the two latter as a single false taxon was phylogenetically illogical and would necessarily have prevented establishing Euglypha’s true position. Fast site removal by Burki et al. (2016) was theoretically sounder than the He et al. method and yielded a contradictory tree with Rhizaria sister to heterokonts, but their tree was far less well sampled than ours for Rhizaria (6 versus 31 species). We deliberately omitted the longest branch alveolates to reduce artefacts, but the Burki et al. tree included three alveolates with much longer branches than the other two harosan groups, which might have led to artefactual long-branch exclusion of alveolates in their shorter-sequence trees causing Rhizaria to group wrongly with the relatively short-branch heterokonts; their choice of proteins also gave greater inter-group branch length disparity than ours. Further evidence for monophyly of Halvaria and against the interpretations of Sierra et al. (2016) and He et al. (2016) comes from Krabberød et al. (2017) who enriched ectoretan taxon sampling by adding a nassellarian and Sticholonche and extra foraminiferan genes (Sorites still omittted) to reduce missing data and whose 255-protein CAT tree has maximal support for a halvarian clade. Because their ML LG tree contradictorily grouped Rhizaria with alveolates as in the sparse trees of Sierra et al. (2016) , Krabberød et al. removed the four fastest evolving rate clases of amino acid sites and found that support for this almost certainly incorrect topology dropped from 96 to 50%. That suggests that rapidly evolving sites misleadingly enhance the incorrrect topology seen on some ML trees. Perhaps the main reason why Krabberød et al. (2017) were unable to eliminate the ML likely artefactual grouping of rhizaria with alveolates entirely may have been serious undersampling of shorter-branch Cercozoa and Endomyxa on their tree compared with ours, and possibly their including Aulacantha contaminants [earlier, we found that its raw data set from Sierra et al. (2013) included many alveolate genes (Cavalier-Smith et al. 2015a) ] which presumably made Aulacantha (one of their only two cercozoa) a much longer branch than on our trees. Finally, the 351protein tree of Brown et al. (2018) has maximal support by ML (two methods) and CAT for clade Halvaria. These authors took great care to test effects of removing fast-evolving sites by reliable methods, not those of He et al. Agreement of the well-sampled CAT trees of Krabberød et al. (2017) for 24 Rhizaria and genically richest ones of Brown et al. (2018) with ours (31 Rhizaria: 14 Cercozoa, 17 Retaria) strongly favours Rhizaria being sister to Halvaria, and argues against both contradictory results from the less well-sampled analyses of Burki et al. (2016) , Sierra et al. (2016) and He et al. (2016) . Previously, we suggested that in corticate multigene trees, the deepest branching within kingdoms Chromista and Plantae may be mutually distorted when red algae are included (Cavalier-Smith et al. 2015a) . The evidence for mutual interkingdom distortion is that basal topology of each kingdom differs when the other is included or excluded without any other changes in taxon or gene sampling or analytic method (Cavalier-Smith et al. 2015a) . In the absence of chromists Plantae are a robust clade, with Viridiplantae strongly sisters of Rhodophyta, and Glaucophyta the deepest branch, as all other evidence indicates is evolutionarily correct; when chromists are added, Rhodophyta are commonly pulled away from Viridiplantae to become the deepest branch of Plantae with some or all Hacrobia often intruding into Plantae or becoming their apparent sisters or ancestors. Conversely, in the absence of Plantae, Chromista are a robust clade with Hacrobia a clade and sister to Harosa; but when Plantae including Rhodophyta are added, they intrude into or become mixed with it. This mutual topological distortion is a fact irrespective of how it can be explained (Cavalier-Smith et al. 2015a) . We suggested it is caused by chromists being chimaeras of two eukaryotes, one a red alga, coupled with the practical impossibility of distinguishing all chromist genes as being of host or red algal origin, because the host and red algae diverged so closely in time and so long ago that phylogenetic signal, which in theory might allow their identification, has been either erased or become so faint as to prevent their accurate placement on single-gene trees (CavalierSmith et al. 2015a) . Figure 4 is a novel test of this explanation of the mutual distortion phenomenon. In addition to testing the position of the possibly mixed ‘Minchinia’, we wanted to see if we represent Plantae solely by the shortest branch Glaucophyta, whether Chromista and Hacrobia remain holophyletic as they are in the complete absence of Plantae. The answer is that on one CAT chain (Fig. 4), Chromista was a maximally supported clade, and (except for exclusion of the single undersampled, highly divergent species Telonema subtilis that lacks close relatives to stabilise its position) Hacrobia is a strongly supported (0.98) clade; but on the other chain, Glaucophyta intrude into Chromista and group with Hacrobia (only 0.64 support) and even more weakly group with core Cryptista (0.54). Burki et al. (2016) agree that relatively poorly sampled Telonema and Picomonas single-species lineages are currently hard to place accurately on trees and excluded them from some of theirs—though their excluding the third corbihelian Microheliella (Cavalier-Smith et al. 2015a, genically better sampled than Picomonas) would hinder correct placement of this cryptist subphylum. Together, our Figs. 1 and 4 completely refute the criticism by Burki et al. (2016) that we get wellsupported holophyletic Hacrobia only when ‘a large part of the diversity’ is removed. These 159–162 taxon trees have more deep eukaryote diversity than the 150-taxon tree of Burki et al. (2016) , differing in removing just one clade (Plantae in Fig. 1 and Rhodophyta/Viridiplantae in Fig. 4). In marked contrast to Burki et al. (2016) whose trees all included both red algae and chromists, we found no convincing evidence for the deeper branching of Cryptista compared with other chromists as their trees suggested (some with insignificant and some with strong support). Their trees were mutually contradictory with respect to the exact position of Cryptista. Their full data set split them into three with Telonema grouping with other chromists as in our Fig. 1, core Cryptista being sisters of the probably false Viridiplantae/Glaucophyta clade (by ML and 2 CAT chains but sister to holophyletic Plantae on chain 3) and Picomonas sister to Rhodophyta. When Telonema and Picomonas are excluded, ML still groups core Cryptista with Viridiplantae/Glaucophyta, but only one of the three CAT chains did so the other two showing clade Plantae with core Cryptista their sisters. When faster sites were removed, ML still showed the same Viridiplantae/Glaucophyta pseudoclade (without support) with Cryptista their sisters (no support by one ML method) as does one CAT chain, the other showing holophyletic Plantae. As site-heterogeneous trees are evolutionarily more realistic, the Burki et al. (2016) results on balance argue that Plantae are a clade and intrusion of Cryptista into Plantae on some of their trees (and one of our Fig. 4 chains) is phylogenetically misleading. Neither we nor they have any compelling evidence against holophyly of Plantae. All their trees agree in showing a Viridiplantae/Glaucophyta clade that is absent when Chromista are excluded, and thus probably arteficial. Interestingly, their earlier trees using slightly more (258) genes but only 68 eukaryotes did not show that clade (Burki et al. 2013) and strongly support (both ML and site-heterogeneous trees) the Viridiplantae/Rhodophyta clade that we found when chromists were omitted and distortion was impossible (Cavalier-Smith et al. 2015a) . We do not understand why their two papers are so strongly contradictory for basal topology of Plantae, but note that for ML the 2016 paper used IQ-TREE and bootstrap approximation methods not yet widely used, not RAxML as in 2013, and the 2016 CAT trees did not converge. Until reasons for this major discrepancy are understood, we should be cautious and not overconfident in either. Possibly, they exemplify our argument that observed distortion effects are expected to be highly sensitive to taxon and gene sampling (Cavalier-Smith et al. 2015a) and the principle that resolution is expected to be extremely difficult and disturbable by the slightest biases (that can so easily add up in multigene trees to give strong support for the wrong tree) when several taxa diverge almost simultaneously, as is so for Hacrobia and Plantae. Nonethless, our previous and present work establish that adding chromists to the tree distorts the branching order of Plantae, often pulling red algae away from Viridiplantae to the base in multigene trees (e.g. Burki et al. 2007, 2009, 2016) , though not invariably (e.g. Burki et al. 2012, 2013; Cavalier-Smith et al. 2014, all with the arguably correct Rhodophyta/Viridiplantae clade) . It is important for kingdom exclusion tests (as in Cavalier-Smith et al. 2015a) to be repeated with the Burki et al. (2016) data set and algorithms. If they omit all chromists, we predict topology of Plantae would change to move red algae one node higher to become sister of Viridiplantae; conversely, omitting all Plantae should produce a Hacrobia clade. Only if neither happened and wellsupported Harosa/Haptista and Viridiplantae/Glaucophyta clades remained could their belief that the chimaeric chromists are not distorting the topology of Plantae on their trees and red algae not distorting that of chromists be defensible. The fact that all genes in their dataset showed ‘no strong affinity between Cryptista and red algae’ does not, as Burki et al. (2016) mistakenly assert, ‘suggest that this relationship [i.e. core Cryptista/Plantae] is a reflection of vertical inheritance rather than owing to a cryptic contamination of endosymbiont genes’. As Cavalier-Smith et al. (2015a) more fully explained, one cannot reasonably expect single-gene trees (mostly entirely unresolved at the base of corticates) to show such a relationship strongly, because basal divergence times within Hacrobia and Plantae were so close together in time and so long ago. Single-gene trees are good for detecting distant contaminants or recent lateral transfers, but not for discriminating amongst groups that diverged nearly simultaneously around 750 My ago. The problem with the founding chromist symbiogenesis is that host and symbiont were so closely related at the time. This is entirely unlike green algal plastid enslavement by euglenoids long after these groups diverged or the symbiogenetic origins of plastids and mitochondria where the symbionts were even more distantly related to the hosts, so single-gene trees easily strongly resolve their foreignness. Furthermore, the assertion that if Cryptista have such contaminants one would expect all other chimaeric chromists to have many such genes and all to be equally attracted to red algae (Sierra et al. 2016) was also erroneous and refuted in detail previously (see Cavalier-Smith et al. 2015a) ; the key points are (a) that independent losses in diverging lineages need not be quantitatively similar and (b) the more rapid protein evolution of Harosa will erode the phylogenetic signal for any that remain much more than in slower evolving Hacrobia. These considerations should not continue to be ignored; doing so gives excessive faith in the reliability of basal branching orders on corticate multigene trees. It is imperative to test these distortions further and reach a consensus explanation for them. As they did no reciprocal kingdom removal tests, Burki et al. (2016) provided no convincing disproof of the likely holophyly of Chromista and Hacrobia. However, they provided further evidence for holophyly of Haptista (contrary to Burki et al. 2009, 2012) and that Haptista are related to other chromists (i.e. subkingdom Harosa), not to Plantae. Our present and previous taxon exclusion analyses (Cavalier-Smith et al. 2015a) provide strong evidence that the grouping of core Cryptista with Plantae in Burki et al. (2012, 2016) and the branching order of the three Plantae clades in Burki et al. (2016) only are both phylogenetically seriously misleading. In contrast, Krabberød et al. (2017) with 255 genes (almost as many as Burki et al. 2016 and slightly more amino acid positions) found holophyletic Plantae without Cryptista intruding and with the expected branching order by both CAT and ML (more strongly supported than the contradictory Burki et al. results) without excluding any major chromist lineages except Corbihelia and Centrohelia (included by us and some by Burki et al.). Perhaps when these groups’ taxa and genes are better sampled, the effects of potentially misleading sequences and undersampling will be reduced enough to recover holophyly of Plantae, Hacrobia and Chromista more consistently. Whatever the reasons for the marked contradictions between their trees and those of Burki et al. (2016) with respect to plant holophyly and basal branching, they reinforce our thesis that effects of different taxon and gene sampling are widely underestimated and poorly understood, and more caution is needed in using the latest multigene tree with extremely short basal branches for both Plantae and a Plantae/Cryptista (likely pseudo)clade (see especially Krabberød et al. 2017) as an excuse for rejecting well-corroborated ideas like monophyly of Plantae, Hacrobia and Chromista (for Chromista, see discussion in CavalierSmith 2018) . Brown et al. (2018) using 351 proteins found 2 contradictory CAT consensus tree pairs for 61 eukaryotes both with every node maximally resolved: in 1 Hacrobia is a maximally supported clade, in the other it is maximally supported as polyphyletic with Cryptista wrongly branching within Plantae, which ultrastructure, periplastid derlin evolution and the shared hacrobian lateral gene transfer all strongly contradict (Cavalier-Smith 2018) . This decisively shows that a maximally supported bipartition on a CAT multiprotein tree can sometimes be wrong and that sometimes synthetic cladistic arguments can be more reliable than maximally supported 351-protein trees. Contrary to Janouškovec et al. (2017) , ‘phylogenomics’ (an often misleading catchphrase; their study was not genomic but partial transcriptomic) is sometimes less reliable and less decisive than such cladistic reasoning. Exclusion of Cryptista from Chromista as first done by Cavalier-Smith (1995, giving too much emphasis to poorly resolved 18S rDNA trees) is not justified by present evidence. Probable holophyly of Metamonada and scotokaryote basal branching order Figures 1 and 4 are the most broadly sampled eukaryote-wide trees to date using a CAT-GTR-Γ model. Though Metamonada are not our main focus, Fig. 1 provides the first strong site-heterogeneous tree evidence for monophyly of Metamonada (0.96) and more weakly suggests they are sister to Malawimonas. This confirms previous evidence for excavate paraphyly (Cavalier-Smith et al. 2014, 2015a) and supports the view that failure of both metamonad clades to group together on previous CAT trees (and our 158- and 162-taxon trees) is more likely statistical artefact than true phylogeny (Cavalier-Smith et al. 2015a) . Often low support for conflicting positions of the metamonad clades makes more extensive data necessary to confirm this. All present trees support the conclusion that the putative Malawimonas/metamonad clade is sister to podiates, forming a scotokaryote clade, and does not belong in Eozoa with Eolouka, Percolozoa and Euglenozoa (Cavalier-Smith et al. 2015a) . They also confirm that Mantamonas and Collodictyon are related and planomonads branch more deeply than any other Sulcozoa (Cavalier-Smith et al. 2015a) , both initially less clear (Cavalier-Smith et al. 2014) and both independently strongly confirmed by Brown et al. (2018) with 351 proteins. All previous CAT trees including Malawimonas, Anaeromonadea and Trichozoa showed them as the deepest scotokaryote branches but their relative branching order varied. Brown et al. (2013) first saw a clade comprising Malawimonas and metamonads (represented solely by T r i m a s t i x ) . A n a e r o m o n a d e a ( T r i m a s t i x , Monocercomonoides) alone were sisters of Malawimonas with Trichozoa one node higher in Cavalier-Smith et al. (2015b; Fig. 3 of 2015a, Fig. 2 showing all Metamonada and Figs. 6 and 2 of 2014 both included metamonads one node below Malawimonas) . Support for holophyly of Malawimonas plus Metamonada, though only 0.79 on Fig. 1, i s h i g h e r t h a n p r e v i o u s l y. P r e v i o u s l y, N e o l o u k a (Malawimonas only) and Metamonada were cautiously treated as distinct phyla (Cavalier-Smith et al. 2015a) , but as they are not morphologically distinct enough for separate phylum rank [as Cavalier-Smith (2013b) explained], we hereby formally reverse this and retain Metamonada and Neolouka as subphyla of phylum Loukozoa (as in Cavalier-Smith 2013b p. 121) but Loukozoa is here revised by excluding former subphylum Eolouka (Jakobea and Tsukubamonas) and raising it to phylum rank, with Eolouka Cavalier-Smith phyl. n. having the same diagnosis and composition as subphylum Eolouka (Cavalier-Smith 2013b p. 121) . This formalises revisions that Cavalier-Smith (2018) informally made when redefining neokaryotes as comprising scotokaryotes, Neozoa and Corticata, and more precisely defining excavates as a purely informal group (organisational grade, not clade) comprising only Loukozoa and Jakobea. Thus Loukozoa (in protozoan subkingdom Neozoa) are now split into only two morphologically and phylogenetically distinct subphyla, whereas Eolouka is one of only three phyla of protozoan subkingdom Eozoa, now more narrowly defined than in Ruggiero et al. (2015) in recognition of thus-revised Loukozoa being probably the deepest branch of the scotokaryote clade and thus sister to podiates. The centriole/ciliary transition zone junction in Rhizaria The ciliary transition zone (tz) is distinctively different in Rhizaria from other eukaryotes but varies significantly in length and detailed structure amongst lineages (CavalierSmith et al. 2008, 2009; Cavalier-Smith and Karpov 2012b; Hess and Melkonian 2014) . Cavalier-Smith et al. (2008, 2009) discovered a highly distinctive hub-lattice structure in Sainouron and Helkesimastix at the base of the tz just distal to where centriolar triplets start and argued that similar structures can be seen less distinctly in all other Rhizaria where this region is not completely obscured by dense material and argued that it is a synapomorphy for Rhizaria. Hess and Melkonian (2014) demonstrated an asymmetric acorn/Y-filament complex at the extreme distal end of the centriole of the glissomonad Viridiraptor indistinguishable from that discovered in Chlamydomonas and other green algae, and previously certainly otherwise known only in the metamonad Pseudotrichonympha and chytrid fungus Phlyctochytrium (Geimer and Melkonian 2004) . Geimer and Melkonian (2004) proposed that the acorn/Y-filament complex is present in all centrioles, giving them the asymmetry essential for morphogenesis of non-radially symmetric centriolar roots and the ciliary axoneme, but has been overlooked in most organisms because its very thin, tenuous filaments are generally obscured by other structures. However, if Metamonada are scotokaryotes, and if the eu karyote tree is correctly rooted within Eozoa (whether as in Figs. 1 and 4 or between Tsukubamonas and all other eukaryotes as Cavalier-Smith (2018) suggested, or between jakobids and other eukaryotes as mitochondrial genomes might suggest), then we cannot take the Pseudotrichonympha acorn/Yfilament as evidence for its universality in eukaryotes. This is because it has not yet been found in any authentic Eozoa— hypothetically, it might be absent and evolved only in the common ancestor of scotokaryotes and corticates, though the idea that it is the universal determinant of centriolar asymmetry originating in the ancestral centriole (Cavalier-Smith 2014) is more attractive. What is clear is that an acorn/Y is probably universal in scotokaryotes, corticates and specifically Rhizaria. Though the acorn-Y overall is asymmetric, a granule at the Y’s fork is at the centre of rotational symmetry, immediately proximal to the hub of the hub-lattice structure in Helkesida and arguably all Rhizaria (but see caveat in Hess and Melkonian 2014) . We suggest that in Rhizaria, this Ygranule is the binding/nucleation site for hub assembly of the Rhizaria-specific hub-lattice complex at the onset of ciliary growth; if so, this universal eukaryotic distal centriolar structure preadapted the rhizarian ancestor for evolving the unique hub-lattice structure. All four non-alveolate corticate groups evolved their own specialised and contrasting ciliary tz structures ancestrally absent in simpler transition zones of alveolates, scotokaryotes and Eozoa, e.g. the star-cylinder of Viridiplantae (Cavalier-Smith 1974, 1981a) , transitional helix of Heterokonta (actually just distal to the tz: Hibberd 1979) and double transition plate of many Hacrobia (CavalierSmith et al. 2015a) , whose evolution Cavalier-Smith (2018) discussed, so ciliary changes played a key role in their basal divergences from the corticate last common ancestor that would have had a simple tz like Colponema and ciliates. Alveids confirm rhizarian origin from chromists with cortical alveoli and ciliary vanes Of key importance for understanding the phenotype of the early chromist ancestor of Rhizaria is the new biciliate protist Ancoracysta twista that branches strongly by ML as sister to Haptista on a 201-protein tree for eukaryote ‘host’ genes and as sister to haptophytes on a 36 mitochondrial protein ML tree (that lacked centrohelid data) (Janouškovec et al. 2017). Ancoracysta represents a fourth, previously unrecognised haptist lineage for which we establish new order Alveida and class Alveidea, as it differs from all others sufficiently to merit this. Alveid ultrastructure has a unique combination of characters that greatly illuminate early chromist diversification, but contrary to these authors is not directly relevant to the root of eukaryotes. Unfortunately, the authors misinterpreted aspects of its structure and failed to notice it was not a novel kind of protist, as claimed, but morphologically so close to ‘Colponema’ marisrubri that two of them discovered earlier (Mylnikov and Tikhonenkov 2009) as to be in the same genus. Given this ultrastructural near identity and the fact that rDNA does not group Ancoracysta with authentic Colponema (Janouškovec et al. 2013) or any other alveolates or Harosa, but weakly with the hacrobian biciliate Telonema (Janouškovec et al. 2017), C. marisrubri was undoubtedly misidentified as a Colponema; we rename it Ancoracysta marisrubri and group it with A. twista in new hacrobian family Ancoracystidae. The name Alveida refers to the most striking shared character of both Ancoracysta species: a large inflated cortical alveolus (CA) that underlies the entire cell surface except the ciliary pits and cytopharynx. Unlike all other CAs, its lumen has numerous smooth vesicles and a very extensive dense sheet, somewhat thicker than a cell membrane but (unlike membranes) no trilaminar substructure. The inner face of the alveolar membrane was incorrectly labelled Bplasmalemma^ (pl = plasma membrane, PM) in Janouškovec et al. (2017 Fig. 1 C, E, F); the true PM is the outermost of the two layers labelled envelope (en) in their Fig. 1E, F, the inner of these being the intralumen sheet; it was wrong to lump these completely dissimilar structures together as envelope or theca. The second trilaminar membrane just inside the PM is the outer face of the CA. The authors refer to these layers as four in A. twista but incorrectly as three in A. marisrubri; however, all four are obvious in Mylnikov and Tikhonenkov (2009) Figs. 2 Г, Д and the sheet and some lumenal vesicles are also clear in Figs. 3Д and 3Ж where the CA outer membrane and PM appear as one thicker unresolved line. Disposition of the CA is essentially the same as in haptophytes (discussion and references in Cavalier-Smith et al. 2015a) which differ primarily in lacking the lumenal sheet and vesicles, which corroborates alveid placement in Haptista. Both alveids have a ventral groove with similar lips, which is shallower than that of colponemids and underlain by a curved sheet (A. twista Fig. 1d; A. marisrubri Figs. 3Б, B), dissimilar from colponemid posterior centriolar roots. This transversely C-shaped band (Cb in A. twista) apparently has a substructure of microtubules associated with dense fibrillar material that often obscures them; at the apical end, it is associated both with the cytostome and with smooth cortical and/ or alveolar membranes and apparently enters the zone between the ciliary pits. Thus the anterior extension of Cb is positioned similarly to the axoneme of the haptophyte haptonema and like it is associated with smooth cortical membranes. This organellar complex is therefore an ultrastructurally credible precursor of the haptonemal skeleton and we suggest evolved into it when the groove was lost and the haptonema evolved instead for prey capture, as discussed by Cavalier-Smith et al. (2015a). Our conclusion that Ancoracysta are deep branching Haptista thus greatly strengthens our earlier theory of haptonemal origin and inference that the ancestral haptist had extensive CA (CavalierSmith et al. 2015a) . A third distinctive alveid-shared feature is the lateral vane on the posterior cilium only, which in A. marisrubri points away from the cell surface (Mylnikov and Tikhonenkov 2009) and is thus a ventral vane like those of alveolate Colponema and neoloukan Malawimonas, not dorsal as in jakobid Eozoa. Janouškovec et al. (2017) say that the A. twista vane points into the groove, which would make it a dorsal vane, but none of their micrographs shows that. Perhaps that is a misinterpretation: their Fig. 1U shows it pointing sideways and slightly out from the groove, marginally consistent with it being ventral as in A. marisrubri. Such inferences need care as the posterior cilium can probably twist distally on its axis (see drawing of Colponema loxodes: Mignot and Brugerolle 1975) . Vane morphology and size is indistinguishable between the two Ancoracysta species, both narrower than in Colponema loxodes or vietnamica (Mignot and Brugerolle 1975; Janouškovec et al. 2013 ), but similar to C. aff. loxodes (Myl’nikova and Myl’nikov 2010) . The fourth distinctive alveid-shared feature is their ciliary tz, which is essentially identical with prominent wellseparated proximal and thicker distal plates, but clearly differs from all other protists and radically so from Colponema! As tzs are invariably first class phylogenetic markers, it is puzzling that Janouškovec et al. (2017) did not realise that A. marisrubri cannot be a Colponema and must have been misidentified and be congeneric with A. twista, especially as they mentioned that A. twista cell cortex is more like that of A. marisrubri than any other protist. Prominent double, well-separated, tz transverse plates are found in the majority of hacrobian lineages but not in Colponema or most Myzozoa, so are consistent with alveid grouping with Haptista on ML trees. They also pointed out that A. twista’s extrusome is more like that of A. marisrubri than any other protist! In fact, they are so similar that we see no qualitative difference not attributable to technical artefacts of differential fixation (less good for A. marisrubri). Only their relative proportions differ sufficiently to make the extrusome slightly more elongated and pear shaped in A. marisrubri. The exaggerated claim that A. twista has a novel type of extrusome is wrong. Both species share an extrusome that closely resembles the toxicyst of Colponema loxodes (Mignot and Brugerolle 1975) and C. vietnamica and aff. loxodes in its mid and posterior parts, but as it differs in its anchor-like anterior, its new name ancoracyst is not altogether superfluous, even though the anterior part seems like a larger version of the anterior part of kinetocysts of centrohelid haptists: the ‘ancoracyst’ appears not radically different from kinetocysts of the centrohelid Raphidiophrys (Brugerolle and Mignot 1984) and substantially more like it than Colponema toxicysts. They are much less similar to cryptist extrusomes. Ancoracysta twista ciliary hairs are restricted to the anterior cilium. Those labelled mn on Fig. S1O of A. twista near its base are thin simple hairs like those at the base of some cryptomonad cilia or more extensively in Colponema and many Myzozoa; however, there are hints that thicker possibly tubular ones may be present more distally in Figs. S1O and 1T (Janouškovec et al. 2017). Not noted in A. marisrubri, hairs could easily have been overlooked, but this might be a genuine difference between the species. It is unfortunate that both cultures are lost as it is important to know if tubular hairs are present, which would be the first example in putative Haptista; tubular hairs are found in heterokonts and several cryptist lineages and were postulated as present in the hacrobian ancestor and lost by haptophytes when the haptonema evolved and ventral groove was lost (Cavalier-Smith et al. 2015a) and to have been also lost independently in early Rhizaria and Alveolata. Janouškovec et al. (2017) said mitochondrial cristae are lamellar, but that is misleading as some micrographs showed narrow tubular cristae in cross section (e.g. Fig. S1X); in A. marisrubri, narrow tubular cristate and more lamellar ones are also evident (Mylnikov and Tikhonenkov 2009) . Thus both Ancoracysta species are unusual in having a mixture of tubular and flat cristae. If they are the deepest branching haptist lineage (as multiprotein ML trees for both host- and mitochondrial-origin proteins concordantly show), this odd mixture is unsurprising as haptophyte cristae are tubular and centrohelid ones strongly lamellar. Hacrobian mitochondria generally have undergone more changes in crista shape than other major eukaryote lineages, as Cryptista also include lineages with flattened tubules (cryptomonads) or rounded tubules (others). In size, cell shape and light microscope morphology, position of nucleus and ciliary pits and ciliary length and orientation both Ancoracysta are indistinguishable. Both are marine and from warmer seas, whereas known Colponema species are all from fresh water or soil. Overall, apart from the possible (unproven) absence of hairs in A. marisrubra, the only difference between A. twista and marisrubri is their slight difference in extrusome shape, just enough to justify retaining them as separate species, but not genera. Calling A. twista ‘structurally unique’ was exaggerated. Janouškovec et al. (2017) underplayed the consistent evidence from their ML trees for Ancoracysta being the deepest branching haptist lineage, giving too much emphasis to their non-converged CAT trees that placed them one node lower as sister to Haptista plus Harosa (‘between’ Haptista and Cryptista), and did not even assign it to superkingdom Corticata to which its CA prove it belongs or to Chromista within which it grouped on all their trees. It was misleading to say it did not branch within a recognised lineage: Chromista has been recognised since 1981 and evidence for its holophyly grows stronger and stronger (Cavalier-Smith 2018) . Their trees excluded the deepest branching cryptist group, Corbihelia, so did not strictly demonstrate that Ancoracysta is really a novel lineage distinct from Corbihelia. Though on our best CAT trees Corbihelia was sister to other cryptists, on some other trees some or all of its representatives sometimes move down one node to lie between haptists and cryptists or even become sisters of haptists similarly to Ancoracysta (Cavalier-Smith et al. 2015a) . Nonetheless, on ultrastructural grounds, we do not consider Ancoracysta a corbihelian or other deep branching cryptist, as it is more like Haptista in extrusome structure and having an extensive CA system— only one cryptist spcies, Lateronema antarctica has cortical flat cisternae, which are probably not CA though sometimes suggested to be related (see Cavalier-Smith et al. 2015a) . As Ancoracysta is ultrastructurally and concordantly phylogenetically significantly closer to Haptista than to Cryptista, we classify it within chromist subkingdom Hacrobia and phylum Haptista in a third subphylum Alveidia. Diagnosis of new family Ancoracystidae Cavalier-Smith: Biciliate, uninucleate, heterotrophic chromists with heterodynamic cilia with a long transition zone having widely spaced distal and proximal dense transverse plates; posterior cilium with short lateral vane in mid region associated with shallow-lipped ventral groove supported by promient Cshaped microtubular/microfibrillar sheet; cilia in deep pits; cytopharynx with cytostome opening into anterior pit associated with one or more large extrusomes similar to kinetocysts but with larger anterior anchor-shaped region. All cell surface other than pits and cytopharnyx underlain by largely continuous inflated cortical alveolar sytem whose lumen contains a broad dense fibrillar sheet or sheets and numerous smooth membrane vesicles. Type genus Ancoracysta Janouškovec et al. (2017 electronic supplement p. e1). Other species Ancoracysta marisrubri comb. n. Cavalier-Smith; basionym Colponema marisrubri Mylnikov and Tikhonenkov (2009 p. 6). Diagnoses of new order Alveida Cavalier-Smith, new class Alveidea Cavalier-Smith, and new subphylum Alveidia Cavalier-Smith all as for family Ancoracystidae. Etymol. Alveus L. cavity, referring to CA system, a much more distinctive feature than the extrusome used for genus; ‘alveids’ avoids confusion with alveolates. As alveids belong in Hacrobia, they are as distantly related to Colponema in Harosa as is possible for any chromists. As both retain Malawimonas-like posterior ciliary vanes and ventral feeding groove and have an extensive cortical alveolar system, both characters are arguably ancestral for all chromists, greatly reinforcing earlier arguments that chromists evolved from a vaned Malawimonas-like excavate ancestor by the evolution of CA in the common ancestor of Chromista and Plantae (i.e. the common ancestor of superkingdom Corticata) and that chromist phyla that lack CA (Cercozoa, Cryptista) lost them (Cavalier-Smith 2013a, b; Cavalier-Smith et al. 2015a; Cavalier-Smith 2018) . This makes the idea that Ectoreta retained CA as their central capsule forming alveoli more plausible also. Strong new mitochondrial protein-paralogue evidence for chromist monophyly Discovery that Ancoracysta mitochondria retain 47 αproteobacterial genes (Janouškovec et al. 2017) and the deep-branching scotokaryote sulcozoan Diphylleia kept 51 (Kamikawa et al. 2016) implies that the neokaryote ancestor had at least 55 protein-coding mitochondrial genes, but jakobid Eozoa with 65 have the most gene-rich mitochondrial genomes. The greater primitiveness of jakobids than any neokaryotes is shown most strongly by their being the only eukaryotes that retain eubacterial 4-protein RNA polymerases and SecY. Eozoa also remain the only supergroup that has no known lineages having the neokaryote-specific derived single-molecule cytochrome c haem lyase (holocytochrome c synthase: HCCS) inferred to have evolved in the last common ancestor of corticates and scotokaryotes (Cavalier-Smith 2010a) . This argument from cytochrome c biogenesis for the eukaryote root lying within Eozoa or between Eozoa and neokaryotes (Cavalier-Smith 2010a, b) is as strong as ever. Even then, it was known that some chromist mtDNA (i.e. ciliates) and some plants retain genes for some subunits of the ancestral eubacterial multiprotein cytochrome lyase (Ccm) subunits in addition to the novel HCCS (Allen et al. 2008), so the ancestral neokaryote must have had both mechanisms. Discovery that Ancoracysta has four Ccm proteins and also has HCCS more directly confirms the idea that ancestral neokaryotes had both HCC and multiprotein lyases and that the α-proteobacterial Ccms were differentially lost independently in different neokaryote lineages (Cavalier-Smith 2010a) . Janouškovec et al. (2017) unconvincingly argue that discordances in their single-molecule HCCS tree do not support this, assuming that the tree is dominated by lateral gene transfer (LGT). The tree does have two plausible examples of LGT: one from a chromist to the apusomonad scotokaryote Thecamonas trahens and one from a chromist to the percolozoan eozoan Percolomonas cosmopolitus. But setting these aside, the rest of the HCCS tree is consistent, given its inevitably poor basal resolution, with vertical inheritance and the origin of the HCCS in the last common ancestor of neokaryotes (i.e. Corticata and scotokaryotes). Paralogy, misrooting and weak resolution of any single-gene tree collectively explain most seeming ‘discordances’ of their HCCS tree much better than does LGT. Janouškovec et al. (2017) wrongly rooted the tree within Chromista which cannot be the ancestral eukaryote group. If rooted between scotokaryotes and corticates as it should have been, we see that all major groups of Chromista (Cryptista, Haptista, Alveolata, Heterokonta and even Rhizaria) have at least two distinct paralogues, whereas many groups within scotokaryotes and Plantae seem to have only one (as animals at least are known to) and Eozoa none (except for the acknowledged Percolomonas LGT). If properly rooted, it fits predominantly vertical inheritance of the two major HCCS paralogues throughout neokaryotes. Janouškovec et al. (2017) overlooked the evidence that fungi have separate paralogues for cytochromes c and c1 and heterokonts and apicomplexa have two paralogues (Allen et al. 2008) . Their tree showed for the first time that at least some red algae have HCCS and that most hacrobian groups have two, yet they failed to realise that if host and red algal symbiont each had two, one for each mitochondrial cytochrome c, the ancestral chromist would have had four. Inevitably, this redundancy would lead to random losses of one c- and one c1-specific HCCS as lineages diverged, which would yield four different combinations of ancestral paralogues, so every chromist lineage should have two of the four original paralogues and chromists as a whole would lie on four different paralogue branches on the tree, essentially as Janouškovec et al. (2017) found. Their tree did not fully sample all paralogues for each group as it shows only one paralogue for chlorophytes, choanoflagellates, apicomplexa and diatoms, yet all were known to have two (Allen et al. 2008) , so some other groups with only one on their tree may also actually have two. Differential paralogue sorting following chromist-chimaera generated redundancy, not LGT, is likely the major reason for complexity of their HCCS tree for heterokonts and Hacrobia. Treating the five seemingly distinct heterokont clades as evidence for polyphyletic LGTs into heterokonts from other chromists was unwarranted. We suggest that most chromists have two paralogues because each lineage still uses different paralogues for cytochrome c and c1. Chromist paralogues with slightly shorter branches and sparser taxonomic representation group with the also shorter branch obazoan scotokaryotes (opisthokonts/Amoebozoa); the other paralogues with slightly longer branches group with Plantae as a corticate clade, in which only Thecamonas is an intruding scotokaryote. Presence of heterotrophic oomycetes in four places and of cryptists in four (with heterotrophs in three) is best seen as evidence for ancestral chromist chimaerism and plastid loss by oomycetes, goniomonads and kathablepharids, not LGT. The longer branch corticate cluster is highlighted as having four distinct groups but these represent the four different deepest branching heterokont subgroups (Ochroph yt a, Pseudofungi, Labyrinthulea, Bicoecia) which often do not group together on single-gene trees and whose correct branching order (more clearly shown than hitherto in their 201-protein tree, including holophyly of phyla Bigyra and Gyrista) is rarely recovered by rDNA, so there being three ‘separate’ clades simply reflects poor basal resolution on all singlegene chromist trees. Allen et al. (2008) debated whether the c/c1 paralogue divergence occurred in the ancestral neokaryote and was lost by the animal ancestor or evolved polyphyletically. The generality of at least two paralogues in chromists, even in Rhizaria, makes it more likely that HCCS duplication ocurred immediately after this protein originated and was lost independently in animals and perhaps some other groups. More thorough taxon sampling, especially in Rhizaria, red algae, glaucophytes and deep branching scotokaryotes, and biochemical studies to test whether dual paralogues in hacrobians (e.g. Roombia, Guillardia and Emiliania even in the same species) do indeed function separately for c and c1 (as in Saccharomyces) should clarify evolution of mitochondrial cytochrome c biogenesis and also have the potential to provide clear evidence from differential paralogue retention that even Rhizaria retain proteins from both the chromist host and red algal symbiont despite retaining no chromophyte plastid. That already appears to be so for the oomycetes (which collectively appear to have three HCCS paralogues, though as expected no one genus has more than two), whose ancestry by plastid loss from an ancestral heterokont eukaryote-eukaryote chimaera has often been doubted. Presence of deep-branching heterotrophic cryptists (Roombia, Goniomonas) in three different paralogue clades (Janouškovec et al. 2017) is also best explained as differential retention from the ancestral chromist eukaryote-eukaryote chimaera and thus evidences ancient multiple plastid loss in early diverging cryptist lineages. The sole retarian sequence (Plasmodiophora) groups strongly with a cryptist sequence (Goniomonas), whereas the sole cercozoan one (Lotharella) groups strongly with the more distant of the two haptophyte paralogue clusters, suggesting that both rhizarian paralogues share closer common ancestry with hacrobian chromist paralogues than with Plantae or scotokaryotes, consistent with chromist monophyly and no LGT. Janouškovec et al. (2017) suggested that HCCS arose in the ancestral eukaryote, but their tree instead strongly supports the original idea that it arose in the common ancestral neokaryote (Cavalier-Smith 2010a) . Its complete absence in Eozoa except for the likely Percolomonas LGT favours absence as the ancestral eukaryote condition, and still favours Eozoa as the earliest branching eukaryotes (Cavalier-Smith 2010a) , contrary to Janouškovec et al.’s (2017) interpretation. Ciliates, which have none of the putatively four ancestral chromist paralogues, and Malawimonas (both previously known to have the multiprotein prokaryote system like jakobids: Allen et al. 2008) appear to be the only major neokaryote lineages that lost all HCCS, apart from secondary anaerobes like metamonads that lost both cytochromes. ( Janouškovec et al. (2017) ought to have excluded the sequence ostensibly from the ciliate Pseudokeronopsis (wrongly labelled ‘ochrophytes’) as it is likely contamination from the ciliate’s diatom food not a unique LGT into ciliates.) Timing of rhizarian origins and basal radiations If Harosa are ~ 750 My old (Cavalier-Smith 2013b; CavalierSmith et al. 2015a) , the proportions of the Fig. 1 tree suggest that Rhizaria may have originated and diverged into Cercozoa and Retaria ~ 620 My ago. This closely fits the earliest fossil e v i d e n c e a s t h e f o r a m i n i f e r a n b i o m a r k e r 2 4 - n propylcholestane was first found ~ 650 My ago between the two Cryogenian snowball global freezing episodes (Brocks et al. 2017) . It is sometimes suggested that eukaryotic algae could not have survived below a kilometre of ice during the ‘snowball earth’ Cryogenian glaciations—Sturtian (717–660 Mya) and Marinoan (640–635 Mya). However, contrary to that assumption, eukaryotic algae, notably chromist diatoms and green algal Plantae within sea ice and glacial ice respectively, survive far better and are much more abundant there than cyanobacteria (Boetius et al. 2015) . Moreover, presence of simple macroalgae in Marinoan age rocks (Ye et al. 2015) implies that there were local patches of open sea maintained by volcanic heating at least during the shorter Marinoan snowball. If the complex phosphatic scales from the Fifteenmile Group that must be at least 717 My old (Cohen and Knoll 2012) are from an extant corticate group—possible but not certain (Cavalier-Smith 2013b suggested they are Viridiplantae) , Plantae and Chromista probably originated shortly before the Sturtian glaciation; their origin may even have triggered that glaciation (Feulner et al. 2015; Tziperman et al. 2011) . The last Cryogenian glaciation ended ~ 635 My ago (Le Heron 2012; Rooney et al. 2015) so Rhizaria may have originated immediately Snowball Earth melted. If our estimated date for the origin of Rhizaria (~ 620 Mya) is approximately correct, their secondarily heterotrophic harosan ancestors possibly originated by losing the chromist plastid and evolving pseudopodia for a benthic existence below the near-global ice cap at a time when being a benthic predator and consumer of ocean floor detritus may have been more advantageous than an algal mode of life that must have become almost impossible under that global km or more thick ice for scores of millions of years. As the ocean was largely anoxic (Hood and Wallace 2015) but probably well mixed (Ashkenazy et al. 2013), this presumably happened in local exceptional, somewhat separated regions of above average oxygenation, probably shallower zones in volcanic areas of thinner and periodically broken and melted ice that must have existed to allow survival of aerobic protists, and whose genuine existence on ocean margins is independently indicated by local glacier-associated banded iron formations (Guilbaud et al. 2015) . However, neither chromists nor scotokaryotes could have become abundant and diversified into most modern groups till after the global thaw and reoxygenation at the beginning of the Ediacara period 635 My ago. No undoubted fossil Rhizaria are known before deep oceanic waters also became oxic after 580 My ago (Canfield et al. 2007) . The oldest generally accepted fossil Foraminifera (~ 545 My ago) and Polycystinea (~ 505 My ago) date only from close to the Precambrian/Cambrian boundary (542 My ago) (Groussin et al. 2011; Maletz 2011; Won and Below 1999) , which is consistent with the relative time of origins of scotokaryote and corticate deepest lineages on Fig. 1. The 100 My older biomarkers may therefore have come from stem Rhizaria rather than from Foraminifera which are not the deepest branch on the trees. However, we can interpret Fig. 1 sensibly in relation to geological time only if we contract the excessively long retarian branches so their termini line up with more averagely evolving clades. Without such contraction, basal crown Foraminifera and Polycystinea would appear to have not yet originated (!) in comparison with the pediglissid or phytomyxid clades that have more typical rates of sequence evolution for eukaryotes. Contracting the whole ectoretan branch including its stem uniformly by a factor of two would roughly line up branch termini with those of other Rhizaria. But that would not solve the problem of unequal evolutionary rates, as the base of crown Foraminifera and crown Polycystinea would still be many times younger than the origin of Pediglissa inferred above to be only ~ 350 My ago. However, simply contracting the bare stem of Ectoreta to almost zero would be almost enough to reduce the acantharean branch to the same length as most Endomyxa. Further contractions of the stems alone of Polycystinea and Foraminifera and of their common stem would then be sufficient to bring the base of each fossilogenic group to a date in line with the base of the bilateral animal clade whose origin effectively defines the Precambrian/Cambrian boundary (543 My ago). Thus comparison of our multigene tree with the fossil record for Ectoreta provides direct evidence that rates of protein evolution were episodically many times higher in the stems at the base of major branches than in the later terminal branches after their primary radiations. This conclusion is strongest for Foraminifera as we now have data for all three classes and their base (last common ancestor) is earlier than that for the sofar sequenced polycystines, Spongosphaera and Collozoum, which probably do not represent the full phylogenetic depth of extant Polycystinea (Gilg et al. 2010; Krabberød et al. 2011) , Collozoum especially belonging to the rather young order Collodarida (Ishitani et al. 2012) . The final three sections below further discuss timing of the origin and major diversifications and cell innovations of Rhizaria. We emphasise that sequence trees contain information not only about cladistic aspects of phylogeny, upon which most authors focus, but also about historical patterns of innovations and timing, but are liable to misinterpretation unless using fossil evidence critically also. Multiprotein trees are superior for this purpose to rDNA or other single-gene trees, but can be properly understood only by mapping them onto the fossil record and carefully assessing the likelihood of accelerations and decelerations in sequence evolution across the tree, using objective fossil dates as controls. Marked novelty, explosive radiations, and molecular coevolution affect sequence rate changes Darwin (1859) argued that major innovations would immediately cause much more rapid evolutionary radiation than at other times and also stressed under the rubric ‘correlation of growth’ that different organismal features necessarily coevolve, writing ‘when slight variations in any part occur ... other parts become modified’ (Darwin 1859 p. 143) . This is highly relevant to multigene trees, because a majority of the 187 proteins in our trees are parts of very few large macromolecular complexes—they are not 187 functionally independent proteins (nor 50,964 independent amino acids). Their ability to bind to each other must be amongst the most important influences on their evolution. As supplementary Table S4 shows, 16 form the proteasome; 43 are part of the large ribosomal subumit; 30 part of the small ribosomal subunit (so must bind specifically to 18S rDNA and/or each other); 2 part of the signal recognition particle (effectively a third ribosomal subunit for secreted/membrane proteins); 13 more function in translation, so several will bind temporarily to ribosomes; and 8 are part of the chaperonin TCP complex. Thus, 88 are ribosome-associated and 24 more with proteasomes or TCP—so these 112 proteins are concerned with the function of just 3 macromolecular complexes, a very narrow range of cell functions all related to the synthesis or proper folding of proteins or their degradation if improperly folded. Many others are parts of smaller macromolecular complexes: microtubules, chaperones like Hsp70 and 90 also related to protein folding so have many hundreds of interactors, or chromatin histones or Mcms, their close associations explaining the coevolutionary constraints on their evolution that generally make them evolve so slowly as to be useful deep phylogenetic markers. The roughly equal branch lengths in most clades on Fig. 1 probably reflect the normal constraints where mutations changing components of what Mayr called ‘coadapted gene complexes’ too much are removed by purifying selection, but small intrinsically deleterious changes evade such purification if compensated by reciprocal changes in interactor molecules (probably themselves mildly deleterious in the absence of phenotypically corrective suppressor mutations). Such regions of trees appear to evolve in semi-clock-like fashion giving rise to the always extremely oversimplified and sometimes grossly false idea of a ‘molecular clock’. Sometimes, radical changes in adaptive zone cause major changes during protist organismal and molecular evolution, affecting so many genes that genuine ‘genetic revolutions’ (Mayr 1954, 1960) are associated with them: transitions from aerobic to anaerobic life and to parasitism are two such revolutions. When both occur at the same time, the outcome is particularly dramatic. The anaerobic retarian haplosporidian parasite Mikrocytos (Burki et al. 2013) and Microsporidia (protozoa related to Fungi) are both anaerobic animal parasites that independently converted their mitochondria to mitosomes, losing their mitochondrial genomes, and miniaturised their cells and reduced their nuclear genome size to a greater extent than any other eukaryotes (both came from ancestors that also had lost cilia and thus also around a thousand genes encoding their proteins). Not surprisingly, both have the longest branches by far on eukaryote multigene trees, so we omitted them from our trees to avoid distorting other branches—such omission is rational and permissible, but it conceals the grossly non-clock-like nature of much protein evolution. In microsporidia at least, not only stems of sequence trees are grossly stretched, but so clearly are many branches. Thus when genomes become unusually small, they can attain equilibrium at permanently elevated evolutionary rates compared with other eukaryotes, and still survive. It is likely that reducing the number of gene product interactions weakens the conservative effect of coadapted gene complexes. Molecular evolutionary rates are immensely more variable and unpredictably erratic than assumed by even the best computer-based ‘molecular clock’ dating efforts (e.g. Eme et al. 2014) , which makes them probably less accurate for backward extrapolation than direct careful comparisons of branch lengths on trees with the more reliable fossil ages interpreted critically (e.g. Cavalier-Smith 2013b) . Multigene trees inform us about macroeveolutionary processes as well as phylogeny Cavalier-Smith (1980 b) argued that rRNA evolved nothing like a molecular clock and that the radical differences between eukaryotic 18S rDNA on the one hand and eubacterial and chloroplast and mitochondrial 16S rRNA are attributable to an episodic short term dramatic increase in the rate of molecular evolution in the ancestor of eukaryotes when the nucleus originated and transcription and translation were first segregated into separate cellular compartments. This idea was developed in more detail subsequently (Cavalier-Smith 1981b, 1987, 2002a, b) in relation to details of the radical transformations during eukaryogenesis, the greatest megaevolutionary destabilisation and restabilisation under a novel structural and adaptive regime in the history of life (Cavalier-Smith 2014) . Unfortunately, these arguments are often overlooked or ignored, especially by mathematicians and biochemists specialising in sequence evolution, often too deeply imbued with the seductive but grossly oversimplified idea of a molecular clock, and less familiar with the extensive evidence from fossils and organismic evolution that overall evolutionary change on the macroscale is highly episodic with long periods of relative stasis punctuated by short periods of ultrarapid, socalled quantum evolution (Simpson 1944, 1953; Eldredge and Gould 1972; Gould and Eldredge 1977, 1986; Mayr 1988b) . Stackebrandt and Woese (1981) admitted there was no evidence that archaebacterial and eubacterial rDNA has evolved at the same rate, yet their assumption that a uniform rate can be applied not only within their three ‘domains’ of life, but (even more doubtfully) to the stems that connect their bases to one point, has been widely adopted as the core of the erroneous Woesian paradigm of their equal antiquity, always grossly contradicted by the fossil record, as Cavalier-Smith (1981b, 1987, 2002a, b, 2006a, b, 2014) argued in detail—and has never been refuted by serious argument or specific evidence. We raise this issue here because it is extremely important to proper evolutionary interpretation of multigene trees, and because Fig. 1 provides the first compelling relatively direct evidence for the idea that long bare stems on sequence trees must at least sometimes be explained by relatively short periods of many times more rapid evolution than usual, followed by a relatively quick return to more normal rates. In other words, macroevolutionary ideas of punctuated equilibria and quantum evolution are indeed applicable to molecular sequence evolution as well as to cell and organismal evolution. This important molecular evolutionary conclusion has been evident for decades from 18S rDNA trees; for example Pawlowski and Berney (2003) stressed the episodic nature of stem lineage rDNA evolution of Foraminifera, but its radical contradiction of many prevailing assumptions, especially about the large-scale structure and root of the tree of life (Cavalier-Smith 2014) and how sequence trees should be mapped onto the fossil record, is not widely realised. Cavalier-Smith et al. (1996) argued that within animals the long stem at the base of Bilateria, and within plants the long stem at the base of florideophyte red algae, are the prime reasons why these branches are so much longer than those of their radiate animal and lower red algal relatives respectively. The good fossil record for bilaterian and radiate animals and floridiophytes necessitates the conclusion for both groups of a temporary elevated and later reversed high rate of sequence evolution. In these cases, unlike the even more dramatic long stem at the base of eukaryotes in rDNA trees, there is no specific cell evolutionary explanation for such an episodic acceleration, so they may be essentially evolutionary accidents. As similar long bare stems are found at the base of many rDNA trees, Cavalier-Smith et al. (1996) argued that in most cases these are probably better explained by episodic dramatic elevations in rate in a single lineage followed by return to normal low rates subsequently, rather than by a permanent more moderate rate increase that persisted in all daughter lineages. Distinguishing hugely accelerated episodic quantum evolution (frequent on taxonomically broad sequence trees) from a single more permanent evolutionary increase can be difficult, but for lineages with a good fossil record it can be done by critically cross-correlating fossil dates derived from the atomic decay clock that dates igneous rocks with great accuracy (and by interpolation objectively gives necessarily somewhat less accurate dates for interbedded fossiliferous ones) with taxonomically well-sampled phylogenetic trees. The example of Retaria here does this for our protein trees, just as the fossil evidence that eukaryotes are much younger than eubacteria did for the universal small subunit rDNA tree (Cavalier-Smith 1991, 1997, 2002b) and Bilateria and Floridiophytina did for 18S rDNA (Cavalier-Smith et al. 1996) . However, for some even more dramatic rate elevations for rDNA for which no fossils are available, we cannot at present get such direct evidence for episodic quantum evolution; two recent examples from Cercozoa for immensely long branches for 18S rDNA of Helkesida (Cavalier-Smith et al. 2008, 2009; Bass et al. 2016) , and Reticulamoeba (Bass et al. 2012) . Both seem most likely examples of grossly stretched stems, followed in Reticulamoeba by no higher rates than usual within the subsequent clades but remaining high at least in the basal three branches in helkesids. Our multiprotein trees show that Helkesimastigoidea also have the longest branch of all Cercozoa, but the evolutionary rate for Helkesimastix is much less inflated than for rDNA; this milder branch distortion enabled us to establish their correct position confidently for the first time. Separate trees could be run for the numerous ribosomal proteins and non-ribosomal proteins to test whether coevolution of ribosomal proteins and ultrarapidly evolving rDNA contributed disproportionally to the length of the helkesean branches in Figs. 1, 2, 3 and 4, or whether the acceleration is genome-wide. Cavalier-Smith (2015) discussed numerous other equally extreme examples of clade-specific rDNA accelerated evolution in other phyla, especially Percolozoa and gregarines within Miozoa. As all five cases with a good fossil record support quantum evolution in stem lineages and none supports the more widespread assumption of a single unreversed more moderate increase in rate as the explanation for exceptionally long branches, the idea that stem quantum evolution is the best general explanation of bare stems at the base of ultralong branches (Cavalier-Smith et al. 1996) has held up rather well. We therefore suggest that it is also the most likely explanation for the bare basal stem of the long helkesean branch on the multiprotein trees (Figs. 1, 2, 3 and 4). Combining fossil and sequence evidence for the age of Rhizaria Within Rhizaria, the 200 My discrepancy between the fossilcalibrated clock-inferred date for the origin of Foraminifera (650–920, mean 770 My) and the oldest accepted cell fossils (545 My) (Groussin et al. 2011) (the former markedly lower than an earlier estimate of 690–1150 My (Berney and Pawlowski 2003) suggests that clock estimates may still be inflated, possibly by the long bare stems at the base of most foraminiferan orders—only the minimum 650 My date coincides with the biomarker date. Their trees are also probably topologically distorted by the long unbroken branch of Reticulomyxa, which our multiprotein trees show to be more closely related to rotaliids than to miliolids and not to branch well below their common ancestor as their 4-gene trees arguably incorrectly show. However, the age discrepancy would be less if fossil agglutinated tubes dated 635–716 My ago really are Foraminifera (Bosak et al. 2012). Unfortunately, though they resemble extant Foraminifera more than other known protists, we cannot rule out the possibility that they were made by other amoeboid Rhizaria or Amoebozoa, as the later date for the foraminferan biomarker suggests—identifying agglutinated protist fossil tests is problematic (Streng et al. 2005) . Such tests tell us when pseudopodia able to manipulate particles to create them evolved, but are often not unambiguously assignable to Rhizaria or Amoebozoa, as noted previously (Cavalier-Smith 2006a, 2013b; Cavalier-Smith et al. 2016) . The stem at the base of Rhizaria is longer than for any other major eukaryote group (Figs. 1 and 4) and might indicate temporarily accelerated protein sequence evolution— useful for making monophyly of Rhizaria far easier to recover than for a group like Hacrobia or scotokaryotes with extremely short stems, but possibly distorting apparent relative timing of evolutionary events. Several recent reviews give inflated dates for the origin of Rhizaria and crown eukaryotes, and ignore arguments for there being no compelling fossil evidence for any crown eukaryotes before ~ 800 My ago (Cavalier-Smith 2002b, 2006a, 2013b) . Showing Rhizaria and Amoebozoa as 760 My old in Fig. 1 of Lenton et al. (2015) is presumably based on the assumption that vase-shaped marine fossils belong to these groups, which is questionable for both groups and almost certainly wrong for those misidentified as euglyphids. Not only is the morphological argument for Precambrian euglyphids totally unconvincing (see Cavalier-Smith 2006a, 2013b) , but as previously explained euglyphids nest so shallowly in rDNA trees that it is likely that they originated only in the Mesozoic not in the Proterozoic (Fig. S8). Our multiprotein trees, the first with good cercozoan taxon sampling to include a euglyphid (Paulinella), are consistent with euglyphids being much younger than Rhizaria and animals. Unsurprisingly, both Lenton et al. (2015) and Knoll (2014) p e r s i s t i n r e g a r d i n g t h e p o s s i b l y 1 . 1 – 1 . 2 G y o l d Bangiomorpha (Butterfield 2000) as a bangiophyte red alga, despite reasoned morphological arguments against that (Cavalier-Smith 2002b, 2006a; there is no compelling evidence it is not a cyanobacterium) and despite arguments from sequence trees that such an identification/age is highly discordant with other evidence (Berney and Pawlowski 2006; Eme et al. 2014) , as earlier trenchantly stressed for rDNA trees (Cavalier-Smith 2002b) . For both Bangiomorpha and the claimed 760 My ‘euglyphids’, the discrepancy is two-fold, much too great to justify it being repeatedly ignored. The arguments of Feulner et al. (2015) for a temporal and hypothetically causal link between eukaryote algal origins and the Sturtian glaciation are weakened by their Fig. 1 exclusive comparison with a multigene ‘clock’ study (Parfrey et al. 2011) that probably erroneously placed their origin hundreds of millions of years earlier than did the best analyses of Berney and Pawlowski (2006) and Eme et al. (2014) and discordantly gave a crown eukaryote ‘date’ twice as old. That gross discrepancy arises only partly because Parfrey et al. (2011) ignored the valid criticisms of Berney and Pawlowski (2006) over widespread use of doubtful Proterozoic ‘calibrations’, unwisely using more than earlier studies: six of their seven earliest fossil ‘calibrations’ are highly dubious, arguably all substantially too early. One is Bangiomorpha; the other five are: claimed sponge biomarkers probably are not specific ( A n t c l i f f e 2 0 1 3 ; s e e a l s o C a v a l i e r- S m i t h 2 0 1 7 ) ; Palaeoarcella, unlikely actually to be arcellinid (see Cavalier-Smith 2013b; Cavalier-Smith 2015) ; Kimberella that may not be a bilaterian (Brasier 2009); Palaeastrum is not obviously a green alga, and could be another eukaryote or a cyanobacterium (as could other claimed green algae in the same reference); gammacerane is not a reliable biomarker for ciliates, as it is also made by proteobacteria (CavalierSmith 2006a; Kleemann et al. 1990) as well as by a fungus and a fern. Eme et al. (2014) rejected the ciliate calibration but retained the others; they found that using soft bounds for calibration points that can allow the programme to overide discordant wrongly dated/identified fossils gives markedly younger dates for most models but noted that the autocorrelated lognormal relaxed clock model is insensitive to whether calibration is soft or hard and also gives the lowest ages, most consistent with the fossil record. Eme et al. (2014) noted: ‘all recent molecular clock analyses seem to concur that the Bangiomorpha fossil date or taxonomic assignment is problematic’. The Parfrey et al. trees used only 3–15 genes, not 159 as in the more thorough Eme et al. (2014) analysis; primarily used an uncorrelated clock model (less accurate than correlated ones: Lepage et al. 2007) ; and inferior sitehomogeneous models (often not even including gamma distribution intersite rate variation); and thus were necessarily less reliable, quite apart from almost all earliest ‘calibrations’ systematically biasing their scale: omitting them reduced most clade dates only about 200–300 My. The age conflicts amongst their methods covered a 2–3-fold range, making conclusions virtually meaningless. Eme et al. showed that uncorrelated clock models and site-homogeneous models (used by Parfrey et al. and most other investigators) can systematically inflate inferred ages by hundreds of millions of years and have severalfold wider error bars than trees using an autocorrelated lognormal clock model that yields much younger dates closer to those of unambiguously identified fossils, and that the CIR method for PhyloBayes (the only autocorrelated model used by Parfrey et al.) yields dates that are similarly much older than by the lognormal model even though under simulation CIR appeared as reliable (Lepage et al. 2007) . This suggests that for real data, a lognormal relaxed clock model may be greatly superior, but it is still unlikely that any statistical model can reliably reflect unique evolutionary events like the grossly non-clock like phenomenon of idiosyncratic episodically dramatically increased rates discussed above. Neither analysis included any Retaria (the Rhizaria with the best fossil record), as they too blatantly infringe the assumption of a clock. Figure 2 of Parfrey et al. inferred far too early dates for Cercozoa (1100 My) and Harosa/ haptophytes (1600 My—older than that inferred for red algae despite using an inflated ‘calibration’ for the latter), and incorrectly claimed that their analysis was congruent with the fossil record; their Fig. 2 shows all major crown eukaryote clades older than the dates of any unambiguous fossils, in many cases 2–3 times as old! Corticates are shown as ~ 1650 My old, whereas the oldest acceptable body fossil corticate calibration point was 550 My (Doushanto red algae) a 1.1 billion year discrepancy—the error was double the size of the signal. Omitting the dubious Proterozoic calibrations (their Fig. S1) reduced inferred ages for Cercozoa to 900 My, Harosa/ haptophytes to 1250 My and corticates to 1350 My, but all deep nodes were still seriously inflated by unrealistic models. Yang et al. (2016) also grossly overestimated corticate age as ~ 1693 My, red algae as ~ 1500 My and subphylum Eurhodophytina (likely sisters of chromist nucleomorphs) at ~ 1194 My. Post-Marinoan steranes ~ 622 ± 20 Mya have reasonably been interpreted as produced by a red algal plankton bloom (Elie et al. 2007) , but if early chromists had similar steranes to red algae, they might be from early chromophytes. Absence of detectable stigmastane or ergostane before ~ 650 My ago (Brocks et al. 2017) shows that green or red algae were not abundant before then, making it unlikely that the slightly deeper-branching Glaucophyta and therefore chloroplasts and Plantae evolved substantially before the beginning of the Sturtian glaciation ~ 717 My ago. If the Bangiomorpha fossils are little older than their minimum inferred date of ~ 722 Mya, as Yang et al. (2016) recognise is likely, their original identification as eukaryotic would be less problematic than if they were 1200 My old as claimed for decades, but even if they are eukaryotic (which their morphology allows but does not require), there is no reason to consider them as Bangiophyceae rather than an earlier-branching group of red algae, e.g. Compsopogonophyceae or Stylonematophyceae with analogous filaments, or as an extinct group of stem Plantae. Using Bangiomorpha to ‘calibrate’ sequence trees when its identity is unknown and its date uncertain by 500 My is ridiculous (Cavalier-Smith 2002b) . There is no good reason to think that red algae or chromists originated before ~ 750 Mya; both could be more recent. Fossil-calibrated dates of Eme et al. (2014) using siteheterogeneous phylogenetic and lognormal clock models for Plantae, Hacrobia and Harosa (~ 900 My ago) agree more than the contradictory Parfrey et al. (2011) trees with a reasonable statement of Feulner et al. (2015) that the first major eukaryote algal radiation was around 800 My ago, but even 900 My could be somewhat inflated for reasons noted above, steranes suggesting ~ 650 My only. All palaeontologists agree there was a major eukaryotic radiation 800–750 My ago of more complex fossils than previously (e.g. Butterfield 2015; Knoll 2014) ; we now suggest this reflects basal radiation of the scotokaryote and corticate eukaryote clades (Cavalier-Smith 2013b; Cavalier-Smith et al. 2015b) collectively denoted neokaryotes (Cavalier-Smith 2018), not the very first eukaryote radiation as Cavalier-Smith (2002b, 2006a) had argued before concluding that the eukaryote root probably lies within Eozoa, making them somewhat older than corticates and scotokaryotes (Cavalier-Smith 2010a) . The now more critical sterane biomarker work increasingly suggests that the first major rise of eukaryote algal abundance did not occur before ~ 650 My in the interglacial between the Sturtian and Marinoan glaciations (Brocks et al. 2017) . Comprehensive fossil-calibrated site-heterogeneous multigene trees place excavates and the crown eukaryote root close to 1 Gy (Eme et al. 2014 Fig. 4) , which closely fits the thesis that the putatively basal eukaryote supergroup Eozoa is no older than ~ 1 Gy and that no fossils older than 1.1–1.2 Gy provide significant evidence even for stem eukaryotes (Cavalier-Smith 2013b) , which assumed that the eukaryote root was within Eozoa. However, if the root of the eukaryote tree were between Eozoa and neokaryotes (all eukaryotes except Euglenozoa, Percolozoa and Eolouka: Cavalier-Smith 2018) not within Eozoa, then crown eukaryotes might be no older than 800– 850 My as previously argued (Cavalier-Smith 2002b, 2006a) ; so far, the oldest biomarker age for archaebacteria (likely sisters of eukaryotes, not their ancestors: Cavalier-Smith 2014) is ~ 820 My (Schintele and Brocks 2017) but earlier evidence is wanting. Irrespective of the position of the root of the eukaryotic tree, it is likely that Rhizaria are substantially more than half as old as eukaryotes and could be over three quarters of their age, significantly older than animals and other opisthokonts. Twenty-five major conclusions 1. Classical Retaria, recently made new subphylum Ectoreta, generally branch as sisters to Endomyxa, and are probably not sisters of Gromiidea alone; the recent transfer of Endomyxa from Cercozoa to Retaria fits their shared reticulose pseudopodia and absence of biciliate trophic phase. 2. Radiozoa is a clade on most trees; their occasional apparent paraphyly with only Polycystinea sister to Foraminifera may be a sparse sampling artefact. 3. Endomyxa are a clade, contrary to some multigene trees. 4. Monadofilosa are a clade but Reticulofilosa are probably paraphyletic with Chlorarachnea (not Granofilosea as some 18S rDNA trees suggested) sister to Monadofilosa. 5. Metromonadea are the deepest branching monadofilosan class, probaby a clade. 6. The non-ciliate non-filose amoeba Guttulinopsis is sister to the essentially non-amoeboid uniciliate gliding zooflagellate Helkesimastix (clade Helkesida). Episodic dramatic increases in sequence evolutionary rates in stem helkesids affected both rDNA and most proteincoding genes. 7. Thecofilosea (including ventricleftids as the most divergent lineage) are probably a clade. 8. Cercomonads are probably polyphyletic, so their suborders are made separate orders: Cercomonadida and Paracercomonadida. 9. Cercomonadida are sisters of Glissomonadida, forming a robust clade of gliding soil flagellates here made new sarcomonad subclass Pediglissa. 10. Pediglissa are probably sister to euglyphids plus Nudisarca, not to Paracercomonadida, which are therefore assigned to a separate new sarcomonad subclass Paracercomonadia. 11. Imbricatea appear to be nested within Sarcomonadea; subclasses Placonuda and Placoperla are possibly not sisters. 12. The earlier possibility that scaly, filoreticulose Kraken are sister to paracercomonads is not supported by siteheterogeneous rDNA trees or ultrastructure, so we establish new order Krakenida for them, grouping them with environmental DNA clade eSarcomonad as new subclass Krakenia within possibly ancestrally scaly Imbricatea. 13. The position of Thaumatomonas (Placoperla) is inconsistent on multiprotein trees, but on trees without distant outgroups it is within clade Monadofilosa and sister to Placonuda (here revised by adding Discomonadida). 14. As protein trees suggest that Sarcomonadea/Imbricatea are a sister clade to Thecofilosea, we transferred Sarcomonadea from the basal monadofilosan superclass Eoglissa to derived superclass Ventrifilosa to make Ventrifilosa a robustly supported, ancestrally filose, clade. This also makes Eoglissa phenotypically more homogeneous as it now contains only the non-filose Metromonadea plus Helkesea. 15. Some transcriptome sequences of ‘Minchinia’ sp. are not from Minchinia (an endomyxan haplosporidian) but from contaminants; after eliminating human, green plant and diatom contaminants, the residue group on trees as a distant relative of core Cryptista; instead of being a previously unknown deep-branching hacrobian lineage close to Palpitomonas, it is more likely to represent a mixture of a cryptist and/or other contaminants and Minchinia genes that collectively group in that position on multiprotein trees. 16. New strongly holophyletic superclasses Proteomyxia and Marimyxia are established for Endomyxa. 17. Endomyxan Filoreta marina has both actin1 and actin2, suggesting that this gene duplication occurred earlier than previously thought and that actin2 may contribute to the reticulose body form of all Retaria, not just Ectoreta, providing the first molecular synapomorphy for the now expanded Retaria. 18. Combining fossil evidence for Cambrian Foraminifera with the remarkable length of retarian and especially foraminiferan branches on multigene trees, and crosscorrelation of dates with outgroups provides strong evidence for episodic dramatic increases in molecular evolutionary rates in stem lineages of both Ectoreta and Foraminifera, with similarly large rate reductions to relatively normal levels within crown Foraminifera. 19. Similar episodically grossly accelerated evolutionary rates in stem lineages are probably the major cause of long bare stems at the base of ultralong branches on sequence trees; we discuss their implications for fossilcalibrated ‘relaxed-clock’ statistical phylogenetic estimates of extrapolated dates for the origins of deepbranching clades like Retaria. 20. Patterns of evolution on multigene trees provide evidence for extensive molecular coevolution; these make it unrealistic to consider molecular evolution solely in terms of individual molecules and illuminate debates about megaevolution and quantum evolution 21. We discuss the origins of Rhizaria and the contrasting phenotypes that arose during their earliest divergences, including the evolutionary significance of the basic contrast between the recently revised phyla (trophically reticulose non-flagellate Retaria; largely filose predominently gliding flagellate Cercozoa) and within Retaria between subphylum Endomyxa without and subphylum Ectoreta (Foraminifera, Radiozoa) with marked segregation by tests or capsules of endoplasm and reticulose ectoplasm. 22. Evidence that the alveid biciliate Ancoracysta is sister to other Haptista and has extensive cortical alveoli and a posterior ciliary vane and simple anterior ciliary hairs and a ventral feeding groove (similarly to the alveolate Colponema) strengthens the idea that these characters were all present in the common ancestor of both chromistan subkingdoms Harosa and Hacrobia and also in the stem lineage from which infrakingdom Rhizaria evolved. Thus Cercozoa likely lost the ancestral neokaryote ventral ciliary vane when they evolved posterior ciliary gliding, just as happened convergently in ancestral Sulcozoa. We correct misinterpretations of Ancoracysta ultrastructure. Rhizaria and other chromists have at least two p aralogues of the unimolecular neo karyote ccytochrome lyase suggesting that one is used for cytochrome c, the other for c1. Their distribution in four clades across the sequence tree probably results from differential retention and loss from the four paralogues deduced to have been present in the ancestral chromist eukaryote-eukaryote chimaera. Cercozoa and Retaria have non-homologous insertions in ribosomal protein L1 which show each is a clade; contrary to previous ideas, there is no shared rhizarian insertion. Contamination by foreign protists is much commoner than often realised in rhizarian samples, especially uncultured ones, so more effort is needed to remove it before analysing sequences. Acknowledgements We thank NERC for grant NE/E004156/1 and an earlier Professorial Fellowship for TCS, T. Nakayama and Ken-Ichiro Ishida for providing Paulinella protein sequences and R. Sierra for some rhizarian sequences and information about them. RL thanks NERC for a research studentship. We thank Oxford University Advanced Research Computing (ARC: for access to their clusters and efficient advice and Cédric Berney, Anna Maria Fiore-Donno and Libby Snell for early help in editing or aligning some new protein sequences. Compliance with ethical standards Conflict of interest The authors declare they have no conflict of interest. Open Access This article is distributed under the terms of the Creative C o m m o n s A t t r i b u t i o n 4 . 0 I n t e r n a t i o n a l L i c e n s e ( h t t p : / /, which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. Adl SM , Simpson AGB , Lane CE , Lukeš J , Bass D , Bowser SS , Brown MW , Burki F , Dunthorn M , Hampl V , Heiss A , Hoppenrath M , Lara E , le Gall L , Lynn DH , McManus H , Mitchell EAD , MozleyStanridge SE , Parfrey LW , Pawlowski J , Rueckert S , Shadwick L , Schoch CL , Smirnov A , Spiegel FW ( 2012 ) The revised classification of eukaryotes . J Eukaryot Microbiol 59 : 429 - 493 . https://doi. org/10.1111/j.1550- 7408 . 2012 . 00644 .x Allen CA , Jackson AP , Rigden DJ , Willis AC , Ferguson AC , Ginger ML ( 2008 ) Order within a mosaic distribution of mitochondrial c-type cytochrome biogenesis systems? FEBS J 275 : 2385 - 2402 Antcliffe JB ( 2013 ) Questioning the evidence of organic compounds called sponge biomarkers . Palaeontology 56 : 917 - 925 Arnold ZM ( 1966 ) Observations on the sexual generation of Gromia oviformis Dujardin . J Protozool 13 : 23 - 27 Arnold ZM ( 1972 ) Observations on the biology of the protozoan Gromia oviformis Dujardin . Univ Calif Publ Zool 100 : 1 - 168 Ashkenazy Y , Gildor H , Losch M , Macdonald FA , Schrag DP , Tziperman E ( 2013 ) Dynamics of a snowball Earth ocean . Nature 495 : 90 - 93 Bass D , Cavalier-Smith T ( 2004 ) Phylum-specific environmental DNA analysis reveals remarkably high global biodiversity of Cercozoa (Protozoa) . Int J Syst Evol Microbiol 54 : 2393 - 2404 Bass D , Moreira D , López-García P , Polet S , Chao EE , von der Heyden S , Pawlowski J , Cavalier-Smith T ( 2005 ) Polyubiquitin insertions and the phylogeny of Cercozoa and Rhizaria . Protist 156 : 149 - 161 Bass D , Chao EE , Nikolaev S , Yabuki A , Ishida K , Berney C , Pakzad U , Wylezich C , Cavalier-Smith T ( 2009a ) Phylogeny of novel naked filose and reticulose Cercozoa: Granofilosea cl . n. and Proteomyxidea revised. Protist 160 : 75 - 109 Bass D , Howe AT , Mylnikov AP , Vickerman K , Chao EE , Smallbone JE , Snell J , Cabral C Jr, Cavalier-Smith T ( 2009b ) Phylogeny and classification of Cercomonadidae: Cercomonas, Eocercomonas, Paracercomonas, and Cavernomonas gen . n. Protist 160 : 483 - 521 Bass D , Yabuki A , Santini S , Romac S , Berney C ( 2012 ) Reticulamoeba is a long-branched granofilosean (Cercozoa) that is missing from sequence databases . PLoS One 7 ( 12 ):e49090. 1371/journal.pone.0049090 Bass D , Silberman JD , Brown MW , Pearce RA , Tice AK , Joiusset A , Geisen S , Hartikainen H ( 2016 ) Coprophilic amoebae and flagellates, including Guttulinopsis, Rosculus and Helkesimastix, characterise a divergent and diverse rhizarian radiation and contribute to a large diversity of faecal-associated protists . Environ Microbiol 18 : 1604 - 1619 Bass D et al. ( 2018 ) Rhizarian 'Novel Clade 10' revealed as abundant and diverse planktonic and terrestrial flagellates, including Aquavolon n. gen . J Euk Microbiol in press Berney C , Pawlowski J ( 2003 ) Revised small subunit rRNA analysis provides further evidence that foraminifera are related to Cercozoa . J Mol Evol 57 ( suppl .1): 120 - 127 Berney C , Pawlowski J ( 2006 ) A molecular time-scale for eukaryote evolution recalibrated with the continuous microfossil record . Proc Roy Soc Lond B 273 : 1867 - 1872 Berney C , Geisen S , Van Wichelen J , Nitsche F , Vanormelingen P , Bonkowski M , Bass D ( 2015 ) Expansion of the 'Reticulosphere': diversity of novel branching and network-forming amoebae helps to define Variosea (Amoebozoa) . Protist 166 : 271 - 295 Berney C , Romac S , Mahe F , Santini S , Siano R , Bass D ( 2013 ) Vampires in the oceans: predatory cercozoan amoebae in marine habitats . Isme J 7 : 2387 - 2399 Biard T , Pillet L , Decelle J , Poirier C , Suzuki N , Not F ( 2015 ) Towards an integrative morpho-molecular classification of the Collodaria (Polycystinea, Radiolaria) . Protist 166 : 374 - 388 . 1016/j.protis. 2015 . 05 .002 Boetius A , Anesio AM , Deming JW , Mikucki JA , Rapp JZ ( 2015 ) Microbial ecology of the cryosphere: sea ice and glacial habitats . Nat Rev Microbiol 13 : 677 - 690 . nrmicro3522 Bosak T , Lahr DJG , Pruss SB , Macdonald FA , Gooday AJ , Dalton L , Matys ED ( 2012 ) Possible early foraminiferans in post-Sturtian (716 −635 Ma) cap carbonates . Geology 40 : 67 - 70 Brasier CM ( 2009 ) Darwin's lost world . Oxford University Press, Oxford Brocks JJ , Jarrett AJM , Sirantoine E , Hallmann C , Hoshino Y , Liyanage T ( 2017 ) The rise of algae in Cryogenian oceans and the emergence of animals . Nature 548 : 578 - 581 . nature23457 Brown MW , Kolisko M , Silberman JD , Roger A ( 2012 ) Aggregative multicellularity evolved independently in the eukaryotic supergroup Rhizaria . Curr Biol 22 : 1123 - 1127 Brown MW , Sharpe SC , Silberman JD , Heiss AA , Lang BF , Simpson AG , Roger AJ ( 2013 ) Phylogenomics demonstrates that breviate flagellates are related to opisthokonts and apusomonads . Proc R Soc B 280 : 0131755 Brown MW et al ( 2018 ) Phylogenomics places orphan protistan lineages in a novel eukaryotic super-group . Genome Biol Evol 10 : 427 - 433 . Bulman SR , Kühn SF , Marshall JW , Schnepf E ( 2001 ) A phylogenetic analysis of the SSU rRNA from members of the Plasmodiophorida and Phagomyxida . Protist 152 : 43 - 51 Burki F , Berney C , Pawlowski J ( 2002 ) Phylogenetic position of Gromia oviformis Dujardin inferred from nuclear-encoded small subunit ribosomal DNA . Protist 153 : 251 - 260 Burki F , Corradi N , Sierra R , Pawlowski J , Meyer GR , Abbott CL , Keeling PJ ( 2013 ) Phylogenomics of the intracellular parasite Mikrocytos mackini reveals evidence for a mitosome in Rhizaria . Curr Biol 23 : 1541 - 1547 Burki F , Inagaki Y , Bråte J , Archibald JM , Keeling PJ , Cavalier-Smith T et al ( 2009 ) Large-scale phylogenomic analyses reveal that two enigmatic protist lineages, Telonemia and Centroheliozoa, are related to photosynthetic chromalveolates . Genome Biol Evol 1 : 231 - 238 Burki F , Kudryavtsev A , Matz MV , Aglyamova GV , Bulman S , Fiers M , Keeling PJ , Pawlowski J ( 2010 ) Evolution of Rhizaria: new insights from phylogenomic analysis of uncultivated protists . BMC Evol Biol 10 : 377 Burki F , Okamoto N , Pombert JF , Keeling PJ ( 2012 ) The evolutionary history of haptophytes and cryptophytes: phylogenomic evidence for separate origins . Proc R Soc B 279 : 2246 - 2225 Burki F , Shalchian-Tabrizi K , Minge M , Skjaeveland A , Nikolaev SI , Jakobsen KS et al ( 2007 ) Phylogenomics reshuffles the eukaryotic supergroups . PLoS One 2 : e790 Burki F , Shalchian-Tabrizi K , Pawlowski J ( 2008 ) Phylogenomics reveals a new 'megagroup' including most photosynthetic eukaryotes . Biol Lett 4 : 366 - 369 Burki F , Kaplan M , Tikhonenkov DV , Zlatogursky V , Minh BQ , Radaykina LV , Smirnov A , Mylnikov AP , Keeling PJ ( 2016 ) Untangling the early diversification of eukaryotes: a phylogenomic study of the evolutionary origins of Centrohelida, Haptophyta and Cryptista . Proc R Soc B 283 :20152802. rspb. 2015 .2802 Butterfield NJ ( 2015 ) Frontiers in paleontology: early evolution of the Eukaryota . Palaeontology 58 : 5 - 17 Butterfield NJ ( 2000 ) Bangiomorpha pubescens n. gen., n. sp.: implications for the evolution of sex, multicellularity, and the Mesoproterozoic/Neoproterozoic radiation of eukaryotes . Paleobiology 26 : 386 - 404 Cachon J , Cachon M ( 1978 ) Sticholonche zanclea Hertwig: a reinterpretation of its phylogenetic position based upon new observations of its ultrastructure . Arch Protistenkunde 120 : 148 - 168 Cachon J , Cachon M , Tilney LG , Tilney MS ( 1977 ) Movement generated by interactions between the dense material at the ends of microtubules and non-actin-containing microfilaments in Sticholonche zanclea . J Cell Biol 72 : 314 - 338 Cachon J , Cachon M , Estep KW ( 1990 ) Phylum Actinopoda . In: L M, Corliss JO , Melkonian M , Chapman DJ (eds) Handbook of Protoctista. Jones and Bartlett , Boston, pp 334 - 346 Canfield DE , Poulton SW , Narbonne GM ( 2007 ) Late-Neoproterozoic deep-ocean oxygenation and the rise of animal life . Science 315 : 92 - 95 Cavalier-Smith T ( 1974 ) Basal body and flagellar development during the vegetative cell cycle and the sexual cycle of Chlamydomonas reinhardii . J Cell Sci 16 : 529 - 556 Cavalier-Smith T ( 1980a ) r- and K-tactics in the evolution of protist developmental systems: cell and genome size, phenotype diversifying selection, and cell cycle patterns . BioSystems 12 : 43 - 59 Cavalier-Smith T ( 1980b ) Cell compartmentation and the origin of eukaryote membranous organelles . In: Schwemmler W, Schenk HEA (eds) Endocytobiology: endosymbiosis and cell biology, a synthesis of recent research . de Gruyter, Berlin, pp 893 - 916 Cavalier-Smith T ( 1981a ) Eukaryote kingdoms: seven or nine ? BioSystems l4 : 461 - 481 Cavalier-Smith T ( 1981b ) The origin and early evolution of the eukaryotic cell . In Molecular and cellular aspects of microbial evolution . In: MJ Carlile, J Collins, BEB Moseley (eds) Society for General Microbiology Symposium 32 . Cambridge University Press, Cambridge, pp 33 - 84 Cavalier-Smith T ( 1986 ) The kingdom Chromista: origin and systematics . In: Round FE , Chapman DJ (eds) Progress in Phycological Research , vol 4 . Biopress Ltd ., Bristol , pp 309 - 347 Cavalier-Smith T ( 1987 ) The origin of eukaryote and archaebacterial cells . Ann N Y Acad Sci 503 : 17 - 54 Cavalier-Smith T ( 1989 ) The kingdom Chromista . In: Green JC , Leadbeater BSC , Diver WL (eds) The Chromophyte algae: problems and perspectives . Clarendon Press, Oxford, pp 381 - 407 Cavalier-Smith T ( 1991 ) Intron phylogeny: a new hypothesis . Trends Genet 7 : 145 - 148 Cavalier-Smith T ( 1993a ) Kingdom Protozoa and its 18 phyla . Microbiol Rev 57 : 953 - 994 Cavalier-Smith T ( 1993b ) Evolution of the eukaryotic genome . In: Broda P, Oliver SG , Sims P (eds) The eukaryotic genome . Cambridge University Press, Cambridge, pp 333 - 385 Cavalier-Smith T ( 1995 ) Zooflagellate phylogeny and classification . Tsitologiia 37 : 1010 - 1029 Cavalier-Smith T ( 1997 ) Amoeboflagellates and mitochondrial cristae in eukaryote evolution: megasystematics of the new protozoan subkingdoms Eozoa and Neozoa . Arch Protistenk 147 : 237 - 258 Cavalier-Smith T ( 1998a ) A revised six-kingdom system of life . Biol Rev 73 : 203 - 266 Cavalier-Smith T ( 1998b ) Neomonada and the origin of animals and fungi . In: Coombs GH , Vickerman K , Sleigh MA , Warren A (eds) Evolutionary relationships among Protozoa . Kluwer, London, pp 375 - 407 Cavalier-Smith T ( 1999 ) Principles of protein and lipid targeting in secondary symbiogenesis: euglenoid, dinoflagellate, and sporozoan plastid origins and the eukaryote family tree . J Euk Microbiol 46 : 347 - 366 Cavalier-Smith T ( 2000a ) Membrane heredity and early chloroplast evolution . Trends Plant Sci 5 : 174 - 182 Cavalier-Smith T ( 2000b ) Flagellate megaevolution: the basis for eukaryote diversification . In: Green JC , Leadbeater BSC (eds) The flagellates . Taylor and Francis, London, pp 361 - 390 Cavalier-Smith T ( 2002a ) The phagotrophic origin of eukaryotes and phylogenetic classification of Protozoa . Int J Syst Evol Microbiol 52 : 297 - 354 Cavalier-Smith T ( 2002b ) The neomuran origin of archaebacteria, the n e g i b a c t e r i a l r o o t o f t h e u n i v e r s a l t re e a n d b a c t e ri a l megaclassification . Int J Syst Evol Microbiol 52 : 7 - 76 Cavalier-Smith T ( 2003a ) Genomic reduction and evolution of novel genetic membranes and protein-targeting machinery in eukaryoteeukaryote chimaeras (meta-algae) . Philos Trans R Soc Lond B 358 : 109 - 134 Cavalier-Smith T ( 2003b ) Protist phylogeny and the high-level classification of Protozoa . Eur J Protistol 39 : 338 - 348 Cavalier-Smith T ( 2004 ) Chromalveolate diversity and cell megaevolution: interplay of membranes, genomes and cytoskeleton . In: Hirt RP , Horner DS (eds) Organelles, genomes and eukaryote phylogeny . CRC Press, London, pp 75 - 108 Cavalier-Smith T ( 2006a ) Cell evolution and earth history: stasis and revolution . Philos Trans R Soc Lond B Biol Sci 361 : 969 - 1006 Cavalier-Smith T ( 2006b ) Rooting the tree of life by transition analysis . Biol Direct 1 : 19 Cavalier-Smith T ( 2007 ) Evolution and relationships of algae: major branches of the tree of life . In: Brodie J , Lewis J (eds) Unravelling the algae . CRC press, Boca Raton, pp 21 - 55 Cavalier-Smith T ( 2009 ) Megaphylogeny, cell body plans, adaptive zones: causes and timing of eukaryote basal radiations . J Euk Microbiol 56 : 26 - 33 Cavalier-Smith T ( 2010a ) Kingdoms Protozoa and Chromista and the eozoan root of the eukaryotic tree . Biol Lett 6 : 342 - 345 Cavalier-Smith T ( 2010b ) Deep phylogeny, ancestral groups, and the four ages of life . Philos Trans R Soc Lond B Biol Sci 365 : 111 - 132 Cavalier-Smith T ( 2013a ) Symbiogenesis: mechanisms, evolutionary consequences, and systematic implications . Ann Rev Syst Ecol Evol 44 : 145 - 172 Cavalier-Smith T ( 2013b ) Early evolution of eukaryote feeding modes, cell structural diversity, and classification of the protozoan phyla Loukozoa, Sulcozoa, and Choanozoa . Eur J Protistol 49 : 115 - 178 . 2012 . 06 .001 Cavalier-Smith T ( 2014 ) The neomuran revolution and phagotrophic origin of eukaryotes and cilia in the light of intracellular coevolution and a revised tree of life . In: Keeling PJ , Koonin EV (eds) The Origin and Evolution of Eukaryotes . Cold Spring Harbor Perspectives Biol, Cold Spring Harbor. cshperspect.a016006 Cavalier-Smith T ( 2015 ) Mixed heterolobosean and novel gregarine lineage genes from culture ATCC 50646: long-branch artefacts, not lateral gene transfer, distort α-tubulin phylogeny . Eur J Protistol 51 : 121 - 137 Cavalier-Smith T ( 2017 ) Vendozoa and selective forces on animal origin and early diversification: response to Dufour and McIlroy comment . Philos Trans R Soc Lond B Biol Sci 373 : 20170336 . 10.1098/rstb. 2017 .0336 Cavalier-Smith T ( 2018 ) Kingdom Chromista and its eight phyla: a new synthesis emphasising periplastid protein-targeting, cytoskeletal and periplastid evolution, and ancient divergences . Protoplasma 255 : 297 - 357 Cavalier-Smith T , Chao EE ( 1997 ) Sarcomonad ribosomal RNA sequences, rhizopod phylogeny, and the origin of euglyphid amoebae . Arch Protistenk 147 : 227 - 236 Cavalier-Smith T , Chao EE ( 2003a) Phylogeny and classification of phylum Cercozoa (Protozoa) . Protist 154 : 341 - 358 Cavalier-Smith T , Chao EE ( 2003b) Phylogeny of Choanozoa, Apusozoa, and other Protozoa and early eukaryote megaevolution . J Mol Evol 56 : 540 - 563 Cavalier-Smith T , Chao EE ( 2003c ) Molecular phylogeny of centrohelid heliozoa, a novel lineage of bikont eukaryotes that arose by ciliary loss . J Mol Evol 56 : 387 - 396 Cavalier-Smith T , Chao EE ( 2012a ) Oxnerella micra sp . n. ( Oxnerellidae fam . n.) , a tiny naked centrohelid, and the diversity and evolution of Heliozoa . Protist 163 : 574 - 601 . 2011 . 12 .005 Cavalier-Smith T , Karpov S (2012b) Paracercomonas kinetid ultrastructure, origins of the body plan of Cercomonadida, and cytoskeleton evolution in Cercozoa . Protist 163 : 47 - 75 . protis. 2011 . 06 .004 Cavalier-Smith T , Oates B ( 2012c ) Ultrastructure of Allapsa vibrans and the body plan of Glissomonadida (Cercozoa) . Protist 163 : 165 - 187 . 2011 . 10 .006 Cavalier-Smith T , Allsopp MTEP , Chao EE ( 1994 ) Chimeric conundra: are nucleomorphs and chromists monophyletic or polyphyletic ? Proc Natl Acad Sci U S A 91 : 11368 - 11272 Cavalier-Smith T , Allsopp MTEP , Chao EE , Boury-Esnault N , Vacelet J ( 1996 ) Sponge phylogeny, animal monophyly and the origin of the nervous system: 18S rRNA evidence . Can J Zool 74 : 2031 - 2045 Cavalier-Smith T , Chao EE , Lewis R ( 2015a ) Multiple origins of Heliozoa from flagellate ancestors: new cryptist subphylum Corbihelia, superclass Corbistoma, and monophyly of Haptista, Cryptista, Hacrobia and Chromista . Mol Phylogenet Evol 93 : 331 - 362 Cavalier-Smith T , Chao EE , Lewis R ( 2016 ) 187-gene phylogeny of protozoan phylum Amoebozoa reveals a new class (Cutosea) of deep-branching, ultrastructurally unique, enveloped marine Lobosa and clarifies amoeba evolution . Mol Phylogenet Evol 99 : 275 - 296 Cavalier-Smith T , Chao EE , Oates B ( 2004 ) Molecular phylogeny of Amoebozoa and the evolutionary significance of the unikont Phalansterium . Eur J Protistol 40 : 21 - 48 Cavalier-Smith T , Chao EE , Snell EA , Berney C , Fiore-Donno AM , Lewis R ( 2014 ) Multigene eukaryote phylogeny reveals the likely protozoan ancestors of opisthokonts (animals, fungi, choanozoans) and Amoebozoa . Mol Phylogenet Evol 81 : 71 - 85 Cavalier-Smith T , Fiore-Donno A , Chao EE , Kudryavtsev A , Berney C , Snell EA , Lewis R ( 2015b ) Multigene eukaryote phylogeny resolves deep branching of Amoebozoa . Mol Phylogenet Evol 83 : 293 - 304 Cavalier-Smith T , Lewis R , Chao EE , Oates B , Bass D ( 2008 ) Morphology and phylogeny of Sainouron acronematica sp . n. and the ultrastructural unity of Cercozoa. Protist 159 : 591 - 620 Cavalier-Smith T , Oates B , Lewis R , Chao EE , Bass D ( 2009 ) Helkesimastix marina n. sp. (Cercozoa: Sainouroidea superfam . n.) a gliding zooflagellate of novel ultrastructure and unusual ciliary behaviour . Protist 160 : 452 - 479 Chantangsi C , Esson HJ , Leander BS ( 2008 ) Morphology and molecular phylogeny of a marine interstitial tetraflagellate with putative endosymbionts: Auranticordis quadriverberis n. gen. et sp. (Cercozoa) . BMC Microbiol 8 : 123 Chantangsi C , Hoppenrath M , Leander B ( 2010 ) Evolutionary relationships among marine cercozoans as inferred from combined SSU and LSU rDNA sequences and polyubiquitin insertions . Mol Phylogenet Evol 57 : 518 - 527 Chantangsi C , Leander BS ( 2010 ) An SSU rDNA barcoding approach to the diversity of marine interstitial cercozoans, including descriptions of four novel genera and nine novel species . Int J Syst Evol Microbiol 60 : 1962 - 1977 Cohen PA , Knoll AH ( 2012 ) Scale microfossils from the mid-Proterozoic Fifteenmile Group, Yukon Territory . J Paleontol 86 : 775 - 800 Craigie RA , Cavalier-Smith T ( 1982 ) Cell volume and the control of the Chlamydomonas cell cycle . J Cell Sci 54 : 173 - 119 Darwin C ( 1859 ) On the origin of species by means of natural selection, or the preservation of favoured races in the struggle for life . Murray , London Decelle J , Suzuki N , Mahe F , de Vargas C , Not F ( 2012 ) Molecular phylogeny and morphological evolution of the Acantharia (Radiolaria) . Protist 163 : 435 - 450 Derelle R , Torruella G , Klimes V , Brinkmann H , Kim E , Vlcek C , Lang BF , Elias M ( 2015 ) Bacterial proteins pinpoint a single eukaryotic root . Proc Natl Acad Sci U S A 112 : E693 - E699 de Saedeleer H ( 1934 ) Beitrag zur Kenntnis der Rhizopoden: morphologische und systematische Untersuchungen und ein Klassifikationsversuch . Mém Mus Roy Hist Nat Belg 60 : 1 - 112 Deschamps P , Moreira D ( 2009 ) Signal conflicts in the phylogeny of the primary photosynthetic eukaryotes . Mol Biol Evol 261 : 2745 - 2753 Deschamps P , Lara E , Marande W , López-García P , Ekelund F , Moreira D ( 2011 ) Phylogenomic analysis of kinetoplastids supports that trypanosomatids arose from within bodonids . Mol Biol Evol 28 : 53 - 58 Dujardin F ( 1841 ) Histoire naturelle des zoophytes infusoires . Roret , Paris Dumack K , Muller ME , Bonkowski M ( 2016a ) Description of Lecythium terrestris sp. nov. (Chlamydophryidae, Cercozoa), a soil dwelling protist feeding on fungi and algae . Protist 167 : 93 - 105 . https://doi. org/10.1016/j.protis. 2016 . 01 .001 Dumack K , Schuster J , Bass D , Bonkowski M ( 2016b ) A novel lineage of 'naked filose amoebae'; Kraken carinae gen. nov. sp. nov. (Cercozoa) with a remarkable locomotion by disassembly of its cell body . Protist 167 : 268 - 278 . 2016 . 04 . 002 Dumack K , Mylnikov AP , Bonkowski M ( 2017a ) Evolutionary relationship of the scale-bearing Kraken (incertae sedis , Monadofilosa, Cercozoa, Rhizaria): combining ultrastructure data and a two-gene phylogeny . Protist 168 : 362 - 373 . 2017 . 04 .004 Dumack K , Siemensma F , Bonkowski M ( 2017b ) Rediscovery of the testate amoeba genus Penardeugenia (Thaumatomonadida, Imbricatea) . Protist 169 : 29 - 42 . 2017 . 12 .002 Elie M , Nogueira CR , Nédélec A , Trindade RIF , Kenig F ( 2007 ) A red algal bloom in the aftermath of Marinoan snowball earth . Terra Nova 19 : 303 - 308 Eldredge N , Gould SJ ( 1972 ) Punctuated equilibria: an alternative to phyletic gradualism . In: Schopf TJM (ed) Models in Paleobiology. Freeman, Cooper, San Fancisco, pp 305 - 332 Eme L , Sharpe SC , Brown MW , Roger AJ ( 2014 ) On the age of eukaryotes: evaluating evidence from fossils and molecular clocks . Cold Spring Harb Perspect Biol 6 . a016139 Farley CA , Wolf PH , Elston RA ( 1988 ) A long-term study of "microcell" disease in oysters with a description of a new genus, Mikrocytos (g. n.), and two new species, Mikrocytos mackini (sp . n.) and Mikrocytos roughleyi (sp . n). Fish Bull 86 : 581 - 593 Febvre-Chevalier C ( 1990 ) Phylum Actinopoda: class Heliozoa . In: Margulis L, Corliss JO , Melkonian M , Chapman D (eds) Handbook of Protoctista. Jones & Bartlett, Boston, MA, pp 399 - 416 Feulner G , Hallmann A , Kienert H ( 2015 ) Snowball cooling after algal rise . Nat Geosci 8 : 659 - 661 Flakowski J , Bolivar I , Fahrni J , Pawlowski J ( 2005 ) Actin phylogeny of foraminifera . J Foraminiferal Res 35 : 93 - 102 Flakowski J , Bolivar I , Fahrni J , Pawlowski J ( 2006 ) Tempo and mode of spliceosomal intron evolution in actin of foraminifera . J Mol Evol 63 : 30 - 41 Frolov AO , Karpov SA , Mylnikov AP ( 2001 ) The ultrastructure of Procryptobia sorokini (Zhukov) comb. nov. and rootlet homology in kinetoplastids . Protistology 2 : 85 - 95 Gilg IC , Amaral-Zettler A , Countway PD , Moorthi S , Schnetzer A , Caron DA ( 2010 ) Phylogenetic affiliations of mesopelagic Acantharia and acantharian-like environmental 18S rRNA genes off the southern California coast . Protist 161 : 197 - 211 Geimer S , Melkonian M ( 2004 ) The ultrastructure of t he Chlamydomonas reinhardtii basal apparatus: identification of an early marker of radial asymmetry inherent in the basal body . J Cell Sci 117 : 2663 - 2674 Gensel PG ( 2008 ) The earliest land plants . Annu Rev Ecol Evol Syst 39 : 459 - 477 Glöckner G , Hülsmann N , Schleicher M , Noegel AA , Eichinger L , Gallinger C , Pawlowski J , Sierra R , Euteneuer U , Pillet L , Moustafa A , Platzer M , Groth M , Szafranski K , Schliwa M ( 2014 ) The genome of the foraminiferan Reticulomyxa filosa . Curr Biol 24 : 11 - 18 . 2013 . 11 .027 Gomaa F, Mitchell EA , Lara E ( 2013 ) Amphitremida (Poche, 1913) is a new major, ubiquitous labyrinthulomycete clade . PLoS One 8 ( 1 ): e53046. Gong Y , Patterson DJ , Li Y , Hu Z , Sommerfeld M , Chen Y , Hu Q ( 2015 ) Vernalophrys algivore gen . nov., sp. nov. ( Rhizaria: Cercozoa: Vampyrellida), a new algal predator isolated from outdoor mass culture of Scenedesmus dimorphus . Appl Environ Microbiol 81 : 3900 - 3913 Gould SJ , Eldredge N ( 1977 ) Punctuated equilibria: the tempo and mode of evolution reconsidered . Paleobiology 3 : 115 - 151 Gould SJ , Eldredge N ( 1986 ) Punctuated equilibria at the third stage . Syst Zool 35 : 143 - 148 Grant JR , Katz LA ( 2014 ) Building a phylogenomic pipeline for the eukaryotic tree of life - addressing deep phylogenies with genomescale data . PLoS currents 6 . c24b6054aebf3602748ac042ccc8f2e9 Groussin M , Pawlowski J , Yang Z ( 2011 ) Bayesian relaxed clock estimation of divergence times in foraminifera . Mol Phylogenet Evol 61 : 157 - 166 Guilbaud R , Poulton SW , Butterfield NJ ( 2015 ) A global transition to ferruginous conditions in the early Neoproterozoic oceans . Nat Geosci 8 : 466 - 471 Hartikainen H , Stentiford GD , Bateman KS , Berney C , Feist SW , Longshaw M , Okamura B , Stone D , Ward G , Wood C , Bass D ( 2014 ) Mikrocytids are a broadly distributed and divergent radiation of parasites in aquatic invertebrates . Curr Biol 24 : 807 - 812 . https:// 2014 . 02 .033 He D , Fiz-Palacios O , Fu CJ , Fehling J , Tsai CC , Baldauf SL ( 2014 ) An alternative root for the eukaryote tree of life . Curr Biol 24 : 465 - 470 He D , Sierra R , Pawlowski J , Baldauf SL ( 2016 ) Reducing long-branch effects in multi-protein data uncovers a close relationship between Alveolata and Rhizaria . Mol Phylogenet Evol 101 : 1 - 7 Heger TJ , Mitchell EA , Todorov M , Golemansky V , Lara E , Leander BS , Pawlowski J ( 2010 ) Molecular phylogeny of euglyphid testate amoebae (Cercozoa: Euglyphida) suggests transitions between marine supralittoral and freshwater/terrestrial environments are infrequent . Mol Phylogenet Evol 55 : 113 - 122 . ympev. 2009 . 11 .023 Hess S , Sausen S , Melkonian M ( 2012 ) Shedding light on vampires: the phylogeny of vampyrellid amoebae revisited . PLoS One 7 ( 2 ): e31165. Hess S , Melkonian M ( 2013 ) The mystery of clade X: Orciraptor gen. nov. and Viridiraptor gen. nov. are highly specialised, algivorous amoeboflagellates (Glissomonadida, Cercozoa) . Protist 164 : 706 - 747 Hess S , Melkonian M ( 2014 ) Ultrastructure of the algivorous amoeboflagellate Viridiraptor invadens (Glissomonadida, Cercozoa) . Protist 165 : 605 - 635 Hibberd DJ ( 1979 ) The structure and phylogenetic significance of the flagellar transition region in the chlorophyll c-containing algae . BioSystems 11 : 243 - 267 Hollande A , Cachon J , Cachon M ( 1970 ) La signification de la membrane capsulaire des Radiolaires et ses rapports avec la plasmalemme et les membranes du réticulum endoplasmique . Affinités entre Radiolaires , Héliozoaires et Péridiniens. Protistologica 6 : 311 - 318 Hood AS , Wallace MW ( 2015 ) Extreme ocean anoxia during the late Cryogenian recorded in reefal carbonates of southern Australia . Precambrian Res 261 : 96 - 111 Hoppenrath M , Leander BS ( 2006a) Dinoflagellate, euglenid, or cercomonad? The ultrastructure and molecular phylogenetic position of Protaspis grandis n. sp . J Eukaryot Microbiol 53 : 327 - 342 Hoppenrath M , Leander BS ( 2006b ) Ebriid phylogeny and the expansion of the Cercozoa . Protist 157 : 279 - 290 Howe AT ( 2008 ) Phylogeny, biodiversity and evolution of the sarcomonad Cercozoa (Protozoa) . D. Phil. Thesis , Oxford University Howe AT , Bass D , Vickerman K , Chao EE , Cavalier-Smith T ( 2009 ) Phylogeny, taxonomy, and astounding genetic diversity of Glissomonadida ord. nov., the dominant gliding zooflagellates in soil (Protozoa: Cercozoa) . Protist 160 : 159 - 189 Howe AT , Bass D , Scoble JM , Lewis R , Vickerman K , Arndt H , CavalierSmith T (2011a) Novel cultured protists identify deep-branching environmental DNA clades of Cercozoa: new genera Tremula, Micrometopion , Minimassisteria, Nudifila, Peregrinia. Protist 162 : 332 - 372 Howe AT , Bass D , Chao EE , Cavalier-Smith T ( 2011b ) New genera, species and improved phylogeny of Glissomonadida (Cercozoa) . Protist 162 : 710 - 722 Ishitani Y , Ujilé Y , de Vargas C , Not P , Takahashi K ( 2012 ) Phylogenetic relationships and evolutionary patterns of the order Collodaria (Radiolaria) . PLoS One 7 ( 5 ):e35775. journal.pone.0035775 Janouškovec J , Tikhonenkov DV , Mikhailov KV , Simdyanov TG , Aleoshin VV , Mylnikov AP , Keeling PJ ( 2013 ) Colponemids represent multiple ancient alveolate lineages . Curr Biol 23 : 2546 - 2552 . 2013 . 10 .062 Janouškovec J , Tikhonenkov DV , Burki F , Howe AT , Rohwer FL , Mylnikov AP , Keeling PJ ( 2017 ) A new lineage of eukaryotes illuminates early mitochondrial genome reduction . Curr Biol 27 : 3717 - 3724 e3715. 2017 . 10 .051 Kamikawa R , Shiratori T , Ishida K , Miyashita H , Roger AJ ( 2016 ) Group II intron-mediated trans-splicing in the gene-rich mitochondrial genome of an enigmatic eukaryote, Diphylleia rotans . Genome Biol Evol 8 : 458 - 466 . Karpov SA , Bass D , Mylnikov AP , Cavalier-Smith T ( 2006 ) Molecular phylogeny of Cercomonadidae and kinetid patterns of Cercomonas and Eocercomonas gen . nov. (Cercomonadida , Cercozoa). Protist 157 : 125 - 158 Karpov SA , Ekelund F , Moestrup Ø ( 2003 ) Katabia gromovi nov . gen., nov. sp. -a new soil flagellate with affinities to Heteromita (Cercomonadida) . Protistology 3 : 30 - 41 Keeling PJ ( 2013 ) The number, speed, and impact of plastid endosymbioses in eukaryotic evolution . Annu Rev Plant Biol 64 : 583 - 607 Kleemann G , Poralla K , Englert G , Kjøsen H , Liaaen-Jensen S , Neunlist S , Rohmer M ( 1990 ) Tetrahymanol from the phototrophic bacterium Rhodopseudomonas palustris: first report of a gammacerane triterpene from a prokaryote . J Gen Microbiol 136 : 2551 - 2553 Knoll AH ( 2014 ) Paleobiological perspectives on early eukaryotic evolution . Cold Spring Harb Perspect Biol 6 : a016121 Krabberød AK , Bråte J , Dolven JK , Ose RF , Klaveness D , Kristensen T , Bjørklund KR , Shalchian-Tabrizi K ( 2011 ) Radiolaria divided into Polycystina and Spasmaria in combined 18S and 28S rDNA phylogeny . PLoS One 6 ( 8 ):e23526. pone.0023526 Krabberød AK , Orr RJS , Bråte J , Kristensen T , Bjorklund KR , ShalchianTabrizi K ( 2017 ) Single cell transcriptomics, mega-phylogeny, and the genetic basis of morphological innovations in Rhizaria . Mol Biol Evol 34 : 1557 - 1573 . Kühn S , Medlin L , Eller G ( 2004 ) Phylogenetic position of the parasitoid nanoflagellate Pirsonia inferred from nuclear-encoded small subunit ribosomal DNA and a description of Pseudopirsonia n. gen. and Pseudopirsonia mucosa (Drebes) comb . nov. Protist 155 : 143 - 156 Lankester ER ( 1885 ) Protozoa . In: Baynes TS (ed) The encyclopedia Britannica, 9th edn . J.M. Stoddard Co., Ltd , Philadelphia, pp 831 - 865 Lartillot N , Philippe H ( 2004 ) A Bayesian mixture model for across-site heterogeneities in the amino-acid replacement process . Mol Biol Evol 21 : 1095 - 1109 Lartillot N , Rodrigue N , Stubbs D , Richer J ( 2013 ) PhyloBayes MPI . Phylogenetic reconstruction with infinite mixtures of profiles in a parallel environment . Syst Biol Doi 62 : 611 - 615 . 1093/sysbio/syt022 Lasek-Nesselquist E , Gogarten JP ( 2013 ) The effects of model choice and mitigating bias on the ribosomal tree of life . Mol Phyl Evol 69 : 17 - 38 Lee WJ ( 2002 ) Some free-living heterotrophic flagellates from marine sediments of Inchon and Ganghwa Island, Korea . Korean J Biol Sci 6 : 125 - 143 Lee WJ , Patterson DJ ( 2000 ) Heterotrophic flagellates (Protista) from marine sediments of Botany Bay, Australia . J Nat History 34 : 483 - 562 Le Heron DP ( 2012 ) The Cryogenian record of glaciation and deglaciation in South Australia . Sed Geol 243 -4 : 57 - 69 Lenton TM , Boyle RA , Poulton SW , Shields-Zhou GA , Butterfield DA ( 2015 ) Co-evolution of eukaryotes and ocean oxygenation in the Neoproterozoic era . Nat Geosci 7 : 257 - 265 Lepage T , Bryant D , Philippe H , Lartillot N ( 2007 ) A general comparison of relaxed molecular clock models . Mol Biol Evol 24 : 2669 - 2680 Longet D , Pawlowski J ( 2007 ) Higher-level phylogeny of foraminifera inferred from the RNA polymerase II (RPB1) gene . Eur J Protistol 43 : 171 - 177 Longet D , Archibald JM , Keeling PJ , Pawlowski J ( 2003 ) Foraminifera and Cercozoa share a common origin according to RNA polymerase II phylogenies . Int J Syst Evol Microbiol 53 : 1735 - 1739 Longet D , Burki F , Flakowski J , Berney C , Polet S , Fahrni J , Pawlowski J ( 2004 ) Multigene evidence for close evolutionary relations between Gromia and foraminifera . Acta Protozool 43 : 303 - 311 Maier I , Parodi E , Westermeier R , Müller DG ( 2000 ) Maullinia ectocarpii gen . et sp. nov. ( Plasmodiophorea), an intracellular parasite in Ectocarpus siliculosus (Ectocarpales, Phaeophyceae) and other filamentous brown algae . Protist 151 : 225 - 238 Maletz J ( 2011 ) Radiolarian skeletal structures and biostratigraphy in the early Palaeozoic. (Cambrian-Ordovician) . Palaeoworld 20 : 116 - 133 Mayr E ( 1954 ) Change of genetic environment and evolution . In: Huxley J (ed) Evolution as a process . Allen and Unwin , London, pp 157 - 180 Mayr E ( 1960 ) The emergence of evolutionary novelties . In S Tax ed. Evolution after Darwin . vol 1 , pp 349 - 380 Mayr E (1988a) Does microevolution explain macroevolution . In: Mayr E (ed) Towards a new philosophy of biology: observations of an evolutionist . Harvard University Press, Cambridge, pp 403 - 422 Mayr E (1988b) Speciational evolution through punctuated equilibria . In: Mayr E (ed) Towards a new philosophy of biology: observations of an evolutionist . Harvard University Press, Cambridge, pp 457 - 488 Mignot J-P , Brugerolle G ( 1975 ) Étude ultrastructurale du flagellé phagotrophe Colponema loxodes Stein . Protistologica 11 : 429 - 444 Mikrjukov KA , Mylnikov AP ( 1998 ) A study of the fine structure and the life cycle of Gymnophrys cometa Cienkowski, 1876 (Gymnophryea classis nov.) with taxonomical remarks to the amoebo-flagellated genera Gymnophrys and Borkovia . Acta Protozool 37 : 179 - 189 Mikrjukov K , Siemensma FJ , Patterson DJ ( 2002 ) Phylum Heliozoa . In: Lee J , Leedale GF , Bradbury P (eds) An illustrated guide to the Protozoa, vol II, 2nd edn . Society of Protozoologists, Lawrence, pp 860 - 871 Mylnikov AP , Tikhonenkov DV ( 2009 ) The new alveolate carnivorous flagellate (Colponema marisrubri sp . n., Colponemida , Alveolata) from the Red Sea . Zoolog Zhur 88 : 1 - 7 (in Russian) Myl'nikova ZM , Myl'nikov AP ( 2010 ) Biolgy and morphology of freshwater rapacious flagellate Colponema aff. loxodes Stein (Colponema, Alveolata) . Inland Water Biology 3 : 21 - 26 Mylnikov AP , Karpov SA ( 2004 ) Review of the diversity and taxonomy of cercomonads . Protistology 3 : 201 - 217 Nakayama T , Ishida K ( 2009 ) Another acquisition of a primary photosynthetic organelle is underway in Paulinella chromatophora . Curr Biol 19 : 7 Nikolaev SI , Berney C , Fahrni J , Mylnikov AP , Aleshin VV , Petrov NB , Pawlowski J ( 2003 ) Gymnophrys cometa and Lecythium sp. are core Cercozoa: evolutionary imnplications . Acta Protozool 42 : 183 - 190 Nikolaev SI , Berney C , Fahrni JF , Bolivar I , Polet S , Mylnikov AP , Aleshin VV , Petrov NB , Pawlowski J ( 2004 ) The twilight of Heliozoa and rise of Rhizaria, an emerging supergroup of amoeboid eukaryotes . Proc Natl Acad Sci U S A 101 : 8066 - 8071 Okamoto N , Chantangsi C , Horak A , Leander BS , Keeling PJ ( 2009 ) Molecular phylogeny and description of the novel katablepharid Roombia truncata gen . et sp. nov., and establishment of the Hacrobia taxon nov . PLoS One 4 : e7080 Parfrey LW , Grant J , Tekle YI , Lasek-Nesselquist E , Morrison HG , Sogin ML , Patterson DJ , Katz LA ( 2010 ) Broadly sampled multigene analyses yield a well-resolved eukaryotic tree of life . Syst Biol 59 : 518 - 533 Parfrey LW , Lahr DJ , Knoll AH , Katz LA ( 2011 ) Estimating the timing of early eukaryotic diversification with multigene molecular clocks . Proc Natl Acad Sci U S A 108 : 13624 - 13629 Patterson DJ , Zölffel M ( 1991 ) Heterotrophic flagellates of uncertain taxonomic position . In: Patterson DJ , Larsen J (eds) The biology of free-living heterotrophic flagellates . Clarendon press, Oxford, pp 427 - 476 Pawlowski J ( 2008 ) The twilight of Sarcodina: a molecular perspective on the polyphyletic origin of amoeboid protists . Protistology 5 : 281 - 302 Pawlowski J , Berney C ( 2003 ) Episodic evolution of nuclear small subunit ribosomal RNA gene in the stem lineage of foraminifera . In: Donoghue PC , Smith MP (eds) Telling the evolutionary time: molecular clocks and the fossil record . Systematics Association Special Volume No. 66 . Taylor & Francis, London, pp 107 - 118 Pawlowski J , Burki F ( 2009 ) Untangling the phylogeny of amoeboid protists . J Euk Microbiol 56 : 16 - 25 Pawlowski J , Holzmann M , Tyszka J ( 2013 ) New supraordinal classification of foraminifera: molecules meet morphology . Marine Micropaleontol 100 : 1 - 10 Payne JL , Jost AB , Wang SC , Skotheim JM ( 2013 ) A shift in the longterm mode of foraminiferan size evolution caused by the endPermian mass extinction . Evolution 67 : 816 - 827 . 10.1111/j.1558- 5646 . 2012 . 01807 .x Polet S , Berney C , Fahrni J , Pawlowski J ( 2004 ) Small-subunit ribosomal RNA gene sequences of Phaeodarea challenge the monophyly of Haeckel's Radiolaria . Protist 155 : 53 - 63 Raymann K , Brochier-Armanet C , Gribaldo S ( 2015 ) The two-domain tree of life is linked to a new root for the archaea . Proc Natl Acad Sci U S A 112 : 6670 - 6675 Ren R , Sun Y , Zhao Y , Geiser D , Ma H , Zhou X ( 2016 ) Phylogenetic resolution of deep eukaryotic and fungal relationships using highly conserved low-copy nuclear genes . Genome Biol Evol 8 : 2683 - 2701 . Rooney AD , Straus JV , Brandon AD , MacDonald FA ( 2015 ) A Cryogenian chronology: two long-lasting synchronous Neoproterozoic glaciations . Geology 43 : 459 - 462 Roure B , Baurain D , Philippe H ( 2013 ) Impact of missing data on phylogenies inferred from empirical phylogenomic data sets . Mol Biol Evol 30 : 197 - 214 Roy RS , Price DC , Schliep A , Cai G , Korobeynikov A , Yoon HS , Yang EC , Bhattacharya D ( 2014 ) Single cell genome analysis of an uncultured heterotrophic stramenopile . Sci Rep 4 : 4780 Ruggiero M , Gordon D , Bailly N , Bourgoin T , Brusca R , Cavalier-Smith T , Guiry M , Kirk M , Orrell T ( 2015 ) A higher-level classification of all living organisms . PLoS One 10 ( 4 ):e0119248. 1371/journal.pone.0119248 Schintele R , Brocks JJ ( 2017 ) Paleoecology of Neoproterozoic hypersaline environments: biomarker evidence for haloarchaea, methanogens, and cyanobacteria . Geobiology 15 : 641 - 663 Scoble JM , Cavalier-Smith T ( 2014 ) Scale evolution, sequence phylogeny, and taxonomy of thaumatomonad Cercozoa: 11 new species and new genera Scutellomonas, Cowlomonas, Thaumatospina and Ovaloplaca . Europ J Protistol 50 : 270 - 313 Shiratori T , Yabuki A , Ishida K ( 2012 ) Esquamula lacrimiformis n. gen., n. sp., a new member of thaumatomonads that lacks siliceous scales . J Euk Microbiol 59 : 527 - 536 Shiratori T , Yokoyama A , Ishida K ( 2014 ) Phylogeny, ultrastructure, and flagellar apparatus of a new marimonad flagellate Abollifer globosa sp . nov. (Imbricatea , Cercozoa). Protist 165 : 808 - 824 . https://doi. org/10.1016/j.protis. 2014 . 10 .003 Sierra R , Matz MV , Aglyamova G , Pillet L , Decelle J , Not F , de Vargas C , Pawlowski J ( 2013 ) Deep relationships of Rhizaria revealed by phylogenomics: a farewell to Haeckel's Radiolaria . Mol Phylogenet Evol 67 : 53 - 59 Sierra R , Canas-Duarte SJ , Burki F , Schwelm A , Fogelqvist J , Dixelius C , Gonzalez-Garcia LN , Gile GH , Slamovits CH , Klopp C , Restrepo S , Arzul I , Pawlowski J ( 2016 ) Evolutionary origins of rhizarian parasites . Mol Biol Evol 33 : 980 - 983 Simpson GC ( 1944 ) Tempo and mode in evolution . Columbia University Press, New York Simpson GC ( 1953 ) The major features of evolution . Columbia University Press, New York Stackebrandt E , Woese CR ( 1981 ) The evolution of prokaryotes. Molecular and cellular aspects of microbial evolution . Cambridge University Press, Cambridge, pp 1 - 32 Stamatakis A , Ludwig T , Meier H ( 2005 ) RAxML-III: a fast program for maximum likelihood-based inference of large phylogenetic trees . Bioinformatics 21 : 456 - 463 Stamatakis A , Hoover P , Rougemont J ( 2008 ) A rapid bootstrap algorithm for the RAxML web servers . Syst Biol 57 : 758 - 771 . https:// Streng M , Babcock LE , Hollingsworth JS ( 2005 ) Agglutinated protists from the lower Cambrian of Nevada . J Paleontol 79 : 1214 - 1218 Swale EMF ( 1969 ) The fine structure of a species of the amoeboflagellate Pseudospora Cienk . Arch Mikrobiol 67 : 71 - 90 Takahashi Y , Yoshida M , Inouye I , Watanabe MM ( 2014 ) Diplophrys mutabilis sp. nov., a new member of Labyrinthulomycetes from freshwater habitats . Protist 165 : 50 - 65 Tekle YI , Grant J , Cole JC , Nerad TA , Anderson OR , Patterson DJ , Katz LA ( 2007 ) A multigene analysis of Corallomyxa tenera sp . nov. suggests its membership in a clade that includes Gromia, Haplosporidia and foraminifera . Protist 158 : 457 - 472 Tziperman E , Halevy I , Johnston DT , Knoll AH , Schrag DP ( 2011 ) Biologically induced initiation of Neoproterozoic snowball-Earth events . Proc Natl Acad Sci U S A 108 : 15091 - 15096 Vickerman K , Appleton PL , Clarke KJ , Moreira D ( 2005 ) Aurigamonas solis n. gen., n. sp., a soil-dwelling predator with unusual helioflagellate organisation and belonging to a novel clade within the Cercozoa . Protist 156 : 335 - 354 Vørs N ( 1992 ) Heterotrophic amoebae, flagellates and Heliozoa from the Tvarminne area , Gulf of Finland, in 1988-1990. Ophelia 36 : 1 - 109 Ward GM , Neuhauser S , Groben R , Ciaghi S , Berney C , Romac S , Bass D ( 2018 ) Environmental sequencing fills the gap between parasitic haplosporidians and free-living giant amoebae . J Eukaryot Microbiol . Won M , Below R ( 1999 ) Cambrian Radiolaria from the Georgina Basin , Queensland, Australia. Micropal 45 : 325 - 363 Wylezich C , Meisterfeld R , Meisterfeld S , Schlegel M ( 2002 ) Phylogenetic analyses of small subunit ribosomal RNA coding regions reveal a monophyletic lineage of euglyphid testate amoebae (Order Euglyphida) . J Eukaryot Microbiol 49 : 108 - 118 Yabuki A , Ishida K ( 2011 ) Mataza hastifera n . g., n. sp.: a possible new lineage in the Thecofilosea (Cercozoa) . J Eukaryot Microbiol 58 : 94 - 102 Yabuki A , Ishida KI ( 2018 ) An orphan protist Quadricilia rotundata finally finds its phylogenetic home in Cercozoa. J Eukaryot Microbiol . Yang EC , Boo SM , Bhattacharya D , Saunders GW , Knoll AH , Fredericq S , Graf L , Yoon HS ( 2016 ) Divergence time estimates and the evolution of major lineages in the florideophyte red algae . Sci Rep 6 : 21361 . Ye Q , Tong J , Xiao S , Zhu S , An Z , Tian L , Hu J ( 2015 ) The survival of benthic macroscopic phototrophs on a Neoproterozoic snowball Earth . Geology 43 : 5067 - 5510 Yuasa T , Takahashi O , Dolven JK , Mayama S , Matsuoka A , Honda D , Bjørklund KR ( 2006 ) Phylogenetic position of the small solitary phaeodarians (Radiolaria) based on 18S rDNA sequences by single cell PCR analysis . Marine Micropaleontol 59 : 104 - 114

This is a preview of a remote PDF:

Thomas Cavalier-Smith, Ema E. Chao, Rhodri Lewis. Multigene phylogeny and cell evolution of chromist infrakingdom Rhizaria: contrasting cell organisation of sister phyla Cercozoa and Retaria, Protoplasma, 2018, 1-58, DOI: 10.1007/s00709-018-1241-1