Europ. J. Protistol. 39, 338���348 (2003) �� Urban & Fischer Verlag http://www.urbanfischer.de/journals/ejp Protist phylogeny and the high-level classification of Protozoa Thomas Cavalier-Smith Department of Zoology, University of Oxford, South Parks Road, Oxford, OX1 3PS, UK E-mail: tom.cavalier-smith@zoo.ox.ac.uk Received 1 September 2003 29 September 2003. Accepted: 29 September 2003 Protist large-scale phylogeny is briefly reviewed and a revised higher classification of the kingdom Pro- tozoa into 11 phyla presented. Complementary gene fusions reveal a fundamental bifurcation among eu- karyotes between two major clades: the ancestrally uniciliate (often unicentriolar) unikonts and the an- cestrally biciliate bikonts, which undergo ciliary transformation by converting a younger anterior cilium into a dissimilar older posterior cilium. Unikonts comprise the ancestrally unikont protozoan phylum Amoebozoa and the opisthokonts (kingdom Animalia, phylum Choanozoa, their sisters or ancestors and kingdom Fungi). They share a derived triple-gene fusion, absent from bikonts. Bikonts contrastingly share a derived gene fusion between dihydrofolate reductase and thymidylate synthase and include plants and all other protists, comprising the protozoan infrakingdoms Rhizaria [phyla Cercozoa and Re- taria (Radiozoa, Foraminifera)] and Excavata (phyla Loukozoa, Metamonada, Euglenozoa, Percolozoa), plus the kingdom Plantae [Viridaeplantae, Rhodophyta (sisters) Glaucophyta], the chromalveolate clade, and the protozoan phylum Apusozoa (Thecomonadea, Diphylleida). Chromalveolates comprise kingdom Chromista (Cryptista, Heterokonta, Haptophyta) and the protozoan infrakingdom Alveolata [phyla Cilio- phora and Miozoa (= Protalveolata, Dinozoa, Apicomplexa)], which diverged from a common ancestor that enslaved a red alga and evolved novel plastid protein-targeting machinery via the host rough ER and the enslaved algal plasma membrane (periplastid membrane).The branching order of the five bikont groups is uncertain: Plantae may be sisters of or ancestral to chromalveolates (jointly designated corti- cates as they share cortical alveoli) Rhizaria and Excavata (jointly cabozoa) are probably sisters if the formerly green algal plastid of euglenoids and chlorarachneans (Cercozoa) was enslaved in a single event in their common ancestor. Apusozoa may be sisters of Excavata and centrohelid heliozoa may be sisters to Haptophyta. Key words: Protozoa protist classification phylogeny bikonts unikonts. plasmodial in the trophic state, are derived from bilateral animal ancestors by the loss of gut and nervous system (Okamura et al. 2002) and that the anaerobic microsporidia evolved from aerobic fila- mentous zygomycotine fungi (Keeling 2003) as postulated (Cavalier-Smith 2000a). Such phyloge- netically deceptive evolution by simplification and character loss has occurred not only in such para- 0932-4739/03/39/04-338 $ 15.00/0 Protists are polyphyletic Eukaryotes comprise the basal kingdom Proto- zoa and four derived kingdoms: Animalia, Fungi, Plantae, and Chromista (Cavalier-Smith 1998). Protists (unicellular eukaryotes) can no longer be considered a taxon (e.g. a kingdom) as they are found in all five kingdoms and are undoubtedly polyphyletic. This is most strikingly shown by the recent evidence that Myxozoa, most of which are

Phylogeny and classification of Protozoa 339 sites but also in free-living protists. Molecular evi- dence now implies that all known groups of anaer- obic, apparently amitochondrial, protists evolved by the loss of mitochondrial genomes, cy- tochromes, and oxidative phosphorylation and the conversion of mitochondria into double-mem- braned organelles of different function: hy- drogenosomes or mitosomes (Roger 1999 Silber- man et al. 2002) ��� these retain their ancestral mito- chondrial mechanisms of membrane heredity and protein targeting: (Cavalier-Smith 2004a Williams et al. 2002 Embley et al. 2003 van der Giezen et al. 2003). Likewise it is almost certain that all non-cil- iated protists {e.g. many amoebae (Cavalier-Smith et al. 2004), yeasts, Corallochytrium (Cavalier- Smith and Chao 2003a), Myxozoa, microsporidia, Blastocystis} evolved ultimately from ciliated an- cestors by losing cilia (= eukaryotic flagella) and centrioles (= basal bodies). A new systematic synthesis These new insights have come not just from molecular sequence studies but by integrating them with numerous other lines of evidence, ge- netic, structural and biochemical. The classical view developed over two centuries that reliance on a single line of evidence or character is often very misleading for phylogeny and systematics is at last penetrating the previously over-dogmatic and over-self-confident field of molecular systematics. There really ought to be no such field, for good systematics should be fully integrative of all avail- able evidence. ���Molecular systematics��� that con- centrates on trees from just one molecule and ig- nores other evidence is poor systematics. Al- though there have long been conclusive theoretical reasons for thinking that sequence trees can some- times be profoundly misleading and all too easily misinterpreted (Felsenstein 1978) and that riboso- mal rRNA is certainly not a molecular clock (Cav- alier-Smith 1980), the recent spread of a more criti- cal approach to protist molecular phylogenies owes much to the balanced and integrative per- spective of Andr�� Adoutte (Baroin et al. 1988 Philippe and Adoute 1996) and critical analyses by his former collaborators (Philippe and Adoutte 1998 Philippe 2000 Philippe and Germot 2000 Philippe et al. 2000a Philippe et al. 2000b Lopez et al. 2002). It is now widely, but by no means uni- versally, accepted that single-gene trees often lack the resolution to group together taxa that really are related, sometimes group together those that are not and can be profoundly misleading about where the root of a subtree really lies. Despite these problems, the overall evidence now allows us to group protists into a relatively small number of phyla (25, counting all four fungal phyla as including protists (Cavalier-Smith 1998), of a total of 48 eukaryotic phyla), most now estab- lished with reasonable confidence as monophyletic - mainly holophyletic, though one is certainly pa- raphyletic (i.e. Chlorophyta, one of the three plant phyla containing protists: Cavalier-Smith 1998) and two probably are (Archemycota, Loukozoa). The present paper summarises this evidence and outlines my current interpretation of relationships among the 11 phyla of the necessarily paraphyletic kingdom Protozoa and the four derived (holo- phyletic) kingdoms (Fig. 1). In this revised system for kingdom Protozoa (see appendix) the phyla Choanozoa and Amoebozoa (both I suspect holo- phyletic, but either or both may be paraphyletic as shown on some trees: resolution is insufficient to decide either way ��� Cavalier-Smith and Chao 2003a Cavalier-Smith et al. 2004) are grouped as the undoubtedly paraphyletic subkingdom Sarco- mastigota, from which animals and fungi indepen- dently evolved (Cavalier-Smith 2000a King et al. 2003). The other nine protozoan phyla (all but Loukozoa probably holophyletic) constitute the paraphyletic subkingdom Biciliata from which the holophyletic kingdoms Plantae and Chromista evolved. Biciliata comprise three probably holo- phyletic infrakingdoms (Alveolata, Rhizaria, Ex- cavata) plus the phylum Apusozoa, which may have affinities with Excavata or Rhizaria or be more deeply branching. Bikonts, unikonts and the eukaryotic root Plantae, Chromista, and Biciliata are all clearly ancestrally biciliate and together constitute a clade designated the bikonts (Cavalier-Smith 2002). A major shared derived character for all three groups (not yet clearly demonstrated for Rhizaria) is cil- iary transformation in which the anterior cilium/centriole and its associated roots are always the first formed in the next cell cycle they undergo often marked changes in structure and function to become the corresponding posterior organelles