Biol. Rev. (1998), 73, pp.

203266

Printed in the United Kingdom # Cambridge Philosophical Society

203

A revised six-kingdom system of life

T. CAVALIER-SMITH
Evolutionary Biology Programme, Canadian Institute for Advanced Research, Department of Botany, University of British Columbia,
Vancouver, BC, Canada V6T 1Z4
(Received 27 March 1996 ; revised 15 December 1997 ; accepted 18 December 1997)

ABSTRACT
A revised six-kingdom system of life is presented, down to the level of infraphylum. As in my 1983 system
Bacteria are treated as a single kingdom, and eukaryotes are divided into only five kingdoms : Protozoa,
Animalia, Fungi, Plantae and Chromista. Intermediate high level categories (superkingdom, subkingdom,
branch, infrakingdom, superphylum, subphylum and infraphylum) are extensively used to avoid splitting
organisms into an excessive number of kingdoms and phyla (60 only being recognized). The two zoological
kingdoms, Protozoa and Animalia, are subject to the International Code of Zoological Nomenclature, the
kingdom Bacteria to the International Code of Bacteriological Nomenclature, and the three botanical
kingdoms (Plantae, Fungi, Chromista) to the International Code of Botanical Nomenclature. Circumscriptions of the kingdoms Bacteria and Plantae remain unchanged since Cavalier-Smith (1981). The kingdom
Fungi is expanded by adding Microsporidia, because of protein sequence evidence that these amitochondrial
intracellular parasites are related to conventional Fungi, not Protozoa. Fungi are subdivided into four phyla
and 20 classes ; fungal classification at the rank of subclass and above is comprehensively revised. The
kingdoms Protozoa and Animalia are modified in the light of molecular phylogenetic evidence that Myxozoa
are actually Animalia, not Protozoa, and that mesozoans are related to bilaterian animals. Animalia are
divided into four subkingdoms : Radiata (phyla Porifera, Cnidaria, Placozoa, Ctenophora), Myxozoa,
Mesozoa and Bilateria (bilateral animals : all other phyla). Several new higher level groupings are made in
the animal kingdom including three new phyla : Acanthognatha (rotifers, acanthocephalans, gastrotrichs,
gnathostomulids), Brachiozoa (brachiopods and phoronids) and Lobopoda (onychophorans and tardigrades), so only 23 animal phyla are recognized. Archezoa, here restricted to the phyla Metamonada and
Trichozoa, are treated as a subkingdom within Protozoa, as in my 1983 six-kingdom system, not as a separate
kingdom. The recently revised phylum Rhizopoda is modified further by adding more flagellates and
removing some rhizopods and is therefore renamed Cercozoa. The number of protozoan phyla is reduced
by grouping Mycetozoa and Archamoebae (both now infraphyla) as a new subphylum Conosa within the
phylum Amoebozoa alongside the subphylum Lobosa, which now includes both the traditional aerobic
lobosean amoebae and Multicilia. Haplosporidia and the (formerly microsporidian) metchnikovellids are
now both placed within the phylum Sporozoa. These changes make a total of only 13 currently recognized
protozoan phyla, which are grouped into two subkingdoms : Archezoa and Neozoa ; the latter is modified in
circumscription by adding the Discicristata, a new infrakingdom comprising the phyla Percolozoa and
Euglenozoa). These changes are discussed in relation to the principles of megasystematics, here defined as
systematics that concentrates on the higher levels of classes, phyla, and kingdoms. These principles also make
it desirable to rank Archaebacteria as an infrakingdom of the kingdom Bacteria, not as a separate kingdom.
Archaebacteria are grouped with the infrakingdom Posibacteria to form a new subkingdom, Unibacteria,
comprising all bacteria bounded by a single membrane. The bacterial subkingdom Negibacteria, with
separate cytoplasmic and outer membranes, is subdivided into two infrakingdoms : Lipobacteria, which lack
lipopolysaccharide and have only phospholipids in the outer membrane, and Glycobacteria, with
lipopolysaccharides in the outer leaflet of the outer membrane and phospholipids in its inner leaflet. This
primary grouping of the 10 bacterial phyla into subkingdoms is based on the number of cell-envelope
membranes, whilst their subdivision into infrakingdoms emphasises their membrane chemistry ; definition of
the negibacterial phyla, five at least partly photosynthetic, relies chiefly on photosynthetic mechanism and
cell-envelope structure and chemistry corroborated by ribosomal RNA phylogeny. The kingdoms Protozoa
and Chromista are slightly changed in circumscription by transferring subphylum Opalinata (classes
Opalinea, Proteromonadea, Blastocystea cl. nov.) from Protozoa into infrakingdom Heterokonta of the

T. Cavalier-Smith

204

kingdom Chromista. Opalinata are grouped with the subphylum Pseudofungi and the zooflagellate
Developayella elegans (in a new subphylum Bigyromonada) to form a new botanical phylum (Bigyra) of
heterotrophs with a double ciliary transitional helix, making it necessary to abandon the phylum name
Opalozoa, which formerly included Opalinata. The loss of ciliary retronemes in Opalinata is attributed to
their evolution of gut commensalism. The nature of the ancestral chromist is discussed in the light of recent
phylogenetic evidence.
Key words : Megasystematics, Bacteria, Protozoa, Archezoa, Fungi, Animalia, Plantae, Chromista, Eomycota,
Mycetozoa.
CONTENTS
I. Introduction ............................................................................................................................
II. Philosophic preliminaries .........................................................................................................
(1) Darwinian evolutionary classification contrasted with Hennigian cladification................
(2) Naming taxa and clades ...................................................................................................
(3) Principles of ranking taxa .................................................................................................
(4) The need to weigh and integrate phylogenetic evidence from diverse sources .................
III. In defence of Bacteria : the sole primary kingdom of life.........................................................
(1) The concept of a bacterium (synonym : prokaryote) ........................................................
(2) Changing views on Archaebacteria...................................................................................
(3) The importance of cell structure in bacterial classification...............................................
(4) Characters important in the high-level classification of Bacteria ......................................
(5) Bacterial subkingdoms and infrakingdoms ........................................................................
IV. Protozoa, the basal eukaryotic kingdom..................................................................................
(1) Status of Archezoa, early diverging amitochondrial eukaryotes .......................................
(a) Changes in circumscription of the Archezoa ..............................................................
(2) The protozoan subkingdoms : Archezoa and Neozoa........................................................
(3) Demarcation between the two zoological kingdoms : Protozoa and Animalia ..................
(4) Classification of the Neozoa ..............................................................................................
(a) The infrakingdom Sarcomastigota..............................................................................
(b) The infrakingdom Alveolata.......................................................................................
(c) The infrakingdom Actinopoda....................................................................................
V. The kingdom Animalia and its 23 phyla.................................................................................
(1) Radiata, the ancestral animal subkingdom.......................................................................
(2) The derived subkingdom Mesozoa....................................................................................
(3) The number of animal phyla ............................................................................................
(4) A broadened phylum Annelida.........................................................................................
(5) The pseudocoelomate phyla Nemathelminthes and Acanthognatha phyl. nov. ...............
(6) The new phyla Brachiozoa and Lobopoda .......................................................................
(7) Phylogenetic assumptions behind the new bilaterian infrakingdoms and superphyla.......
(8) New animal subphyla and infraphyla...............................................................................
VI. The kingdom Fungi and its four phyla ...................................................................................
(1) Circumscription of the Fungi ............................................................................................
(2) The trichomycete origin of microsporidia .........................................................................
(3) Revision of higher fungal classification .............................................................................
VII. The kingdom Plantae and its five phyla .................................................................................
(1) New red algal subphyla ....................................................................................................
(2) New chlorophyte infraphyla..............................................................................................
VIII. Chromista, the third botanical kingdom, and its five phyla....................................................
(1) Multiple losses of chloroplasts by chromists ......................................................................
(2) Transfer of Opalinata to Chromista : the new heterokont phylum Bigyra........................
(3) The origin of ciliary retronemes and the nature of the ancestral chromist.......................
(4) Mitochondrial and cell-surface evolution in chromists......................................................
(5) Retroneme loss in Opalinata.............................................................................................
(6) Are there other unidentified chromists lurking within the kingdom Protozoa ? ................

205
210
210
213
214
215
215
215
216
217
220
222
226
226
226
229
231
234
234
238
238
239
239
239
239
240
240
241
241
243
244
244
245
245
249
249
251
251
252
253
254
255
256
257

A revised six-kingdom system of life

IX.
X.
XI.
XII.

205

Envoi .......................................................................................................................................
Conclusions ..............................................................................................................................
Acknowledgements ..................................................................................................................
References................................................................................................................................

I. INTRODUCTION

The idea that nature can be divided into three

kingdoms, mineral, vegetable and animal (Lemery,
1675), was popularized by Linnaeus in the eighteenth century. Although separate kingdoms for fungi
(Necker, 1783), Protozoa (Owen, 1858) or bacteria
(Enderlein, 1925) were proposed later, the seventeenth century conception of just two kingdoms of
life dominated biology for three centuries. The
discovery of protozoa (Leeuwenhoek, 1675) and
bacteria (Leeuwenhoek, 1683) eventually undermined the two kingdom system. But general agreement that the living world must be classified into at
least five kingdoms (Margulis & Schwartz, 1988 ;
Cavalier-Smith, 1989 a ; Mayr, 1990) was reached
only after the dramatic discoveries made by electron
microscopy in the second half of the twentieth
century. These conclusively confirmed the fundamental differences between bacteria and eukaryotes,
and revealed the tremendous ultrastructural diversity of protists. Acceptance of the necessity for
several kingdoms also owes much to the systematic
synthesis of Copeland (1956), and the influential
writings of Stanier (1961 : Stanier & van Niel, 1962)
and Whittaker (1969).
Whether more kingdoms than five are needed
(Leedale, 1974), and if so how many, has been
debated for over 20 years (Cavalier-Smith, 1978,
1981 a ; Mo$ hn, 1984 ; Corliss, 1994). Even more
important than the sheer number of kingdoms is the
fact that the definition and circumscription of the
recognized kingdoms are not yet agreed by different
systematists. Phylogenetic advances have been so
great over the past 30 years, however, that we can
now define monophyletic kingdoms, the circumscription of which ought to be widely acceptable and
therefore lead to a much desired stability in the
macrosystem of life.
Some years ago, I argued that the minimum
number of kingdoms suitable for general purposes
was six (Cavalier-Smith, 1981 a), and that in
addition to the previously accepted five kingdoms it
was necessary to recognize a third botanical kingdom, Chromista (mainly comprising the oomycetes
and those numerous algae that, remarkably, have
their chloroplasts located within the lumen of the
rough endoplasmic reticulum, not in the cytosol as in

258
258
259
259

the plant kingdom). That paper also made major

changes in the circumscription of the existing
botanical kingdoms Fungi and Plantae. The sixkingdom system of Cavalier-Smith (1981 a) was later
slightly modified (Cavalier-Smith, 1983 a) by using
the older and more widely familiar name Protozoa,
rather than Protista, for the basal eukaryotic
kingdom [in effect, broadening the kingdom Protozoa of Cavalier-Smith (1981 a)] and creating a new
protozoan subkingdom, Archezoa, for putatively
primitively amitochondrial protozoa. Since then the
boundaries of the kingdom Protozoa have not been
totally stable : the taxon Archezoa was later
(Cavalier-Smith, 1987 a) modified by exclusion of
the Parabasala (trichomonads and hypermastigote
flagellates), and the revised Archezoa were segregated from Protozoa as a separate kingdom. At the
same time the bacterial or prokaryotic taxa Eubacteria and Archaebacteria, treated as subkingdoms by Cavalier-Smith (1981 a, 1983 a), were
raised in rank to kingdoms, thus creating an eight
kingdom system (Cavalier-Smith, 1987 a, 1989 a, b),
which I subsequently advocated (Cavalier-Smith,
1991 ac, 1993 a, 1995 a) and which has been adopted
in certain general works (Gould & Keeton, 1996 ;
Maynard Smith & Szathmary, 1995).
The central purpose of the present review is to
reconsider the merits of the earlier six-kingdom
system in comparison with the eight-kingdom system. After discussing the advantages and disadvantages of each, it will be concluded that the simpler
six-kingdom classification is preferable as a general
reference system. Finer phylogenetic discrimination
can be achieved by making better use of intermediate-level categories such as subkingdoms and
infrakingdoms, without multiplying the number of
kingdoms. Some changes in circumscription of four
of the six kingdoms are also needed ; in the light of
recent molecular sequence evidence it is necessary to
remove four groups from the Protozoa and to place
them in higher kingdoms ; thus Microsporidia are
transferred to the Fungi, Myxozoa and Mesozoa to
the Animalia, and Opalinata to the Chromista. This
makes the kingdom Protozoa substantially more
homogeneous. The postulated phylogenetic relationships between the kingdoms, subkingdoms, and the
most deeply divergent infrakingdoms of the present
system are shown in Fig. 1.

T. Cavalier-Smith

206

ANIMALIA
Mesozoa

CHROMISTA

FUNGI
Neomycota

Cryptista

PLANTAE

Chromobiota

Viridiplantae

Bilateria
Myxozoa
Eomycota

Radiata

Biliphyta

Discicristata
Alveolata
Actinopoda

Sarcomastigota
chloroplast

Neozoa

tubular mitochondrial cristae

spliceosomal introns; bikonty

neokaryotes
peroxisomes
Archezoa

PROTOZOA
mitochondria

Empire
EUKARYOTA
Empire
PROKARYOTA

endomembrane system, cytoskeleton

tetrakont centrioles, nucleus, mitosis, sex

loss of murein

isoprenoid
ether lipids
Posibacteria

Archaebacteria
Unibacteria

loss of outer membrane

grade
Eubacteria

Glycobacteria
lipopolysaccharide
BACTERIA
Negibacteria
Lipobacteria

acyl ester membrane lipids

murein walls; outer membrane

Fig. 1. Postulated phylogenetic relationships between the six kingdoms (upper case) and their subkingdoms. Infrakingdoms are also shown for the two basal paraphyletic kingdoms, as are some key innovations (within heavy boxes).
The four major symbiogenetic events in the history of life are shown with dashed arrows : (1) the symbiogenetic origin
of mitochondria from an -proteobacterium, (2) the probable origin of peroxisomes from a posibacterium (CavalierSmith, 1990), (3) the monophyletic origin of chloroplasts from a cyanobacterium, and (4) the origin of the ancestral
chromist as a eukaryote-eukaryote chimaera between a rhodellophte red algal endosymbiont and a biciliated protist
host ; whether the host was a protozoan, as assumed here, or a very early plant is unclear. For clarity infrakingdoms
are not shown for the four higher holophyletic kingdoms, Animalia, Fungi, Plantae, and Chromista. The diagnoses of
the taxa are given in Table 1 and their composition is summarized in Tables 27. Chloroplasts are assumed to have
originated in the latest common ancestor of Plantae, Discicristata and the alveolate dinoflagellate protozoa (CavalierSmith, 1982), and the plastids of euglenoids and alveolates are assumed to have arisen divergently from this primary
endosymbiosis. However, the alternative possibility that dinoflagellates and}or euglenoids obtained their chloroplasts
secondarily by lateral transfer from a chromobiote and green alga respectively, though considered distinctly less likely,
cannot currently be excluded (Cavalier-Smith, 1995 a) ; the source of the non-photosynthetic plastid of coccidiomorph
Sporozoa is equally uncertain (McFadden & Waller, 1997). For simplicity a fifth important case of symbiogenesis is
not shown : secondary symbiogenetic lateral transfers of green algal, possibly ulvophyte (Ishida et al., 1998), chloroplasts into a cercozoan (sarcomastigote) host to create the algal class Chlorarachnea (e.g. Chlorarachnion). The least
certain features of the tree are the position of its root (some authors favour a root within the Unibacteria rather than
the Negibacteria) and of the Discicristata (18S rRNA and EF 1- trees put them below the Sarcomastigota, whereas
other proteins place them as shown).

A revised six-kingdom system of life

When establishing the kingdom Chromista
(Cavalier-Smith, 1981 a) I mentioned that, as discussed earlier by Taylor (1978) and Hibberd (1979),
there were reasons for thinking that proteromonads
may be closer to Chromobiota than to Protozoa.
They had some ultrastructural similarities with
chromists and some with the opalinids, so it appeared
that they might be phenotypically intermediate
between chromists and protozoa. In order to make
the definition of the Chromista as sharp as possible,
I conservatively kept the proteromonads in the
kingdom Protozoa, placing them together with
opalinids and cyathobodonids in the protozoan
group Proterozoa, but recognized that the boundary
between chromists and protozoans might require
slight adjustment in future. Patterson (1985) subsequently placed proteromonads and opalinids alone
together in the order Slopalinida. Later Patterson
(1989) argued that slopalinids should be grouped
with heterokonts as stramenopiles. I have since
argued that the genus Proteromonas is less closely
related to Opalinida than is the genus Karotomorpha
and placed the orders Karotomorphida and Opalinida together in the class Opalinea (CavalierSmith, 1993 a, b) but put proteromonads in a
separate class. As part of a reappraisal of the
Proterozoa stimulated by recent rRNA (ribosomal
RNA) sequences (Cavalier-Smith & Chao, 1995,
1997) I have created a new class Proteromonadea
for the Proteromonadida, abandoned the taxon
Proterozoa and grouped Opalinea with Proteromonadea in the subphylum Opalinata (CavalierSmith, 1997 a). In order to increase the homogeneity
of the new protozoan subkingdom Neozoa and the
zooflagellate phylum Neomonada, I excluded the
revised Opalinata from them as a protist taxon
incertae sedis. Here, I accept the view (Patterson,
1989) that Opalinata (a subphylum now compositionally the same as the order Slopalinida) are
probably secondarily modified heterokonts, and
therefore place them within the chromist infrakingdom Heterokonta and group them with the
heterotrophic subphylum Pseudofungi (CavalierSmith, 1986 a, 1989 b) to form a new heterotrophic
chromist phylum, Bigyra, in which I also place the
recently discovered distinctive heterokont protist
Developayella elegans (Tong, 1995), for which I create
a new order, class and subphylum.
In the system of Cavalier-Smith (1981 a) I refined
the kingdom Fungi by removing oomycetes, hyphochytrids, and thraustochytrids and transferring them
into the Heterokonta within the new kingdom
Chromista, and arguing that Chytridiomycetes, by

207

contrast, should be included in the kingdom Fungi.

This circumscription of the kingdom Fungi (see also
Cavalier-Smith, 1987 b) is now almost universally
accepted by phylogenetically oriented mycologists
(Bruns, White & Taylor, 1991), and adopted by the
authoritative Dictionary of the Fungi (Hawksworth et
al., 1995). It will, therefore, come as a considerable
surprise to mycologists to learn that the kingdom
Fungi now needs to be drastically expanded by the
addition of the Microsporidia, a phylum of minute
intracellular parasites of animals. Though Microsporidia have chitinous spores and are not phagotrophic, they were traditionally thought of as
protozoa rather than fungi because their vegetative
cells lack cell walls. Like Opalinata, the Microsporidia lack peroxisomes ; but as they also lack
mitochondria they were postulated to be archezoan
Protozoa (Cavalier-Smith, 1983 a) ; however, recent
protein sequence data (Li et al., 196l ; Edlind et al.,
1996 ; Keeling & Doolittle, 1996 ; Roger, 1996 ;
Germot et al, 1997) strongly suggest that they are
actually highly degenerate fungi, which have secondarily lost mitochondria, like the rumen fungi
(Neocallimastigales), which have also been mistakenly classified as protozoa in the past. I here
remove the Microsporidia from the Archezoa and
place them instead in the kingdom Fungi, where I
group them with the Archemycota in a new
subkingdom Eomycota. As the higher level classification of the kingdom Fungi needs some revision, I
divide it into two new subkingdoms (Eomycota and
Neomycota) and create a few other new high level
fungal taxa, and also validate some of those proposed
earlier (Cavalier-Smith, 1987 b). I present a detailed
classification of the kingdom Fungi, down to the
level of subclass.
The third refinement made here to the circumscription of the Protozoa compared with the earlier
six-kingdom system (Cavalier-Smith, 1983 a) is the
transfer of the Mesozoa and Myxosporidia into the
kingdom Animalia, as briefly indicated earlier
(Cavalier-Smith, 1995 b, Cavalier-Smith et al.,
1996 a).
The reasons for the above changes in circumscription of the kingdoms Protozoa, Chromista,
Fungi and Animalia will be explained in detail. In
view of the numerous changes in high-level classification over the past 15 years, I shall present a
detailed summary of this revised six-kingdom classification of the living world, down to the level of
infraphylum.
My redefinition of the kingdom Plantae to include
all Viridaeplantae (green plants), Glaucophyta, and

208

T. Cavalier-Smith

Table 1. The revised six-kingdom system of life

Empire or Superkingdom 1. PROKARYOTA** Dougherty 1957 stat. nov. Allsopp 1969 (ribosomes in the
same compartment as the usually circular chromosome).
Kingdom Bacteria** Cohn 1870 stat. nov. Cavalier-Smith 1983 (syn. Procaryotae Murray 1968 : no internal
cytoskeleton, endomembrane system, nuclei, mitosis or true sex).
Subkingdom 1. Negibacteria* Cavalier-Smith 1987 (outer membrane present ; acyl ester lipids ; with small signalrecognition particle RNA).
Infrakingdom 1. Lipobacteria* new infrakingdom (diagnosis : no lipopolysaccharide in outer membrane ;
murein cell wall).
Infrakingdom 2. Glycobacteria* new infrakingdom (diagnosis : lipolysaccharide in outer membrane ; cell wall of
murein or, rarely, protein).
Subkingdom 2. Unibacteria* new subkingdom (diagnosis : bacteria with no outer membrane ; with large signalrecognition particle RNA as in eukaryotes).
Infrakingdom 1. Posibacteria* Cavalier-Smith 1987 stat. nov. (acyl ester lipids ; often with teichoic acids).
Infrakingdom 2. Archaebacteria Woese & Fox 1977 stat. nov. (isoprenoid ether lipids ; murein absent).
Empire or Superkingdom 2. EUKARYOTA (cytoskeleton, endomembrane system, nucleus, sex).
Kingdom 1. Protozoa** Goldfuss 1818 stat. nov. Owen 1858 em. (phagotrophs primitively without plastids,
collagen or chitinous vegetative cell walls ; unicellular, plasmodial or colonial).
Subkingdom 1. Archezoa** Cavalier-Smith 1983 em. (kinetid tetrakont ; mitochondria absent ; genes unsplit).
Subkingdom 2. Neozoa** Cavalier-Smith 1993 stat. nov. 1997 (kinetid typically bikont ; tubular or rarely flat
mitochondrial cristae ; spliceosomal introns common).
Infrakingdom 1. Sarcomastigota** Cavalier-Smith 1983 stat. nov. em. (kinetid bikont or secondarily unikont
or absent ; mitochondria usually with tubular cristae, rarely flat (typically non-discoid) or vesicular ; cortical alveoli
absent ; pseudopodia when present non-eruptive ; axopodial microtubules if present, not spirally or hexagonally
arranged].
Infrakingdom 2. Discicristata new infrakingdom (kinetid tetrakont or bikont ; mitochondria or hydrogenosomes
present ; cristae discoid ; genes mostly unsplit ; pseudopodia eruptive when present ; cortical alveoli and axopodia
absent).
Infrakingdom 3. Alveolata Cavalier-Smith 1991 [cortical alveoli (rarely secondarily absent) ; mitochondrial
cristae tubular or ampulliform].
Infrakingdom 4. Actinopoda Calkins 1902 stat. nov. Cavalier-Smith 1996 (axopodia with hexagonal or spirally
arranged microtubules ; cilia absent or for dispersal only ; possibly polyphyletic).
Kingdom 2. Animalia Linnaeus 1758 em. Cavalier-Smith 1995 (unnecessary synonym Metazoa Haeckel 1874)
(ancestrally phagotrophic multicells with collagenous connective tissue between two dissimilar epithelia).
Subkingdom 1. Radiata** Linnaeus 1758 stat. nov. em. Cavalier-Smith 1983 (multicellular animals with radial
or biradial symmetry ; no anus).
Infrakingdom 1. Spongiaria* De Blainville 1816 (choanocytes line body cavity ; nervous system primitively
absent : sponges).
Infrakingdom 2. Coelenterata* Leuckart 1847 em. auct. (nerve net).
Infrakingdom 3. Placozoa infraking. nov. (without body cavity, gut or nervous system).
Subkingdom 2. Myxozoa Grasse! 1970 stat. nov. (secondarily unicellular parasites of bilateral animals ; spores
multicellular ; cilia absent : myxosporidia).
Subkingdom 3. Bilateria Hatschek 1888 stat. nov. Cavalier-Smith 1983 (bilateral, primitively with anus).
Branch 1. Protostomia* Grobben 1908 (blastopore becomes mouth).
Infrakingdom 1. Lophozoa* new infrakingdom (primitively sessile with U-shaped gut and ciliated oral
tentacles with coelomic extensions ; early ciliated larvae trochophores, later often bivalved).
Infrakingdom 2. Chaetognathi Leuckart 1854 stat. nov. (non-ciliated ; thin cuticle not moulted ; embryonic
enterocoel absent in adult).
Infrakingdom 3. Ecdysozoa new infrakingdom (non-ciliated ; thick cuticle that is moulted ; haemocoel or
pseudocoel).
Infrakingdom 4. Platyzoa new infrakingdom (secondarily acoelomate worms without vascular system ;
ancestrally ciliated).
Branch 2. Deuterostomia Grobben 1908 (blastopore becomes anus).
Infrakingdom 1. Coelomopora stat. nov. (trimeric coelom, anterior compartment with external pore ;
enterocoelous ; ciliated pelagic larvae ; nerve net).

A revised six-kingdom system of life

209

Table 1. (cont.)
Infrakingdom 2. Chordonia Haeckel 1874 em. Hatschek 1888 (schizocoelous or coelom absent ; larvae
tadpole-like ; hollow dorsal nerve cord).
Subkingdom 4. Mesozoa Van Beneden 1877 stat. nov. (ciliated multicellular parasites ; no nervous system or gut).
Kingdom 3. Fungi Linnaeus 1753 stat. nov. Nees 1817 em. (vegetative and}or spore cell walls of chitin and glucan ; no phagocytosis).
Subkingdom 1. Eomycota** subking. nov. (diagnosis : hyphae without perforate septa ; without dikaryotic phase :
cellulae non binucleatae ; septa usitate absens ; si praesens non perforata).
Subkingdom 2. Neomycota subking. nov. (diagnosis : usually with a dikaryotic phase ; hyphae, if present, with
perforate septa : cellulae binucleatae plerumque praesentes ; endospora aut ascospora aut basidiospora instructa).
Kingdom 4. Plantae Haeckel 1866 em. Cavalier-Smith 1981 (plastids with double envelope in cytosol ; starch ; no
phagocytosis).
Subkingdom 1. Biliphyta* Cavalier-Smith 1981 (phycobilisomes ; single thylakoids ; starch in cytosol).
Infrakingdom 1. Glaucophyta infraking. nov. (diagnosis : peptidoglycan in plastid envelope : plastidae
peptidoglycanum instructae).
Infrakingdom 2. Rhodophyta infraking. nov. (sine peptidoglycano : plastid envelope lacks peptidoglycan).
Subkingdom 2. Viridaeplantae Cavalier-Smith 1981 (chlorophyll a and b ; thylakoids stacked ; starch in plastid
stroma ; ciliary transition nine-fold star : green plants).
Infrakingdom 1. Chlorophyta Cavalier-Smith 1993 (green algae).
Infrakingdom 2. Cormophyta Endlicher 1836 stat. nov. (embryophytes).
Kingdom 5. Chromista Cavalier-Smith 1981 em. (chloroplasts with chlorophyll c inside a periplastid membrane
within the rough endoplasmic reticulum lumen and}or with rigid bipartite or tripartite ciliary hairs).
Subkingdom 1. Cryptista Cavalier-Smith 1989 (ejectisomes ; usually with bipartite hairs on both cilia,
nucleomorph and phycobilins ; cristae flattened tubules ; pellicular plates).
Subkingdom 2. Chromobiota Cavalier-Smith 1991 (cristae tubular ; posterior cilium often autofluorescent ; no
nucleomorph, ejectisomes, pellicular plates or phycobilins).
Infrakingdom 1. Heterokonta Cavalier-Smith 1986 stat. nov. 1995. em. (rigid tripartite or bipartite hairs on
anterior cilium only or, rarely, on cell body ; no haptonema).
Infrakingdom 2. Haptophyta Cavalier-Smith 1995 (haptonema ; hairs unipartite tubules, knob hairs or absent).
* Probably paraphyletic taxon.
** Almost certainly paraphyletic.
The classification of each of the kingdoms is given in more detail in Tables 27, with examples of representatives
of the major taxa. It is doubtful if the modification of the spelling of the name Archaebacteria in Bergeys Manual to
Archaeobacteria (Gibbons & Murray, 1978) was correct ; if archae- was derived from the Greek archaios (ancient) then
it is a stem and the o should be inserted according to the rules of the botanical code but this is not required by
the bacteriological code ; but if it was derived from the Greek arche (beginning), Archaebacteria is a correctly
formed pseudocompound word in which archae retains the case ending -ae, and as archae is a whole word (not the
stem arch-) it would be incorrect to insert the o . For stability, I prefer to keep the original spelling and to assume
that Woese & Fox (1977 a) made no error. The more recent change to Archaea (Woese, Kandler & Wheelis, 1990)
is both totally unnecessary and highly undesirable ; its use should be most strongly discouraged (Cavalier-Smith,
1992 b).

Rhodophyta, and these three groups alone (Cavalier-Smith, 1981 a), is not yet widely accepted, largely
because of a formerly widespread belief in the
polyphyletic symbiotic origins of chloroplasts
(Mereschkowsky, 1910 ; Margulis, 1970, 1981 ;
Raven, 1970). But as the theory of the monophyletic
symbiogenetic origin of chloroplasts (CavalierSmith, 1982), upon which my circumscription of
Plantae was based, is increasingly widely accepted
(Bhattacharya & Medlin, 1995), and as it logically
follows that the three major plant groups are derived

from a single photosynthetic common ancestor, I

think that it will be only a matter of time before the
logic of accepting the monophyly of the kingdom
Plantae sensu Cavalier-Smith (1981 a) will also be
generally recognized, as it already has been by
Ragan & Guttell (1995). I have therefore seen no
reason to modify the circumscription of the kingdom
Plantae or its subdivision into two subkingdoms
(Viridaeplantae and Biliphyta) since Cavalier-Smith
(1981 a), and expect the kingdom to continue to be
stable in the future.

210

An overview of the revised six-kingdom system is

given in Table 1, which includes all the supraphyletic
taxa, with diagnoses showing how they may be
distinguished. In order to save space in the later
tables dealing with the phyla, subphyla and infraphyla of each kingdom diagnoses are given only for
new or substantially emended taxa. Despite the
creation here of five new phyla, I accept only 60
phyla for the entire living world, many fewer than
the 92 recognized by Margulis and Schwartz (1988)
or the 100 used by Mo$ hn (1984). I am not actually
a born splitter (Corliss, 1994). Viruses are best
thought of as laterally transmissible parasitic genetic
elements, not as living organisms, so I do not attempt
to classify these important biological entities here.
II. PHILOSOPHIC PRELIMINARIES

(1) Darwinian evolutionary classification

contrasted with Hennigian cladification
Ever since Darwin, it has been accepted that the
systematist is primarily a morphologist whose task is
to discover the true phylogenetic relationships as
Cummings (1916 : 254) noted 80 years ago. Darwinian evolutionary classifications take into account
both the branching patterns and the degree of
change along different branches, and are therefore
truly phylogenetic (Mayr & Ashlock, 1991) rather
than purely cladistic or purely phenetic. A balanced
classification aims to subdivide a taxon like Eukaryota into a number of subordinate taxa of lower rank
in such a way as to maximize the degree of similarity
within each taxon, while ensuring that none of them
is polyphyletic. Since organisms can vary in many
different characters, and in ways varying from the
trivial to the profound, it is essential to weigh their
relative importance. Lamarck (1810) initiated the
idea of genealogical classification, but it was Darwin
(1859), Haeckel (1866, 1868) and Lankester (1877)
who first really popularized the notion of phylogenetic classication as the central goal of systematics.
Haeckel (1866) introduced the terms phylogeny,
phylum, monophyletic and cladus, whilst Lankester
(1977) gave us the concept of a grade and encouraged the English-speaking world to substitute
Haeckels term phylum for subkingdom. These early
pioneers of phylogenetic classification used both
grades and clades as taxa in their influential
classifications.
When I first argued that the kingdoms of life
needed reclassifying on phylogenetic principles to
make them monophyletic (Cavalier-Smith, 1978), I

T. Cavalier-Smith
was complimented by certain cladists who mistakenly thought I was applying the principles of
Hennig (1966) because I was using cladistic reasoning and advocating a phylogenetic classification
with strictly monophyletic taxa. However, I had not
then even heard of Hennig or his confusing redefinition of the term monophyletic, which I do not
accept. Like classical phylogeneticists, I use the word
monophyletic to include both holophyletic (Ashlock,
1971, i.e. monophyletic sensu Hennig, 1966) and
paraphyletic. I do, however, accept Hennigs redefinition of Huxleys (1957, 1959) term clade, which was
originally defined using the classical, not the
Hennigian definition of monophyly. (The earlier use
of the word clade in biology by Lankester (1911 :
1031) as an anglicization for Haeckels (1868)
taxonomic category of cladus, equivalent in rank to
infraphylum in the present system, is no longer
current : the Renaissance usage of clade to denote
a plague or disaster is even more obsolete.) Hennig
and his followers have done much to increase the
rigour of phylogenetic analysis, but this advance has
been brought at the cost of some verbal confusion
and much harmful dogmatism. Furthermore, though
cladistic analysis is simply formalized phylogenetic
common sense, and should be strongly encouraged,
Hennigian taxonomy has two grave defects, one
practical and one theoretical.
Practically, it leads to serious instabilities in
classification and nomenclature. The great practical
merit of a Darwinian evolutionary classification is
that it should be much more stable, since whether a
taxon is holophyletic or paraphyletic is irrelevant to
its classificatory role. There are many thousands of
taxa where we do not know whether they are
paraphyletic or holophyletic. According to
Hennigian principles, whenever any taxa become
clearly established as paraphyletic we would have to
split them up, abandon their names and invent two
or more new taxa. This would be especially
nomenclaturally destabilizing for genera, a large
proportion of which may be paraphyletic.
Theoretically, the Hennigian attempt to restrict
taxa to clades, and forbid paraphyletic groups is
incompatible with the basic purpose of phylogenetic
classification, even though it misleadingly masquerades under that name. What a biological
classification aims to do is to arrange organisms in a
hierarchical series of nested taxa, in which each
more-inclusive higher-level taxon is subdivided comprehensively into less-inclusive taxa at the next level
below. There are two key words here : hierarchical
and comprehensive. A set of nested clades is

A revised six-kingdom system of life

211

(a)

(b)
D

H
A
1

B
2

C
6

B
2

C
4

F
10

11
12

10
13

14

12

11
13
14
G

15

Cladification
Clade E
Clade D
Clade A
Clade B
Clade C

15

Classification
Taxon E
Taxon D
Taxon F*
Taxon A
Taxon B
Taxon H*
Taxon G*
Taxon C

Fig. 2. Contrast between (a) cladification and (b) classification. From the same phylogenetic tree with holophyletic
recent species 1 to 6 and paraphyletic ancestral species 7 to 15 one can produce either a list of nested clades or a
comprehensive classification. In the classification taxon D is comprehensively subdivided into three taxa of lower and
equal rank (A, B, F) ; A and B are holophyletic and F is paraphyletic. In the cladification, clade D is subdivided into
two clades A and B with the same composition as the holophyletic taxa A and B plus a residue including species 10
to 12 which is not placed in any subgroup of clade D. The cladification is not a comprehensive classification of clade
D, because this paraphyletic residue comprising species 10 to 12 is not listed under clade D even though it is part of
that clade. The size of this paraphyletic residue could be reduced in an alternative cladification by moving the boundaries of clades A and B from just below species 7 and 8 to just above species 12 ; but as species 12 is ancestral to both
clades A and B, it cannot be included in either. No clade can be comprehensively subdivided into subclades. A
cladification can never be a comprehensive classification since it omits the paraphyletic residues which a comprehensive classification must include (taxa F, G and H in this case). The cladification on the left is congruent with the
classification on the right. When comprehensively subdivided into taxa of equal rank, it is logically inescapable that
every monophyletic taxon (whether holophyletic or paraphyletic) must include at least one paraphyletic taxon. A
cladification is a most useful step towards a phylogenetic classification but it does not constitute one in its own right.
If paraphyletic taxa are explicitly identified in a phylogenetic classification one can reconstruct the cladification on
which it is based by deleting them. But one cannot move from a phylogenetic tree to a classification without exercising
taxonomic judgement, which is necessary both in deciding which clades should be treated as taxa and whether the
paraphyletic residue should be treated as a single taxon or subdivided. Judgement is likewise needed about the
appropriate rank for each taxon.

hierarchical, but because it is not comprehensive at

each level in the hierarchy, it is much too incomplete
to be properly called a classification. Ernst Mayr
(pers. comm.) has recently called such a set of nested
clades a cladification. Figure 2 contrasts the way in
which a Darwinian evolutionary classification subdivides a phylogenetic tree to produce a comprehensive classification of the organisms making up
the tree, whereas Hennigian cladification of the same
organisms yields an incomplete subdivision of higher-

level groups. Cladists have long accepted that the

inability to classify ancestral and many fossil taxa is
the Achilles heel of Hennigian classificatory principles, and refer to it as a problem ; it is not a
problem at all for systematics, but merely a basic
defect in the Hennigian ideas on classification.
Obviously, if you assert that you must not make
paraphyletic groups then you cannot properly
classify ancestral species excluded from a particular
clade. No comprehensive phylogenetic classification

212

is even theoretically possible unless one accepts

paraphyletic as well holophyletic taxa. The dogma
against paraphyletic taxa is logically incompatible
with the acceptance of both evolution by descent
and the goal of taxonomy as the creation of a
comprehensive phylogenetic classification of all
organisms, both extant and extinct.
As it is theoretically unsound, and in practice
increases nomenclatural and classificatory instability, the dogma against paraphyletic groups has been
most detrimental to systematics. Cladists, rightly
immensely impressed with the logic and great value
of Hennigian principles for phylogenetics, have
naturally been reluctant to accept that they can be
so fatally flawed when applied to taxonomy. Some
do recognize this central flaw, but say that in
practice most taxonomists deal only with extant
organisms or that it is never possible to recognize
ancestors with certainty. However, we sometimes
can recognize ancestors and need to classify them.
Furthermore, not all ancestors are extinct. In plants,
new allopolyploid species are created by hybridization between two ancestral species followed by
polyploidization. In several instances, both ancestral
species have been unambiguously identified and
both are extant. A well known example is the
formation of the new grass Spartina anglica by
allopolyploidy from the ancestors S. alterniflora and S.
maritima. The two ancestors are both perfectly good
paraphyletic taxa, and as biological species are
natural taxa, distinct from each other and from their
descendant species S. anglica. Paraphyletic species
are therefore as natural as holophyletic ones. Apart
from the special case of biological species, paraphyletic and holophyletic taxa are both partly artificial
constructs of taxonomists (whether they are asexual
species or taxa of higher rank). From such a
perspective a holophyletic taxon is not created by
the origin of a new synapomorphy, but by the
decision of a systematist to subdivide an historically
continuous phylogenetic tree at that point. As I have
stressed previously (Cavalier-Smith, 1993 a), every
such cutting off of a holophyletic branch necessarily
simultaneously generates a paraphyletic stem. If I
cut a tree into pieces, each piece, whether it is a
terminal piece with one cut or a stem piece with two
or more, is both artificial and natural. Each
continuous segment owes its continuity to natural
processes, whereas its separation from others is the
result of human, i.e. artificial intervention. If I group
together cut pieces that were never joined on the
tree, such joining would be totally unnatural, which
is why polyphyletic grouping has long been accepted

T. Cavalier-Smith
as much more artificial than a monophyletic grouping. For monophyletic groups (whether paraphyletic
or holophyletic), their coherence is the result of
historical genetic continuity between all members of
the group : it is their discontinuity from other groups
that is artificial (except for biological species). The
claim that a cladogram or phenogram can be
converted into a classification without artificial cuts
across the natural phylogenetic continuum (Nelson
& Platnick, 1981) is at best a spurious half truth ;
cladograms and phenograms can be claimed to be as
discontinuous as classifications only because the cuts
across the phyletic continuum have already been
made by classical taxonomists when creating the
discrete taxa that form the raw materials for cladistic
and phenetic analysis. The claim involves an element
of myopic self-deception that allows some cladists to
claim falsely that Hennigian systematics is more
natural than classical phyletic taxonomy (Stuessy,
1990). No classification can be totally natural.
Whether a taxon is paraphyletic or not is
irrelevant to its validity as a taxon. It is also
irrelevant to many of the uses to which classifications
are put, such as arranging specimens in a museum,
organising the chapters in a biology textbook, or
providing a convenient label, e.g. bacteria or fungi,
for a group of similar organisms. But the distinction
is important in certain uses to which classifications
are sometimes put. Two common examples are the
choice of model systems and the study of group
extinction.
A biologist wishing to choose a protozoan as a
model system for understanding animal cell biology
would be quite mistaken in supposing that as all
protozoa are classified in the same kingdom it does
not matter which is chosen. Obviously, some
members of such a paraphyletic group are cladistically more closely related to a derived taxon than
others : for example, choanoflagellates would be
much more similar to animal cells than would
retortamonads. Likewise, the extinction of all members of a paraphyletic group, in contrast to that for
a holophyletic group, does not constitute extinction
of the entire lineage. But the mistaken assumption
that it does, like the assumption that all members of
a group are equally related to another group, are
misuses, not uses, of a classification. Where precise
phylogenetic relationships are important for a
scientific problem, scientists should base their reasoning directly on phylogenies and not use classifications
as a crude surrogate for the real thing. The purpose
of classification is to provide a simplified reference
system that is biologically sound and widely useful.

A revised six-kingdom system of life

It should be compatible with the phylogeny, but it
cannot serve its central simplifying purpose unless it
leaves out some of the fine detail about relationships
that are essential for some phylogenetic purposes.
One can use a phylogeny as a basis for making a
classification, but one cannot logically deduce a fully
detailed phylogeny from a classification. Nor is a
phylogeny sufficient to give a classification. A
phylogeny and a classification must be congruent
(i.e. not contradictory) but they are different ways of
abstracting from and representing biological relationships. It is highly desirable that taxonomists publish
an explicit phylogeny and statement of their phylogenetic assumptions at the same time as their
classification, a practice I have long followed
(Cavalier-Smith, 1978). This will allow users to use
whichever is most appropriate to their needs. In
order to alert users of classifications to the sometimes
important differences between holophyletic and
paraphyletic taxa it would be good practice if, where
known, these are clearly distinguished. In Tables
17 in this review I mark those taxa that are almost
certainly paraphyletic and those which, in my
phylogenetic judgement, are probably paraphyletic.
Taxa not so identified include not only those known
or thought to be holophyletic but also those for
which evidence is too little for me to make a
judgement ; future work is bound to show that some
are paraphyletic, and others polyphyletic, and that
at least some of those thought to be paraphyletic are
actually holophyletic.
(2) Naming taxa and clades
A named taxon such as Protozoa or Plantae has to
serve a dual role. Its name acts as a label to refer
succinctly to a group of related organisms sufficiently
similar to each other that they share a generalized
phenotype, by which they can be readily recognized
and distinguished from other taxa. Secondly, they
are units that can be grouped together hierarchically
to create a smaller number of higher taxa. A list of
nested clades, such as that in Patterson (1994), can
be very useful as a phylogenetic summary, but it is
not the same as a classification.
Stability is desirable in nomenclature, so I have
tried wherever reasonably possible to retain older
and more familiar names. But stability is not a
primary value in classification. If it were, we should
rigidly retain the oldest classification irrespective of
how bad it is ! Especially among microorganisms
traditional classifications have been so grossly incorrect that they must be changed as science

213

advances. This requires some new names of higher

taxa and some shifts in the meaning of old ones.
Because classifications are scientific generalizations
about the varied properties of organisms, and not
just indexes to stored information, they must be
changed if new knowledge shows them to be
fundamentally wrong. During the past half century,
electron microscopy, biochemistry and molecular
sequencing have immensely deepened our understanding of biodiversity. After the revolutionary
changes occasioned by this have been soundly
assimilated, our classifications will become more
stable. Thus stability is a valuable outcome of a
sound classification, but must not dominate systematics by impeding necessary changes.
It is important to have a brief name for all clades
of major evolutionary significance, but one should
not make the mistake of supposing that all such
clades need to be made into taxa. Nor should one
make the converse error, especially common among
cladists, of supposing that only clades can be taxa.
Taxa and clades should be thought of as two partly
overlapping sets that serve somewhat different, but
equally essential, biological purposes. In a given
classification such as the six-kingdom system advocated here, some taxa (e.g. Fungi, Animalia,
Chromista and probably Plantae) are clades, but
paraphyletic taxa like Bacteria, Archezoa and
Protozoa are not. For example, an important clade
which is not a taxon in this system is the clade
comprising Animalia, Fungi and Choanozoa, which
I named opisthokonta (Cavalier-Smith, 1987 b) ;
molecular phylogeny has since confirmed the validity
of this clade (Cavalier-Smith, 1993 a ; Wainright et
al., 1993). I deliberately did not create a taxon
Opisthokonta or give a diagnosis for it. This is
because opisthokonta are far too phenotypically
diverse to be useful as a major unit of eukaryote
classification. One cladist has even told me that I
ought to have created a formal kingdom Opisthokonta. A kingdom Opisthokonta would be much
less useful than the existing kingdoms Animalia,
Fungi and Protozoa as a way of subdividing the
living world into manageable major groups of similar
organisms, i.e. in classification as opposed to phylogeny. The cladists suggestion reflects a misunderstanding of the purposes and functions of classification, which is all too common among cladists,
who are typically very much more interested in
phylogeny than in classification, and often forget
that these are two different branches of systematics.
Clades and taxa are non-equivalent concepts, as is
shown pictorially in Fig. 3.

T. Cavalier-Smith

214

(3) Principles of ranking taxa

Clades

CT

Taxa

CTG
CG

GT
G

Grades

Fig. 3. The non-equivalence of clades and taxa and the

non-exclusiveness of either with grades. Biological groups
may be simply clades, grades or taxa, or any two or three
of these at the same time. Some well known groups, such
as vertebrates and eukaryotes, are simultaneously taxa,
clades, and important grades of organization (region
CTG). Some clearly definable clades should not be made
taxa, partly because they do not constitute clearly defined
grades of organization (region C) (e.g. opisthokonta, i.e.
the clade comprising Fungi, Animalia and Choanozoa
(Cavalier-Smith, 1987 a) ; or neokaryotes, the clade
consisting of all descendants of the most recent common
ancestor of Neozoa). Yet other clades are grades of
organization, but for other reasons are not made taxa
(region CG ; e.g. Metakaryota, which was treated as a
taxon in the eight kingdom system, but not in the sixkingdom system for reasons discussed in the text).
Paraphyletic taxa (regions T and GT) are not clades ; but
they may be grades of organisation of greater (e.g.
Bacteria, Archezoa) or lesser (e.g. Reptilia, Amphibia)
importance (region GT) or they may not be grades of
organization at all (region T) either because they have
secondarily become heterogeneous (e.g. the heterokont
algal phylum Ochrophyta by multiple secondary losses
of plastids) or because they are too similar to related taxa
to be called separate grades of organization (as in
paraphyletic species ancestral to allopolyploids, e.g.
Spartina anglica and S. maritima). Though many holophyletic taxa are grades (region CTG, e.g. Tracheophyta)
some are not, because of secondary changes that have
converted some of their members to a drastically different
grade (region CT ; e.g. Animalia, where the basic
triploblastic body plan has been totally lost by Myxozoa ;
Heterokonta where the Opalinata have lost the ciliary
retroneme synapomorphy that makes other heterokonts
an important grade of organization). Finally many
important grades are neither clades nor taxa (region G) ;
they may be paraphyletic groups (e.g. eubacteria,
invertebrates, fish) or polyphyletic ones (e.g. protists,
zooflagellates, amoebae). All seven types of biological
groups have valuable roles to play in the description and
analysis of organismic diversity ; it is a simplistic and
harmful error to attempt to restrict the scientifically
appropriate use of any of them by naive dichotomies into
allowable and forbidden types of group, and still worse to
pretend that the three types that are not clades are not
even groups at all.

Classically, taxonomists reach a consensus on appropriate ranks for taxa based on a broad knowledge
of diversity, judgement of degrees of structural
disparity, and respect for sensible traditions of what
degree of disparity merits a particular rank. Though
ranking has a subjective element, it is a very useful
device for helping the human mind quickly appreciate the relative position of nested taxa within
the taxonomic hierarchy and, roughly, the relative
magnitude and significance of the differences between them. It is a mistake to denigrate the process
of ranking because it cannot be objectively programmed into a computer.
Human judgement of significance and reasoned
choice between alternatives are required throughout
science, as Singer (1931 : 121123) perceptively
emphasized ; the Baconian error of thinking that one
should be able to make important discoveries or
sensible classifications merely by passing facts
through a sort of automatic logical mill (Singer,
1931 : 121) is unfortunately widespread amongst
cladists.
Even if we were given the true phylogenetic
relationships of all organisms, creating a balanced
classification would require careful taxonomic judgement in delimiting and ranking taxa : classification
should not be based on a dogmatic approach that
prohibits paraphyletic taxa or a simple algorithm or
rule like that of Hennig (1966), which stated that the
categorical rank of a taxon should be determined by
its biological age. The latter is practically impossible,
as we do not know, and may never know precisely,
where the root of the tree of life is located, and there
are numerous places in the tree with so much
uncertainty about branching order that attempts to
apply the rule would be arbitrary and less stable
than the classical method of ranking by taxonomists.
If, for the sake of argument, we knew that the root of
the tree of life was within the cyanobacteria, which
is not unreasonable given the palaeontological
evidence (Schopf, 1993), then it would be absurd to
rank certain cyanobacterial subgroups more highly
than the kingdom Animalia or the superkingdom
Eukaryota, merely because they diverged earlier.
The morphological and chemical uniformity of the
cyanobacteria is so marked that they should be
ranked only as a phylum, and their subgroups lower
still irrespective of their probably immense antiquity ; as in most other groups their molecular trees
reveal so many rapid bush-like radiations (Turner,
1998) that their branching order would be an almost

A revised six-kingdom system of life

worthless guide to ranking, even were it philosophically desirable, which it is not. Significant
evolutionary change is highly non-clock like : because
of periods of stasis or minimal change despite
branching, and an erratic distribution of major
innovations across the tree of life, there would be
no classificatory value in ranking taxa by their
presumed antiquity instead of by the classical
criterion of their structural disparity.

(4) The need to weigh and integrate

phylogenetic evidence from diverse sources
Systematics is, par excellence, a synthetic science
and needs both generalists and specialists. It is a
commonplace that excessive reliance on a single line
of evidence is dangerous and that not all data are of
equal value ; but judgement of their appropriate
value is both difficult and controversial. Congruence
between different types of evidence is the most
important criterion for assigning weight to them. But
there is no magic formula for doing so and the
creation of a macrosystem of life is both an art and
a science (Minelli, 1993) not an art that passively
reflects nature, but one which creates a simplified
picture of its underlying pattern to further its
appreciation by humans. Like any science, macrosystematics advances by trial and error : by intuition
and imagination as well as by careful observation,
experiment, reasoning and critical discussion. I
welcome criticisms of the present system, which
attempts to take some account of all types of
evidence, but deliberately gives them unequal
weight. I have attempted to give reasons for major
decisions, but space does not permit this for every
detail.

III. IN DEFENCE OF BACTERIA : THE SOLE

PRIMARY KINGDOM OF LIFE

It is highly desirable that we keep the long

established group of Bacteria as a major taxon in the
classification of life. Whether Bacteria are ranked as
a kingdom or a superkingdom is less important than
the retention of both the name and the concept of
bacteria in their classical meaning. However, I shall
argue that the advantages of ranking Bacteria as a
kingdom rather than a superkingdom outweigh the
disadvantages. Before explaining my present views, I
need briefly to review some important earlier ideas.

215

(1) The concept of a bacterium (synonym :

prokaryote)
The concept of bacteria was considerably refined by
Picken (1960), Stanier (1961) and Stanier & Van
Niel (1962). These authors and Echlin & Morris
(1965), Allsopp (1969) and Stanier (1970) contrasted the structure of bacteria with that of
eukaryotes in 12 major respects : (1) the fundamental
differences in the structure and function of bacterial
flagella and eukaryotic cilia (a term which, for
clarity, embraces also eukaryotic flagella ; Cavalier-Smith, 1986 b) ; (2) the location of respiratory
chains in the cytoplasmic (surface) membrane rather
than in mitochondria ; (3) the location of photosystems in either the cytoplasmic membrane or free
thylakoids rather than chloroplasts ; (4) the usual
presence of a peptidoglycan cell wall ; (5) the absence
of an internal cytoskeleton of actin microfilaments
and tubulin-containing microtubules ; (6) the absence of cytoplasmic motility mediated by ATPase
molecular motors such as myosin and dynein ; (7)
organisation of the chromosome as a nucleoid in the
cytoplasm, not as a nucleus ; (8) absence of an
endomembrane system (endoplasmic reticulum,
Golgi complex and lysosomes) ; (9) absence of
phagocytosis, endocytosis and exocytosis ; (10) DNA
segregation by association with the cell surface, not
a mitotic spindle ; (11) recombination by parasexual
processes, not syngamy and meiosis ; (12) inability to
harbour cellular endosymbionts.
Each of these twelve character differences between
bacteria and eukaryotes is of profound evolutionary
importance and significance. Collectively, they are
of such overwhelming significance compared with
other character differences among organisms, that
Stanier, Doudoroff & Adelberg (1963) were entirely
correct in proclaiming the bacteria}eukaryote dichotomy as the most profound and significant in the
living world. Nothing that has been discovered since
then has invalidated their thesis that it outweighs in
importance the differences between the eukaryotic
kingdoms and equally strongly outweighs the differences between the various groups of bacteria.
Since then, my own analyses (Cavalier-Smith,
1981 b, 1987 c, 1991 a, b) have added many more
such differences of genetic and cellular organisation,
ranging from the presence of DNA gyrase in bacteria
but not in eukaryotes to the absence of spliceosomes,
peroxisomes and hydrogenosomes in bacteria. The
total list now extends to about 30 major differences
(Cavalier-Smith, 1991 a, b).

216

(2) Changing views on Archaebacteria

The characterization of Archaebacteria and of
their substantial differences from Eubacteria (Woese
& Fox, 1977 a ; Woese, 1987) has been of profound
importance for our understanding of bacterial
diversity. So also have been the methods of rRNA
phylogenetics pioneered by Woese (1987 ; see also
Olsen & Woese, 1993) for the progress of phylogenetic knowledge of both bacteria and eukaryotes.
I do not wish to minimize the great importance of
these very positive contributions to science, but I
must emphasize, as I have rather more mildly before
(Cavalier-Smith, 1986 c), that the systematic importance of the distinction between archaebacteria
and eubacteria has been greatly exaggerated, a view
strongly supported by Mayr (1990). Conversely, the
very much greater importance of Staniers (1961 ;
Stanier & van Niel, 1962) concept of bacteria (or
prokaryotes : the two words became synonyms with
the acceptance of Staniers term cyanobacteria for
what were previously known as blue-green algae),
and of the Bacteria}eukaryote distinction, has been
repeatedly attacked by rhetoric and assertion rather
than by careful argument. A long talk with George
Fox 10 years ago made it clear to me, however, that
when he and Woese first started to criticize the
concept of a prokaryote (Woese & Fox, 1977 b) they
did not really have Staniers ideas (1961 ; Stanier &
Niel, 1962) in mind at all, but a naive view then
widespread among molecular biologists that Escherichia coli was a typical prokaryote and that by
studying its molecular biology the whole of bacterial
biology would be revealed. This naive view clearly
deserved to be criticized. What was most unfortunate
about the way that the criticism was made, though,
was that it was done as a blanket criticism of the
perfectly sound basic idea of what constituted a
prokaryote, rather than of the naive views of
molecular biologists about the matter. Molecular
biologists, being chemically rather than morphologically oriented, never took much notice of
Staniers concepts (1961, 1970 ; Stanier & Niel,
1962), which were mainly cell biological. Instead
they pictured a prokaryote in terms of such molecular genetic features as operon structure and
promoter organization in which E. coli was known to
differ from eukaryotes. Woese and Fox (1977 b) were
correct to criticize the assumption that all bacteria
would necessarily have these features, but failed to
realize that such details played no role whatever in
the basic organismic and cell-biological definition or
concept of bacteria. Exaggerating the biological

T. Cavalier-Smith
distinctiveness of Archaebacteria was, however,
historically useful in that it encouraged a horde of
very competent bacteriologists and molecular biologists to study them in the hope that they might
unearth some really radical differences in genetic
organization from that of E. coli and eukaryotes.
What this enormous enterprise has shown, however, is that Archaebacteria are not really very
different at all from E. coli in the organization of
their genes (e.g. in operons) and genomes, or their
replication, transcription and translation machinery,
as Keeling, Charlebois & Doolittle (1994) have
clearly accepted. The first complete genome sequence of eubacteria and an archaebacterium give
added force to the view that the genetic organization
of all bacteria is fundamentally the same, and
radically different from that of eukaryotes (Edgell et
al., 1996). Admittedly there are some differences,
but these (except for some aspects of transcription)
are ones of relatively small detail, mainly of interest
to the specialist molecular biologist rather than the
general bacteriologist, and these special properties of
archaebacteria are mostly shared with eukaryotes,
rather than indicative of a third form of life . These
differences pale into insignificance in comparison
with the immensely more profound differences
between the genetic organization of eukaryotes and
bacteria, which Stanier (1961, 1970) was largely
unaware of, but which I have discussed in detail
elsewhere (Cavalier-Smith, 1981 b, 1987 c, 1991 a, b,
1993 b). This conclusion clearly refutes the idea that
Archaebacteria and Eubacteria arose independently
from a crude precellular progenote with poorly
developed genetic mechanisms (Woese & Fox,
1977 b). That idea was based on the huge differences
between the two taxa revealed by 16S rRNA
oligonucleotide cataloguing. The way in which this
technique severely exaggerates the quantitative
differences was trenchantly criticized by Hori, Itoh
& Osawa (1982), and the validity of their criticism
has been confirmed fully by total rRNA sequences.
However, even at the time it was proposed it was
probably obvious to any cell biologist, but curiously
was not to most molecular biologists, that the idea
that eubacteria, archaebacteria and eukaryotes had
evolved independently could not possibly have been
correct. Long before then it was known that
eukaryotes and eubacteria share so many features of
cell organization and metabolism that their latest
common ancestor must have been a highly developed
cell with several thousand genes, and highly developed DNA replication machinery (Edgell &
Doolittle, 1997), and therefore could not have been

A revised six-kingdom system of life

a progenote with crudely-developed replication,
DNA repair, transcription and translation machinery (Woese & Fox, 1977 b). Belatedly, several authors
have become aware of this (e.g. Ouzonis & Kyrpides,
1996).
Woeses (1983) statement that it was impossible
for eukaryotes to have evolved from bacteria was a
gross overexaggeration. I previously attempted to
explain in detail how a bacterium could have evolved
into the first eukaryote, and why Woeses criticisms
(1983, 1987 ; see also 1994) of the concept of bacteria
are mistaken (e.g. Cavalier-Smith, 1987 c, 1991 a, b).
The evolution of rRNA is important, but one cannot
understand bacterial evolution by focusing mainly
on just one molecule (e.g. Woese, 1987). Bacteria are
cells and organisms, and it is these biological entities
(not single molecules) that systematists must classify.
Woese (e.g. 1987) has correctly criticized earlier
bacterial systematics for its phenetic rather than
phylogenetic approach, but this earlier approach
was as much through necessity as choice. Ribosomal
RNA has added a wealth of new characters without
which a phylogenetic approach would have been
much more limited. But to understand cell evolution
we need to consider very much more than just
rRNA.
I have felt obliged to go into this history in some
detail because the very early ideas of Woese & Fox
(1977 a, b) about the independent and early primary divergence of archaebacteria, eubacteria and
eukaryotes from a progenote , though not now
accepted even by the authors themselves, and the
thoroughly mistaken criticism of the concept of
bacteria and prokaryotes, have played a larger role
in determining the present views of many biologists
on the status of archaebacteria, than has a critical
assessment of the overall evidence now available to
us. I wish, however, to emphasize that my criticisms
of these particular early views in no way detracts
from the importance of the concept of archaebacteria
for understanding cell evolution and biological
diversity.

(3) The importance of cell structure in

bacterial classification
Though rRNA sequencing has quite clearly caused a
revolution, largely most beneficial, in bacterial
systematics, it is a great pity that it has sometimes
been associated with a neglect (and still worse a
positive disparagement) of morphology, which remains crucial for bacterial systematics. Sequence

217

data, morphology, and non-sequence chemical data

are complementary types of scientific evidence that
have to be integrated in a balanced way for sound
systematic decisions. It is historically incorrect to
imply that before rRNA sequencing there was no
phylogenetically sound bacterial taxonomy (Woese,
1987 ; 1994) : it was even less correct to imply that
this was true of microbiology in general, since for
eukaryotic microbes there are a wealth of good
morphological characters and over a century of good
phylogenetic studies and creation of durable higher
taxa. Even in bacteria it is a serious mistake to
dismiss morphology altogether (Woese, 1987) ; mere
differences in cell shape are indeed often relatively
trivial, but ultrastructural morphology in bacteria is
exceedingly important and phylogenetically informative. Long before any sequences became available
there were several phylogenetically sound groupings
of bacteria based on morphology, either alone or in
combination with chemical characters.
The three most obvious such groupings are
Cyanobacteria, spirochaetes and Gram-positive bacteria, all established in the nineteenth century, and
all supported much later first by electron microscopy
and then eventually also by rRNA phylogeny (for a
summary of the bacterial classification proposed
here see Table 2). The phylum Cyanobacteria is
exactly the same in definition and circumscription
as the Myxophyta of Cohn (1875) : rRNA only
confirmed what we already knew. For leptospiras
and spirochaetes, Stackebrandt & Woese (1981)
using trees based on rRNA oligonucleotide cataloguing suggested that they were not related. I never
accepted this, because the location, uniquely in
bacteria of the flagella in the periplasmic space
seemed such a complex and unusual morphological
character that it was highly unlikely to have evolved
twice. Therefore, my first bacterial classification
(Cavalier-Smith, 1987 a) treated them as a single
phylum. Full rRNA sequences agree with the
morphological conclusion that this phylum (Spirochaetae) is monophyletic (Olsen, Woese & Overbeek, 1994 ; Embley, Hirt & Williams, 1995).
This highlights the point often made by systematists, but still not fully appreciated by some molecular
biologists, that molecular characters are not inherently superior to morphological ones. There are
strong molecular characters (e.g. full 16s rRNA
sequences) and weak ones, just as there are strong
morphological characters (e.g. the periplasmic location of spirochaete flagella and the structure of
cyanobacterial thylakoids) and weaker ones (e.g.
rods versus cocci). What is still sorely needed in

218

T. Cavalier-Smith

Table 2. Classification of the kingdom Bacteria and its 10 phyla

Subkingdom 1. Negibacteria* Cavalier-Smith 1987 [outer membrane present ; acyl ester lipids ; murein wall
between the two membranes usually present ; small recognition particle RNA (based on Proteobacteria only at
present) ; in content nearly the same as the former phylum Gracilicutes].
Infrakingdom 1. Lipobacteria* infrak. nov. (murein cell wall ; outer membrane of phospholipids ;
lipopolysaccharide absent ; no flagellar shaft outside outer membrane).
Superphylum 1. Eobacteria* Cavalier-Smith 1982 (as an infrakingdom) stat. nov. (murein walls with
ornithine ; diaminopimelic acid, cytochrome aa3, and RuBisCo absent ; flagella absent ; small citrate synthetase).
Division or Phylum 1. Heliobacteria Cavalier-Smith 1987 (anaerobic photosynthesisers unable to fix CO ;
#
bacteriochlorophyll g and g ; no chlorosomes or b cytochromes).
Phylum 2. Hadobacteria Cavalier-Smith 1992 em. (thermophiles or radiation resistant ; if photosynthetic fix
CO ).
#
Subdivision or subphylum 1. Chlorobacteria Cavalier-Smith 1992 (aerobic photosynthetic thermophilic
gliders with bacteriochlorophyll c and b cytochromes ; e.g. Heliothrix, Chloroflexus).
Subphylum 2. Deinobacteria Cavalier-Smith 1986 (non-photosynthetic ; thermophiles or radiation
resistant with thick walls, e.g. Deinococcus, Thermus).
Superphylum 2. Endoflagellata Cavalier-Smith 1992 (flagella in periplasmic space).
Phylum 1. Spirochaetae Ehrenberg 1855 stat. nov.
Subphylum 1. Euspirochaetae subphyl. nov. (murein walls with ornithine, e.g. Treponema).
Subphylum 2. Leptospirae subphyl. nov. (murein walls with diaminopimelic acid, e.g. Leptospira)
Infrakingdom 2. Glycobacteria infrak. nov. (outer membrane with phospholipid in the inner leaflet and
lipolysaccharide in the outer leaflet of the bilayer ; RuBisCo present ; flagella pass through outer membrane).
Superphylum 1. Pimelobacteria* Cavalier-Smith 1992 (as infrakingdom) stat. nov. (murein wall with
diaminopimelic acid).
Phylum 1. Sphingobacteria Cavalier-Smith 1987 (sphingolipids ; flagella absent ; usually glide)
Subphylum 1. Chlorobibacteria subphyl. nov. (photosynthetic anaerobes with chlorosomes : Chlorobiaceae,
e.g. Chlorobium).
Subphylum 2. Flavobacteria (non-photosynthetic, e.g. Flavobacterium, Cytophaga).
Phylum 2. Eurybacteria* phyl. nov. (without sphingolipids or photosynthesis ; phylogenetically distinct from
proteobacteria ; monophyly of this phylum is more uncertain than for any other bacterial phyla, as clear
synapomorphies not identified).
Subphylum 1. Selenobacteria* Cavalier-Smith 1992 (as phylum) stat. nov. (non-fusiform ; often flagellate,
e.g. Selenomonas, Sporomusa).
Subphylum 2. Fusobacteria subphyl. nov. (fusiform, non-flagellate ; Fusobacterium, Leptotrichia).
Subphylum 3. Fibrobacteria subphyl. nov. (Fibrobacter).
Phylum 3. Cyanobacteria Stanier 1973 (oxygenic photosynthesizers with chlorophyll a).
Subphylum 1. Gloeobacteria subphyl. nov. (no thylakoids ; phycobilisomes on cytoplasmic membrane ;
Gloeobacter).
Subphylum 2. Phycobacteria subphyl. nov. (with thylakoids, e.g. Anabaena, Nostoc, Prochloron).
Phylum 4. Proteobacteria Stackebrandt et al. 1986 (without sphingolipids ; phylogenetically distinct from
Eurybacteria).
Subphylum 1. Rhodobacteria Cavalier-Smith 1987 (as phylum) stat. nov. (photosynthetic purple bacteria
and their colourless derivatives ; without chlorosomes, phycobilisomes or thylakoids ; photosynthetic machinery in
cytoplasmic membrane invaginations ; bacteriochlorophyll c and d and purple carotenoid in photosynthetic species ;
this subphylum and its three major subgroups are delimited primarily by ribosomal RNA similarity as clear
synapomorphies have not been found).
Infradivision or infraphylum 1. Alphabacteria Cavalier-Smith 1992 (as class) stat. nov. (with or
without photosynthesis, e.g. Rhodospirillum, Rickettsia, Agrobacterium).
Infraphylum 2. Chromatibacteria infraphyl. nov. (with or without photosynthesis ; - and proteobacteria, e.g. Escherichia, Haemophilus, Spirillum, Chromatium).
Subphylum 2. Thiobacteria subphyl. nov. (- and -proteobacteria ; non-photosynthetic, often sulphurdependent, e.g. Desulfovibrio, Thiovulum, Bdellovibrio, Myxobacteria, e.g. Myxococcus).
Superphylum 2. Planctobacteria superphyl. nov. (wall of protein, not murein peptidoglycan).
Phylum Planctobacteria Cavalier-Smith 1987 (Planctomycetales and Chlamydiae).
Subkingdom 2. Unibacteria new subkingdom (diagnosis : bacteria with single cytoplasmic membrane only, but
no outer membrane ; large signal recognition particle RNA).

A revised six-kingdom system of life

219

Table 2. (cont.)
Infrakingdom 1. Posibacteria* Cavalier-Smith 1987 stat. nov. (acyl ester lipids ; murein widespread).
Phylum Posibacteria* Cavalier-Smith 1987.
Subphylum 1. Teichobacteria subphyl. nov. (cell walls, if present, thick and with teichoic acids ; includes
Firmicutes and Mollicutes ; paraphyletic Firmicutes here abandoned as a formal taxon and subdivided between
Endobacteria and Actinobacteria).
Infraphylum 1. Endobacteria infraphyl. nov. (spores endospores ; DNA low in guanine and cytosine ;
classes Clostridea Cavalier-Smith 1982, e.g. Bacillus, and Mollicutes, e.g. Mycoplasma).
Infraphylum 2. Actinobacteria Margulis 1974 (as phylum) stat. nov. (spores exospores ; DNA high in
guanine and cytosine ; e.g. Corynebacterium, Streptomyces).
Subphylum 2. Togobacteria Cavalier-Smith 1992 (as phylum) stat. nov. (teichoic acids absent ; murein
wall very thin ; external non-lipid toga ; obligately anaerobic thermophiles : Thermotogales and Aquifex).
Infrakingdom 2. Archaebacteria Woese & Fox 1977 stat. nov. (isoprenoid ether lipids ; murein absent).
Phylum Mendosicutes Gibbons & Murray 1978.
Subphylum 1. Euryarcheota Woese, Kandler & Wheelis 1990 stat. nov. (energy metabolism various ; not
dependent on elemental sulphur).
Infraphylum 1. Halomebacteria Cavalier-Smith 1986 (methanogens or extreme halophiles, e.g.
Halobacterium, Methanospirillum).
Infraphylum 2. Eurytherma infraphy. nov. (non-methanogenic, non-halophilic thermophiles ;
Thermoplasma, Thermococcales, Archaeoglobales).
Subphylum 2. Sulfobacteria Cavalier-Smith 1986 stat. nov. (syn. Crenarcheota Woese, Kandler & Wheelis
1990 : energy metabolism depends on elemental sulphur ; Sulfolobales and Thermoproteales ; Jim Black Pool
thermophiles : e.g. Sulfolobus, Pyrobaculum).
This classification is modified from that in Cavalier-Smith (1992 a), where its cladistic basis is explained in more
detail. To save space diagnoses in Tables 27 are given only for new taxa.
* Probably paraphyletic.
Almost certainly paraphyletic.
All the phyla here should be treated as divisions for formal bacterial nomenclature, and the superphyla as
superdivisions, subphyla as subdivisions, and infraphyla as infradivisions. For uniformity with the other kingdoms of
life, I have used phylum rather than division in the present paper. The use of division rather than phylum is an historical relic of the origin of the Bacterial Code of Nomenclature from the Botanical Code rather than from the
Zoological Code, which impedes a unified approach to biological nomenclature and systematics. As the Botanical
Code now allows phyla and states that phylum and division are of equivalent rank, the Bacterial Code ought to be
similarly revised. When the Unified Code of Biological Nomenclature, now being planned for the next century
(Hawksworth, 1995), is introduced, I hope that it will adopt phylum, not division, for all organisms and that it
explicitly recognizes subkingdoms, infrakingdoms, superphyla, subphyla and infraphyla so as to fix their relative
rank, and also either superkingdoms or empires or both. By using phylum not division, one is free to use ordinary
English terms like subdivision in a general sense with no connotation of rank or risk of their being confused with the
taxonomic category subphylum. Haeckels (1866) name Monera is an unsuitable name for bacteria as it was
invented for hypothetical non-nucleate amoebae, which do not exist, and is much less widely known than Bacteria.
Originally treated as separate phyla (Cavalier-Smith, 1992 a) ; for molecular evidence for their monophyly see
Olsen, Woese & Overbeek (1994).

bacterial systematics, and has not been provided by

the sequence-oriented school, is the development of a
stronger tradition, like that established over two
centuries in eukaryotic systematics, of the critical
weighing of all available characters, whether molecular, chemical or sequence. Over-emphasis on one
molecule and neglect of important morphological
evidence is poor systematic practice.
The Gram-positive group, as defined by light
microscopy, was slightly heterogeneous. Electron

microscopy showed that the vast majority of Grampositives had only a single bounding membrane,
whereas the vast majority of Gram-negatives were
bounded by two separate membranes. I have long
argued that this morphological difference is of very
profound cell biological and evolutionary importance, and therefore coined the word Posibacteria for
all eubacteria with a single membrane, because their
most numerous members are the true Gram-positive
bacteria (Cavalier-Smith, 1987 c). The taxon Posi-

220

bacteria was originally both a phylum and a

subkingdom but is now ranked as both a phylum
and an infrakingdom. Posibacteria is not a synonym
for true Gram-positive bacteria (formally Firmicutes
or Firmibacteria), as is sometimes incorrectly
thought, because it also includes two other taxa :
these are mycoplasmas (Mollicutes : another morphologically based group confirmed by rRNA sequences both sequences and membrane chemistry
showed that the archaebacterium Thermoplasma
acidophilum was originally wrongly included, so
morphology had to be supplemented by this extra
evidence to refine the group) and Thermotogales.
Sequencing showed that mycoplasmas evolved from
the branch of Firmicutes with DNA of low guanine
and cytosine content by the loss of the murein
peptidoglycan wall (Woese, Stackebrandt &
Ludwig, 1985). This molecular sequence demonstration that Firmicutes and Mollicutes together are
monophyletic is congruent with the morphological
fact that both groups have a single membrane, not
two as in all other eubacteria. Thus the singleness of
the membrane and the acyl ester character of its
lipids are more important than the presence or
absence of the cell wall that was emphasized in the
earlier classification (Gibbons & Murray, 1978).
To contrast all the other eubacteria that are
bounded by two distinct lipid membranes with
Posibacteria, I named them Negibacteria (CavalierSmith, 1987 c). Negibacteria is not an exact synonym
for the vernacular term Gram-negative bacteria,
because it includes a few bacteria such as the genus
Deinococcus that are Gram-positive because they have
a thick wall between the two membranes, rather
than outside the single membrane as in the majority
of Gram-positives. However, it has the same circumscription as the division Gracilicutes (Gibbons
& Murray, 1978), but the higher rank of subkingdom. Therefore it could be argued that it was,
strictly speaking, an unnecessary new word, like the
term Cyanobacteria, a fifth synonym for Myxophyceae, Cyanophyceae, Myxophyta and Cyanophyta.
However, as Gracilicutes was of much lower rank,
not in wide currency, and there was no informal
vernacular equivalent, I thought it useful to create a
subkingdom name that would be euphonious in both
the vernacular (as negibacteria) and formally (as
Negibacteria), and which would contrast the group
directly with Posibacteria by using the prefix Negiand emphasize their bacterial character by using the
suffix -bacteria. The importance of euphony is
illustrated by the widespread use of the name
Archaebacteria in preference to the first formally

T. Cavalier-Smith
valid name Mendosicutes (Gibbons & Murray,
1978).
(4) Characters important in the high-level
classification of Bacteria
Archaebacteria, like Posibacteria, have only a single
bounding membrane. I shall therefore henceforward
refer to both groups collectively as Unibacteria.
Because of the differences in protein and lipid
targeting that it implies, I think that the difference
in membrane number between Negibacteria and
Unibacteria is the most important morphological
difference of all within the Bacteria (Cavalier-Smith,
1987 a, c ; 1991 a, b, 1992 a). By contrast, the differences in morphology between Archaebacteria and
Eubacteria are trivial or non-existent. The cells of
Archaebacteria and Posibacteria are organized in
fundamentally the same way ; there are no known
morphological differences between them by which
all archaebacteria can be distinguished. They have
morphologically indistinguishable flagella and gas
vacuoles. Archaebacteria are undoubtedly bacteria
by all 12 of Staniers (1961, 1970) criteria listed at
the beginning of subsection (1). Their genetic
systems are very similar to those of Posibacteria
(Doolittle & Brown, 1994 ; Keeling, Charlebois &
Doolittle, 1994), and their cell-division apparatus is
fundamentally similar (Wang & Lutkenhaus, 1996).
The complete sequence of the first archaebacterial
genome (Methanococcus jannaschii : Bult et al., 1996) in
comparison with those of the eubacteria Haemophilus
influenzae (a negibacterium : Fleischmann et al., 1995)
and Mycoplasma capricolum (a posibacterium Fraser et
al., 1995) confirms that both eubacteria and archaebacteria are fundamentally similar in genome organization and gene content, and share nearly 1000
different genes. Many, if not most, of the genes that
cannot at present be homologized across the eubacterial}archaebacterial divide may simply be rather
rapidly evolving ones, rather than genuinely unique
to either group. As Edgell et al. (1996) rightly stress,
the most striking thing about the eubacterial and
archaebacterial genomes is how similar they are ; as
these authors explain, the apparently high fraction
of archaebacterial genes not previously present in
databases is almost certainly an artefact of the
present low representation of archaebacterial sequences compared with those of eukaryotes and eubacteria, rather than a sign of the uniqueness of
archaebacteria as Bult et al. (1996) somewhat
misleadingly suggested. As more sequence becomes
available for the crenarcheote archaebacterium

A revised six-kingdom system of life

Sulfolobus solfataricus (Sensen et al., 1996) this artefact
will be reduced.
If one wished to divide Bacteria into two kingdoms
based on morphology then I would argue that
Negibacteria and Unibacteria would be a better
choice than Eubacteria and Archaebacteria.
In what non-morphological ways do archaebacteria and eubacteria differ that might be sufficient to justify their separation into separate
kingdoms ? They differ in wall chemistry. The
absence of murein in archaebacteria has often been
stressed but it is also absent in the posibacterial
mycoplasmas and in the negibacterial Planctobacteria. Probably, as I have previously argued
(Cavalier-Smith, 1987 a, c), this reflects three independent losses of murein : but this negative
character clearly would not even justify their
separation as a subkingdom, still less a kingdom.
More important is the presence in archaebacteria,
but apparently not in eubacteria, of N-linked
glycoproteins, a character that is shared with
eukaryotes and which I used to argue for a sister
group relationship between archaebacteria and
eubacteria (Cavalier-Smith, 1987 c). Given that Olinked glycoproteins are present in a few eubacteria,
as well as archaebacteria and eukaryotes, the
magnitude of the innovation is not immense. There
is also a most interesting difference in the chemistry
of the shaft of the flagellum in archaebacteria
(Fedorov et al., 1994). But none of these differences
is so profound individually or collectively to justify a
separate kingdom. The well known difference in
membrane lipid chemistry (isoprenoid ethers in
archaebacteria and acyl esters in eubacteria) is the
only other important qualitative chemical character
that clearly differentiates archaebacteria from eubacteria. I have long considered that this difference is
sufficiently important, coupled with the internal
metabolic diversity of archaebacteria, to give the
taxon some sort of supraphyletic rank. But why not
a subkingdom, infrakingdom or superphylum rather
than a separate kingdom ? The resemblances between archaebacteria and eukaryotes in transcription factors and some details of rRNA processing,
though supportive of a sister relationship between
the two groups (i.e. of the clade Neomura : CavalierSmith, 1987 c), involve insufficient differences from
the eubacterial pattern to justify any higher rank
than infrakingdom or superphylum.
Bacteriologists and molecular biologists seem to
have given rather little thought to the possibility of
using such taxonomic categories of intermediate
rank. Indeed, before the advent of rRNA sequencing

221

bacteriologists had no tradition of using categories of

high rank such as phyla ; virtually all classifications
used only classes, orders and families. There were no
phyla, subphyla, or subkingdoms at all : nothing
between classes (and precious few of them) and the
kingdom Prokaryota. After the recognition of
Archaebacteria four phyla were introduced : Gracilicutes, Firmicutes, Mollicutes (formerly a class), and
Mendosicutes (Archaebacteria) by Gibbons &
Murray (1978). Then suddenly, on the basis just of
rRNA sequence divergence, a dozen or more
separate bacterial kingdoms were suggested seriously
(Woese, 1987), and still seem to be favoured by
Olsen, Woese & Overbeek (1994), but have fortunately not been formally adopted by bacterial
taxonomists. From a system where major groups
such as spirochaetes and Cyanobacteria were given
much too low a rank in comparison with eukaryotic
taxa, there was a dramatic jump to excessively high
ranking of the major bacterial taxa.
The overall system of life must be well balanced if
it is to serve the needs of the general biologist as well
as those of the specialist in some particular group. It
is important for systematics and biology as a whole
that ranking within bacteria does not get drastically
out of step with that of eukaryotes.
The widespread ranking of Eubacteria and
Archaebacteria as separate kingdoms has never been
really critically discussed. The main justification
usually assumed for such a high rank is the large
difference in 16S rRNA sequence between these two
groups. This originally so impressed Woese that he
thought that archaebacteria must be a radically new
form of life (Woese & Fox, 1977 a). But what does the
large difference in this one molecule really amount
to biologically ? The molecule is over 50 % identical
between the two taxa and functions in much the
same way in both groups. There is no reason to think
that most of the differences are of substantial
functional significance : they probably are mainly
biologically meaningless chemical noise that has
accumulated by random mutation and genetic drift
of neutral or quasi neutral sequences (Kimura,
1983). As such, they are very important as phylogenetic markers, but do not constitute the sort of
organismal characters properly used in the classification and ranking of higher taxa. Quantitative
differences in rRNA would be a very arbitrary basis
for deciding rank, because rates of rRNA evolution
vary considerably, sometimes drastically (up to
about 50-fold), in different lineages and do not
correlate well with the occurrence of substantial
differences in body plan, which have always been

222

(and ought to continue as) the primary basis for

delimiting higher taxa. The differences between
archaebacterial and eubacterial rRNA are comparable in magnitude (though actually slightly less
than) those that separate the two eukaryote phyla
Microsporidia and Basidiomycota, which differ far
more from each other morphologically, and in way
of life, than do the two bacterial groups, yet are now
both included in the kingdom Fungi (see section
VII).
Ribosomal RNA on its own provides no sound
reason to rank Archaebacteria any higher than a
phylum. In fact, I consider that the rRNA differences deserve less weight than do the membrane
chemistry, the N-linked glycoproteins, and the
presence of a single bounding membrane in determining the rank of Archaebacteria. It is entirely
unwarranted to treat the two archaebacterial subphyla as separate kingdoms ; erecting a third
archaebacterial kingdom , Korarchaeota, just on
the basis of rRNA divergence, was most unwise,
especially as we know next to nothing about the
phenotypes of the organisms in question (Barns et al.,
1996). The discovery of these previously unknown
lineages, coupled with the uncertainty of the branching order at the base of the archaebacterial rRNA
tree and the known propensity of rRNA for highly
misleading non-clock like evolution (see discussion
below on microsporidia) should discourage the use of
rRNA branching depth alone as a basis for subdividing bacteria or any other organisms. When we
know more about the biology of the novel uncultured
archaebacteria, it may be appropropriate to subdivide archaebacteria into more than one phylum.
But at present I favour a single phylum for all
archaebacteria.
The above comments, from a predominantly
eukaryotic systematist with extensive experience in
ranking higher level microbial taxa, now entering
the bacterial field more seriously than before, are
intended to be constructive, and should not be
construed as personal criticisms of other biologists,
despite their critical character in places.
(5) Bacterial subkingdoms and
infrakingdoms
Taking both morphology and chemistry into account, we have, in fact, not two but three major
groups of bacteria that have to be grouped and
ranked : Archaebacteria, Posibacteria and Negibacteria. In deciding groupings and ranks systematists traditionally try to determine which differenti-

T. Cavalier-Smith
ating characters are ancestral and which are derived ;
for greater weight in grouping is accorded to shared
derived characters (synapomorphies) than to shared
ancestral ones. This brings me to the vexed question
of rooting the tree of life, which is far more difficult
than is usually thought and has not yet been
unambiguously achieved.
For over a century, most biologists have assumed
that eukaryotes evolved from bacteria, not vice
versa, both because it was much more reasonable to
suppose that the structurally simpler bacterial cell
preceded the much more complex eukaryotic one,
and because, since the 1950s, that view has been the
best interpretation of the now very extensive microbial fossil record (Schopf & Klein, 1992). Woese
& Fox (1977 a), however, relying solely on the rRNA
cataloguing evidence and the simple, but invalid
assumption of a molecular clock, proposed instead
that eukaryotes, archaebacteria and eubacteria were
equally old and had evolved independently from the
hypothetical progenote. For these three taxa, they
coined the novel term Urkingdom to emphasize their
hypothesis that the three groups were derived
independently as separate primary kingdoms. For
the next decade many molecular archaebacteriologists accepted this idea overdogmatically, even
though cell biological and palaeontological arguments were always opposed to it.
The situation was radically changed not so much
by my detailed arguments about cell evolution
(Cavalier-Smith, 1981 a, b, 1987 a, c), and the cladistic considerations that favoured a sister relation
between archaebacteria and eukaryotes (CavalierSmith, 1987 c), but by molecular trees for duplicated
proteins (Iwabe et al., 1989 ; Gogarten et al., 1989).
These clearly supported a sister relationship between
archaebacteria and eukaryotes and suggested that
they had separated from each other a substantial
time after the origin of life. If the topology of these
trees if correct, they provide cladistic evidence that
the eubacterial}eukaryote acyl ester lipids are the
ancestral type and the archaebacterial isoprenoid
ones are derived, as was argued earlier (CavalierSmith, 1987 a, c). If the eubacterial lipids are indeed
the ancestral type, then their common possession by
Negibacteria and Posibacteria is not a very strong
argument for linking them together in the taxon
Eubacteria. Eubacteria are a paraphyletic grade,
not a holophyletic clade as archaebacteria probably
are.
The likelihood that eubacteria are paraphyletic is
not in itself a reason for rejecting the taxon ; I have
treated them as a subkingdom or kingdom for many

A revised six-kingdom system of life

years despite believing them to be paraphyletic. But
it should cause us to ask seriously whether the best
place for the primary subdivision of bacteria is that
between archaebacteria and eubacteria or whether
it might not be better to place it within the
eubacteria between the Negibacteria and Posibacteria. It is really a question of whether one
weights more strongly the difference in number of
membranes or the differences in membrane chemistry and N-linked glycoproteins. I consider the
difference in membrane number to be of more
profound evolutionary significance, because there
have been a number of changes in membrane
chemistry, such as the invention of sterols, but
apparently only one changeover between two and
one bounding membrane (Cavalier-Smith, 1980,
1987 a, c ; 1991 a, b). I therefore here treat Negibacteria and Unibacteria as bacterial taxa of equal
rank, and both Posibacteria and Archaebacteria of
lower rank but equal to each other.
What rank should be assigned to these three taxa ?
As Negibacteria are sufficiently diverse in cell
structure and}or fundamental biochemistry to merit
subdivision into several phyla (eight being adopted
here), all three taxa must be ranked at a level above
that of phylum. While a reasonable case could be
made for treating Negibacteria and Unibacteria as
separate kingdoms, I prefer to treat them as
subkingdoms of a single kingdom Bacteria, for
basically the same reasons given below for the
kingdom Protozoa. It provides a simpler and more
balanced system for general reference purposes, and
maintains a good correspondence between the
kingdom name and the vernacular name bacteria
that has been in perpetual use for over 150 years and
is well known to the layman. If Unibacteria is a
subkingdom, it is most appropriate to make Archaebacteria and Posibacteria infrakingdoms. The proposed broad classification of bacteria and its place
within the six-kingdom system are summarized in
Table 1 ; the major taxa comprising each infrakingdom are indicated in Table 2.
The subkingdom Negibacteria is here divided into
two infrakingdoms based on the chemistry of the
cells outer membrane. One of these I call Glycobacteria, as it has lipopolysaccharide in the outer
leaflet of the bilayer of the outer membrane ; the
other, which lacks lipopolysaccharide and has
phospholipid in both leaflets of the outer membrane,
I name Lipobacteria, and suggest this is the ancestral
condition since lipopolysaccharide synthesis is immensely more complex than phospholipid synthesis,
and is therefore unlikely to have been present in the

223

first negibacterium. Infrakingdom Lipobacteria contains the three phyla which lack lipopolysaccharide
in their outer membrane ; two of these lack flagella
and have ornithine rather than diaminopimelic acid
(DAPA) in the cross-linking peptide of their murein
walls (the photosynthetic Heliobacteria and the
partly photosynthetic and partly heterotrophic
Hadobacteria ), while the Spirochaetae (typical
spirochaetes, with ornithine-containing muramopeptides and leptospiras with DAPA) have flagella
within the periplasmic space. Glycobacteria are
subdivided into two superphyla according to
whether they have murein walls or not. The
glycobacterial superphylum Pimelobacteria, unchanged in circumscription but slightly reduced in
rank compared with my previous classification
(Cavalier-Smith, 1992 a), have murein walls containing diaminopimelic acid and include the two
best-known negibacterial phyla, Cyanobacteria and
Proteobacteria, as well as the lesser known Sphingobacteria and Eurybacteria. Superphylum Planctobacteria includes only the Planctomycetales and
chlamydias (in a single phylum, the Planctobacteria), which lack murein. I have suggested
previously that the absence of murein in Planctobacteria is derived ; and the presence of murein in
the common ancestor of Lipobacteria and Pimelobacteria was the ancestral condition (CavalierSmith, 1987 a). If this and my hypothesis that the
lipobacterial lack of lipopolysaccharide is ancestral
are both correct, then to understand the early
evolution of negibacteria we need to focus attention
on the little-studied Lipobacteria. If my view that
the root of the tree of life should be placed within the
negibacteria is also correct, this means that to
understand the earliest divergences in the history of
life the study of Lipobacteria will be much more
important than that of Archaebacteria. It is, of
course, possible that the absence of lipopolysaccharide, flagella, DAPA and RuBisCo in eobacteria is derived (or derived in some but not other
taxa), but until there is definite evidence for this we
should take seriously the possibility that this was the
ancestral state for all bacteria and regard the
eobacteria as very good candidates for the earliest
diverging cells and thus of great potential evolutionary significance.
The present classification of Bacteria into two
subkingdoms, each further subdivided into two
infrakingdoms, has a pleasing symmetry and balance. The primary subdivision is according to the
number of envelope membranes, and the secondary
subdivision of each relies on membrane chemistry.

224

Each of the four infrakingdoms differs not only in

membrane number and or chemistry, but also in cell
wall chemistry or structure. Archaebacteria alone
have N-linked glycoproteins and sometimes pseudomurein ; Posibacteria alone have teichoic acids, and
typically very thick murein walls with peptide linkers
of very variable chemistry ; Glycobacteria have very
thin murein layers always with DAPA in the
crosslinking peptide ; Lipobacteria have usually thin,
but sometimes thick, murein walls typically with
ornithine in the linker. Thus, the special cytoplasmic
membrane chemistry of archaebacteria is given the
same taxonomic weight as the special outer membrane chemistry of the glycobacteria. Isoprenoid
ethers and lipopolysaccharides are the synapomorphies that can be used to define these two
infrakingdoms most clearly. The four-infrakingdom
system groups the 10 bacterial phyla recognized here
(a slightly lower number than that more tentatively
suggested by Woese, 1987) into organismally more
homogeneous higher taxa than does the archaebacterial}eubacterial dichotomy : the probably paraphyletic Eubacteria are so much more heterogeneous
organismally than archaebacteria, that their splitting into three taxa each of the same rank seems to
me a marked improvement and a more balanced
classification.
The major phylogenetic assumptions behind the
six-kingdom system are explicitly laid out in Fig. 1.
Unlike traditional bacterial classifications, the present system does not undervalue the diversity of
bacterial body plans in comparison with that of
higher kingdoms : as Table 1 makes clear, the
kingdom Bacteria has four major subdivisions of
rank infrakingdom or above, which is the same as in
the kingdom Plantae and more than the kingdoms
Fungi and Chromista, which have only two or three
respectively. Even Protozoa and Animalia, the most
megadiverse kingdoms have, respectively, only six
and 11. In this system the taxa Archaebacteria and
Archezoa are each treated as one of the 30 major
forms of life ; this sufficiently emphasizes their
distinctiveness previously Woese and I were probably both a little overenthusiastic in the use of the
unnecessarily high rank of kingdom. At the phylum
level, Bacteria are the third-most diverse of the six
kingdoms. The fact that five out of the eight
negibacterial phyla (including more than one in
each superphylum) are entirely or partly photosynthetic shows that the diversification of the basic
photosynthetic machinery (Pierson & Olson, 1989)
played a key role in early cell evolution and is
consistent with the view that the first cells were

T. Cavalier-Smith
photosynthetic (Woese, 1979 ; Cavalier-Smith,
1987 a). The presence of chlorosomes in Chloroflexus
(a lipobacterium) and Chlorobiaceae (Glycobacteria) suggests that the transition between the two
infrakingdoms involved a green photosynthetic bacterium with chlorosomes, thus supporting the idea
that these neglected bacteria are of pivotal evolutionary significance (Cavalier-Smith, 1987 a).
Whether the ranking of Archaebacteria, Posibacteria, Eobacteria and Lipobacteria as infrakingdoms and the abandonment of the taxon
Eubacteria will be accepted by others, only time will
tell. However, I hope that the reader will agree that
it is based on a more careful consideration of what
weight should be given to the major differences
within Bacteria, than the suggestion that archaebacteria and eubacteria be ranked equally as domains
(Woese, Kandler & Wheelis, 1990). These are
merely unnecessary new names for the earlier
suggested Urkingdoms or primary kingdoms of
Woese & Fox (1977 a) that were not well based. As
I stressed previously (Cavalier-Smith, 1986 c) there
are no good reasons to rank the three taxa
eukaryotes, archaebacteria and eubacteria equally
as kingdoms : merely renaming them domains
(Woese, Kandler & Wheelis, 1990) does not lessen
the criticisms of equal ranking at all (CavalierSmith, 1992 b) ; the new category domain is also
unnecessary given that we already have superkingdom and empire above the level of kingdom.
Equal ranking of these taxa produces a grotesquely
unbalanced system for living organisms as a whole.
The name eubacteria will, however, continue to be
very useful to denote a grade, as are invertebrate
and fish in zoology or alga in botany. The
suggestion that eubacteria be renamed bacteria
(Woese et al., 1990), was exceedingly confusing and
should not be followed (Cavalier-Smith, 1992 b).
Some may object to my placing the Thermotogales
and Aquifex in the Posibacteria on the grounds that
the rRNA tree does not provide evidence for such a
grouping. However, the rRNA tree neither provides
convincing evidence against such placing nor convincing evidence for any supraphyletic groupings
within eubacteria, as has been made clear by
Embley, Hirt & Williams (1995). On rRNA trees
even all Teichobacteria often do not group together ;
for their two major subgroups (Endobacteria and
Actinobacteria) often, but not always, separate from
each other (Olsen, Woese & Overbeek, 1994 ;
Embley, Hirt & Williams, 1995). It appears that the
eubacterial phyla (and sometimes also subphyla)
underwent an almost simultaneous diversification so

A revised six-kingdom system of life

that molecular trees cannot correctly resolve their
branching order, or even, in the case of the
Teichobacteria, confirm or refute their monophyly.
In such a circumstance, it is perfectly proper to base
phylogenetic reasoning and classification solely on
other evidence, which in some cases may be much
stronger than the sequence evidence. I have argued
that this major eubacterial radiation most likely was
the consequence of the origin of the first photosynthetic eubacterial cell and agree with Woeses
(1979) hypothesis that the first bacterial cell was
photosynthetic.
For reasons discussed in detail earlier (CavalierSmith, 1980, 1987 a, 1992 a), I have argued that the
first cell was a negibacterium with two membranes,
and that Unibacteria are secondarily derived by the
loss of the outer membrane. If this is true, Unibacteria and Eukaryota together form a clade that
may be referred to by the informal term unimembrana. The distinction between unimembrana
and negibacteria is profoundly important for cell
evolution and origins. The first cell must either have
been unimembranous, as is classically assumed (e.g.
Goldacre, 1958 ; Hargraeves & Deamer, 1978), or
negibacterial, as Blobel (1980) and Cavalier-Smith
(1987 a) have suggested. Until we can establish
firmly which idea is correct, we shall continue to
have a very shaky basis for understanding early cell
evolution. It might be thought that a combination of
the Iwabe et al. (1989) trees and the rRNA trees
refute the idea that negibacteria were ancestral since
they appear to place the root between Archaebacteria and Posibacteria, i.e. within Unimembrana.
But I remain unconvinced that we really know
where the root is yet, as do others (e.g. Forterre et al.,
1993 ; Doolittle & Brown, 1994). While the basic
logic behind using a duplicate protein gene as the
outgroup for the tree of its sister duplicate is good,
there is a practical problem arising from the fact that
the outgroup tree for one duplicate is a very long
branch compared with the branches within the tree
for the other duplicate. It is well known that such a
very long branch tends to be misplaced on trees
(Swofford et al., 1996). Even if the correct position of
the root were at the base of the negibacteria, treecalculation algorithms will tend to misplace the root
nearer to the mean position ; the more extreme the
differences in branch length the greater will be this
tendency. The problem is that the gene duplicates so
far studied appear to have diverged very strongly
from each other very soon after the duplication and
prior to the slower divergence of the major taxa. This
necessarily means that the outgroup branches are

225

excessively long in comparison with the ideal

situation for accurate tree-construction. The possibility of such an artefact, in which the root is placed
in the mid position rather than in the correct place,
makes the interpretation of all gene duplication trees
studied up to now rather uncertain.
It is important to stress that the uncertainty
regarding the position of the root of the tree of life
does not affect the validity of the bacterial subkingdoms and infrakingdoms proposed here. The
uncertainty simply means that we do not know
which of them are holophyletic and which are
paraphyletic. If I am correct, then Archaebacteria
are holophyletic and Posibacteria, Unibacteria,
Negibacteria, Glycobacteria and Lipobacteria are
all paraphyletic ; if, however, the tree of Iwabe et al.
(1989) is correctly rooted, then Negibacteria would
become holophyletic, as would one of Glycobacteria
and Lipobacteria (depending on their branching
order) ; the same taxa also become holophyletic if
Lake (1988, 1989) is correct and the root lies
between the two archaebacterial subphyla ; whereas
Archaebacteria become paraphyletic. Neither
grouping would make them polyphyletic. However
certain conceivable branching orders of the eubacterial phyla would make one or more of my taxa
polyphyletic and thus necessitate a revision. Accurately determining their branching order will be
the best test of monophyly of the major taxa in the
present classification. Both this and the fixing of the
position of the root may only be possible when a
complete genome sequence becomes available for at
least one member of each of the nine eubacterial
phyla. As four are already published Proteobacteria [Haemophilus influenzae (Fleischmann et al.,
1995) ; also E. coli and Helicobacter pylori] Cyanobacteria, Spirochaetae, and Posibacteria [Mycoplasma genitalium (Fraser et al., 1995)], we might
know the answer by the end of the century, provided
that whole genomes are sequenced for the other five.
It is however possible, as suggested by the 16S rRNA
tree, that the radiation was so explosively simultaneous that we shall never be able to resolve the
order fully, but I am optimistic that we can.
The fact that if one of these other ideas is correct
and I am wrong about the rooting of the tree, more
of the bacterial high-level taxa proposed here
become holophyletic, means that new knowledge
about the rooting can only increase the attractiveness
of the present system to strict Hennigian cladists, not
decrease it. However, if Archaebacteria are ever
shown to be paraphyletic (Lake, 1988, 1989 ;
Baldauf, Palmer & Doolittle, 1996), contrary to the

226

arguments of both Woese (e.g. 1987) and myself

(e.g. 1986 b, 1987 c), the taxon should not be
abandoned, just because of the dogmas of strict
Hennigians.

IV. PROTOZOA, THE BASAL EUKARYOTIC

KINGDOM

Treating Archezoa as a subkingdom of Protozoa

makes the present kingdom Protozoa very similar in
composition to Owens (1858) original kingdom
Protozoa, apart from diatoms now being properly
placed with other ochrophyte algae in the kingdom
Chromista, and Bacteria being separated into their
own kingdom. It maintains in a single kingdom the
vast majority of the organisms that have been
treated as protozoa over most of the 150 years since
Von Siebold (1845) restricted the term to unicellular
organisms, as summarized in Table 2. For approaching two centuries, since its invention by
Goldfuss (1817), the vernacular term protozoa has
been so widely used by biologist and layman alike
[for example, by the poet Coleridge (1834)], that
there is real value in keeping as close as possible to
the historically dominant meaning without compromising the principles of Darwinian classification.
Protozoa as a major taxon has, in fact, proved to be
more stable than the remarkable variety of different
interpretations we have seen over the past 50 years of
the far more heterogeneous Protista of Haeckel
(1866). Protists in Haeckels sense of unicellular
organisms are now distributed in the present system
across all six kingdoms. However, as I stressed
previously (Cavalier-Smith, 1981 a), the vernacular
word protist remains a most useful term for designating the polyphyletic grade of unicellular eukaryotes, even though a kingdom Protista or Protoctista
(in addition to being polyphyletic) would be much
too heterogeneous to be taxonomically meaningful.
(1) Status of Archezoa, early diverging
amitochondrial eukaryotes
The original concept of the Archezoa was phylogenetic. It was argued that a symbiotic origin of
mitochondria could not have occurred until after the
evolution of phagocytosis (Cavalier-Smith, 1983 b).
Since phagocytosis is restricted to eukaryotes
(Stanier, 1961, 1970), it followed that the origin of
the eukaryote cell and phagocytosis had almost
certainly preceded the symbiotic origin of mitochondria (Cavalier-Smith, 1983 b), not the reverse as

T. Cavalier-Smith
had generally been postulated before (Margulis,
1970, 1981). Thus primitively amitochondrial
eukaryotic cells must once have existed (CavalierSmith, 1983 b). Although there was no guarantee
that they had not all gone extinct after the origin of
mitochondria, I considered it rather unlikely that
this would have occurred, since niches for anaerobic
phagotrophs must always have existed somewhere in
the environment, in which amitochondrial eukaryotes would have been at no disadvantage compared
with eukaryotes with mitochondria. Therefore, I
postulated that at least some of the known amitochondrial eukaryotes were primitively so, and defined the subkingdom Archezoa as comprising all
primitively amitochondrial eukaryotes (CavalierSmith, 1983 b). I was perfectly well aware that some
amitochondrial eukaryotes had secondarily lost
mitochondria (i.e. certain ciliates and fungi) and
that some, or conceivably even all, of the taxa that I
initially placed in Archezoa might well also have lost
mitochondria. But in a Popperian spirit I deliberately chose to frame, what I explicitly referred to as
my taxonomic hypothesis , in the most strong and
therefore most easily refutable form by including in
Archezoa all amitochondrial taxa which then had no
known mitochondrial ancestry.
(a) Changes in circumscription of the Archezoa
Later, after coming to the view that all hydrogenosomes had probably evolved from either peroxisomes or mitochondria (Cavalier-Smith, 1987 c), I
decided that the double-membraned hydrogenosomal envelope of Parabasala was therefore probably
homologous with the mitochondrial envelope, and
therefore removed Parabasala from Archezoa
(Cavalier-Smith, 1987 a). This change in circumscription of the group, contrary to what is sometimes
implied, did not change the phylogenetic definition
of the group. It was simply a change in view as to
whether Parabasala satisfied that definition or not.
As previously discussed (Cavalier-Smith, 1993 a), the
evidence for a mitochondrial origin for parabasalan
hydrogenosomes was, for a long time, rather equivocal. However, there is no good evidence against it,
and what is known about protein-targeting to
hydrogenosomes is consistent with the view that
their targeting system evolved from that of mitochondria. This is simpler than the alternative idea of
an entirely independent symbiogenetic event
(Mu$ ller, 1992), or an even less likely origin from the
endomembrane system. Recent evidence from the
proteobacterial affinities of the heat shock protein

A revised six-kingdom system of life

hsp 60 of trichomonads (Bui, Bradley & Johnson et
al., 1996 ; Germot, Philippe & Le Guyader, 1996 ;
Horner et al., 1996 ; Roger, Clark & Doolitttle, 1996)
provides the first strong molecular evidence that
Parabasala are indeed secondarily amitochondrial,
and therefore gives indirect support for a mitochondrial ancestry for their hydrogenosomes.
To those who believe that classifications should
not be based on hypotheses, I simply say that all
phylogenetic classifications are based on similar sorts
of phylogenetic hypotheses. The degree to which a
phylogenetic hypothesis needs to be corroborated
before being used as a partial basis for a classification
is a matter for scientific judgement by individual
systematists. It is neither philosophically correct nor
good manners to call the phylogenetic hypotheses of
others speculation, and to treat ones own as
accepted facts. It is an error to assert that Archezoa
were compositionally defined (Patterson, 1988).
Although many rRNA trees show Parabasala as
branching just below all mitochondrial eukaryotes,
some show them instead as branching just above the
mitochondria-bearing Percolozoa (Cavalier-Smith,
1993 a,). The latter position is inconsistent with the
idea that Parabasala are primitively amitochondrial
archezoa (that is, if mitochondria are monophyletic),
but both positions are consistent with the view that
they are secondarily amitochondrial ; which position
is correct is uncertain (Cavalier-Smith & Chao,
1996 a), but as several protein trees place the
divergence of Parabasala lower than that of the
Percolozoa, this position is more likely to be correct.
Parabasala are now treated as a subphylum within
the protozoan phylum Trichozoa, all of which have
hydrogenosomes instead of mitochondria (CavalierSmith, 1997 a). Trichozoa, Percolozoa and Euglenozoa were recently grouped together as the protozoan
subkingdom Eozoa to contrast them with the
putatively more advanced protozoa placed in the
subkingdom Neozoa (Cavalier-Smith, 1997 a).
The phylogenetic evidence from rRNA that the
archamoebae Entamoeba, Phreatamoeba (Hinkle et al.,
1994), and Pelomyxa (Morin & Mignot, 1996) are all
secondarily amitochondrial made it desirable to
remove the phylum Archamoebae as a whole from
the Archezoa, and to place them in the Neozoa
alongside the amoebae with mitochondria (Lobosa
and Filosea) (Cavalier-Smith, 1997 a). The
evidence that they are secondarily without mitochondria is relatively clearcut, as they branch well
above the deepest branching point of mitochondrial
taxa. As they sometimes even group with the Lobosa
in maximum-likelihood trees, Lobosa and Arch-

227

amoebae have each recently been ranked as subphyla within the revised phylum Amoebozoa
(Cavalier-Smith, 1997 a). In addition to the positive
cladistic evidence for a mitochondrial ancestor, there
is even more direct evidence from hsp 70 sequences
that Entamoeba histolytica evolved from an ancestor
with mitochondria (Clark & Roger, 1995), but in
contrast to the situation in Parabasala these were not
converted into hydrogenosomes.
Unlike Trichozoa, the Metamonada and Microsporidia have no hydrogenosomes and branch at the
very base of the eukaryotic rRNA tree, and it has
been widely thought that both taxa are primitively
amitochondrial. However, I stressed previously
(Cavalier-Smith, 1993 a ; see also Cavalier-Smith &
Chao, 1996 a), that this conclusion might not be
correct. The first evidence, apart from their obligate
intracellular parasitism, that Microsporidia might
be less primitive than metamonads was the discovery
of spliceosomal RNA in microsporidia (preliminary
data cited in Cavalier-Smith, 1993 a ; DiMaria et al.,
1996) ; this makes it possible that Metamonada are
the only eukaryotes that primitively lack spliceosomal introns. If spliceosomal introns arose from
group II self-splicing introns that entered eukaryotes
within the genes of the -proteobacterium ancestral
to mitochondria, as postulated (Cavalier-Smith,
1991 c), this would imply that microsporidia evolved
from ancestors that had mitochondria as previously
emphasized (Cavalier-Smith, 1993 a). Two lines of
evidence now strongly support this view and suggest
that Microsporidia are really degenerate fungi, and
must be removed from the Archezoa. First were trees
based on sequences of -tubulin (Keeling &
Doolittle, 1996 ; Li et al., 1996) and -tubulin (Edlind
et al., 1996 ; Li et al., 1996 ; Roger, 1996), which
group the Microsporidia very robustly with the
Fungi, not the Protozoa. Second, sequence trees for
the molecular chaperone hsp70 equally convincingly
group the Microsporidia with the Fungi (Germot et
al., 1997) ; since the chaperone in question is the
mitochondrial type related to that of the -proteobacteria, there seems little doubt that microsporidia
are indeed secondarily amitochondrial. A third, but
much less convincing, piece of evidence is that the
protein synthesis elongation factor protein EF 1- of
microsporidia has an insertion at exactly the same
site as does that of all Fungi, animals and choanozoan protozoa (collectively known as opisthokonta) ;
however, this insertion is a little shorter and only
weakly similar in sequence to that in opisthokonta.
Moreover the EF 1- protein sequence tree does not
group the microsporidia with the Fungi but places

228

them near the bottom of the eukaryotic tree, just as

does the small subunit rRNA tree.
Thus there is a dramatic conflict between the
tubulin and hsp70 trees on the one hand and the EF
1- and small subunit rRNA trees on the other.
Since the two latter trees do not specifically group
microsporidia with any other taxa, and in both cases
the microsporidial branches are very long, the
simplest way to reconcile the data are to suggest that
the rRNA and elongation factor genes of microsporidia have undergone a drastically increased rate
of evolution compared with all other eukaryotes and
that the deep position in which they appear is a
grossly misleading systematic error produced by the
phylogenetic algorithms, none of which can give the
correct tree if branches of sisters differ many-fold in
length. There are a number of other instances where
rRNA and certain protein trees have been grossly
misleading, and tree-calculation algorithms have
been proven to give the wrong answer quite
reproducibly in certain circumstances (Felsenstein,
1978 ; Hillis, Huelsenbeck, & Cunningham, 1994 ;
Swofford et al., 1996).
Though initially a surprise, the fungal nature of
microsporidia is not at all unreasonable. Both groups
have spores with walls of chitin, and a fair number of
fungi are parasites of animals. A further similarity
with higher Fungi is that the vegetative cells do not
have Golgi dictyosomes visible in the electron
microscope as stacks of cisternae. Higher Fungi are
unusual among higher eukaryotes in lacking Golgi
stacks ; the only Fungi that have Golgi stacks like
most other eukaryotes are the Dictyomycotina,
which include Chytridiomycetes sensu stricto, the
most basal group (Cavalier-Smith, 1987 b). The
absence of Golgi stacks in microsporidia suggests
that they are not a sister group to the fungi as shown
by some of the protein trees but, as Roger (1996)
suggests, actually evolved from within the fungi after
the allomycete ancestor underwent an evolutionary
unstacking of its Golgi cisternae (Cavalier-Smith,
1987 b).
It has long been known that the mitotic mechanism of microsporidia is remarkably similar to that
of ascomycetes (Raikov, 1982). In fact their mitosis
is indistinguishable from that of ascomycetes ; in
both groups mitosis is closed with an intranuclear
spindle and a flattened centrosomal plaque at the
spindle poles, either embedded in or just outside the
nuclear envelope. Most other fungi do not have this
type of centrosomal plaque ; in basidiomycetes the
centrosomes (the recently fashionable fungal term
spindle pole body is quite unnecessary) are

T. Cavalier-Smith
globular and cytoplasmic, except in Uredomycetidae
where they are a unique trilaminar plaque, whereas
in Archemycota they are very varied in structure :
centrosomal plaques occur in a few Zygomycotina.
The similarity of mitotic mechanism between ascomycetes and microsporidia was earlier assumed to be
convergent, partly because a similar mechanism is
also found in the malaria parasite, Plasmodium, which
must have evolved it independently from Fungi. The
mycetozoan Dictyostelium also has centrosomal
plaques, though mitosis is semi-open, so it is clear
that centrosomal plaques have evolved at least three
times following independent losses of centrioles.
Though the mitotic mechanism is clearly, therefore,
on its own not good evidence for a relationship with
ascomycetes and Zygomycotina, it does in conjunction with the tubulin and hsp70 sequences
render such a relationship highly probable.
Microsporidia commonly have a binucleate phase
in the life cycle, similar to the dikaryophase of
ascomycetes and basidiomycetes (Canning, 1990).
Because in microsporidia the two nuclei are physically attached via the surfaces of their nuclear
envelope this binucleate condition is called diplokaryotic (Canning, 1990). Some protozoologists
have suggested that this diplokaryotic condition is a
modification of the dikaryotic condition of higher
Fungi, and that microsporidia may be derived from
higher Fungi (Desportes & Nashed, 1983). However,
in my view diplokaryosis is distinct from and
convergent with the fungal dikaryophase (CavalierSmith, 1995 a).
The absence of mitochondria and peroxisomes
from microsporidia is no reason for excluding them
from the Fungi. Both organelles are also absent in
the rumen fungi of the order Neocallimastigales ;
however, these anaerobic fungi do have another
respiratory organelle, the hydrogenosome, which
may be evolutionarily derived from mitochondria or
possibly peroxisomes (Cavalier-Smith, 1987 d ;
Marvin-Sikkema et al., 1993). Because of the presence of Golgi stacks, centrioles and cilia in Neocallimastigales it is most likely that microsporidia lost
mitochondria and peroxisomes independently of
them, and had either a zygomycote or a hemiascomycete ancestor (see section VI). The extremely
degenerate character of the microsporidian fungi,
which caused them to be misclassified for over a
century as protozoa, is one of the most striking
examples known of evolutionary degeneration, the
importance of which was first strongly argued by
Dohrn (1875). Such degeneration greatly complicates the task of phylogenetic reconstruction, as

A revised six-kingdom system of life

Lankester (1877, 1880) emphasized, since its attendant simplification and loss of ancestral characters is so easily confused with primitive simplicity.
Time and again secondarily simplified organisms
have been misclassified with genuinely primitive
ones.
The clear demonstration that Trichozoa, Archamoebae and Microsporidia are all secondarily
amitochondrial leaves only the phylum Metamonada, which appears to have diverged from all
other eukaryotes earlier than any others (CavalierSmith & Chao, 1996 a), as possibly primarily
amitochondrial. Transferring Microsporidia to the
kingdom Fungi, left only Metamonada in the
Archezoa, which I now treat as a subkingdom of
Protozoa (Cavalier-Smith, 1997 b), as it was originally (Cavalier-Smith, 1983 a), rather than as a
separate kingdom (Cavalier-Smith, 1987 a ; 1993 a).
If Metamonada lack both mitochondria and peroxisomes primitively (Cavalier-Smith, 1983 a), then
they would be radically different from all higher
eukaryotes (collectively designated Metakaryota :
Cavalier-Smith, 1987 a). But it seems increasingly
unlikely that they are indeed genuinely primitively
amitochondrial, contrary to widespread assumptions. The published evidence that diplomonads
may have arisen by the loss of mitochondria (Keeling
& Doolittle, 1987 ; Soltys & Gupta, 1994) is not yet
very convincing, but firmer sequence evidence for a
secondarily amitochondrial character of the group
may soon be forthcoming from heat shock proteins
(Roger and Sogin, personal communication) and
aminoacyl tRNA synthetases (Hasegawa, personal
communication).
If, as seems likely, Metamonada are indeed
secondarily amitochondrial, then the distinction
between them and the Parabasala becomes less
fundamental than I thought when I removed
Parabasala from the Archezoa on the grounds that
their hydrogenosomes probably arose from mitochondria (which is now generally accepted). Moreover, Archezoa can no longer be distinguished from
other eukaryotes on the grounds that they are
primitively amitochondrial : we must therefore reconsider the definition and circumscription of the
Archezoa and the other protozoan subkingdoms
(Cavalier-Smith, 1997 a). Metamonada and Trichozoa are anaerobic or microaerophilic flagellate
protozoa that lack mitochondria and have kinetids
that ancestrally have four centrioles, not two as in
higher eukaryotes. They also appear to be the two
earliest branching eukaryote phyla on several different protein sequence trees (e.g. RNA polymerase ;

229

heat shock protein ; -tubulin, -tubulin). At present

there is no evidence that they have any spliceosomal
introns, unlike all higher eukaryote phyla.
Given these distinctive features and their apparently early divergence I here revise the subkingdom Archezoa to include both Metamonada
and Trichozoa (as it did originally : but now
Microsporidia and Archamoebae are excluded).
Though it now appears that Archezoa in this revised
sense are secondarily without mitochondria, there is
no evidence that they ever had peroxisomes. It is
possible therefore that they are primarily without
peroxisomes, unlike all higher eukaryotes. If this is
true and if, as De Duve (1984) and I (CavalierSmith, 1987 c) have argued, peroxisomes are derived
symbiogenetically from bacteria, then Archezoa may
be less chimaeric in origin than all higher eukaryotes.
In postulating that the presently redefined Archezoa may be primitively without spliceosomal introns
and peroxisomes, I do not rule out the possibility
that they have lost them, or that some spliceosomal
introns may one day be found. I am, as in my earlier
discussions of the Archezoa, merely following Lankesters (1877) methodologically sound principle that
until we have special reason to take a different view of
any particular case we are bound to make the
smallest amount of assumption by assigning to the
various groups of organisms the places which they
will fit into, on the supposition that they do represent
in reality the original progressive series .

(2) The protozoan subkingdoms : Archezoa

and Neozoa
The recent grouping of the phyla Trichozoa,
Euglenozoa and Percolozoa together as the subkingdom Eozoa (Cavalier-Smith, 1997 a) was heavily
influenced by the fact that on rRNA trees these three
taxa consistently diverge earlier from all other
mitochondrial eukaryotes (neokaryotes : CavalierSmith, 1993 b) than the latter do from each other.
However the recent demonstration that Microsporidia are Fungi not Archezoa, shows just how
grossly misleading the rRNA tree can sometimes be.
This highlights early arguments that rRNA gene
evolution must be substantially non-clock like and
subject to gross systematic biases in evolutionary rate
(Cavalier-Smith, 1980). The recent demonstration
that the histionid flagellate Reclinomonas americana has
the most primitive known mitochondrial genome
organisation (Lang et al., 1997), when coupled with
the fact that rRNA trees place Reclinomonas above

230

both Percolozoa or Euglenozoa (Cavalier-Smith &

Chao, unpublished) adds further fuel to my earlier
suspicions that rRNA trees may place the Euglenozoa substantially too low (Cavalier-Smith, 1993 c,
1995 a). Reclinomonas is the only known eukaryote to
have retained the -proteobacterial RNA polymerase genes in its mitochondrial genome. It seems
unlikely that their replacement by nuclearly encoded
phage T3}T7 type RNA polymerases took place
more than once in other eukaryotes. This makes it
likely that Euglenozoa and Percolozoa arose from
neozoan ancestors in which this had already occurred. If true, then the rRNA tree is grossly
misleading as to their position and the Eozoa are
polyphyletic rather than paraphyletic as earlier
suggested (Cavalier-Smith, 1997 a). In view of this, I
here remove the Euglenozoa and Percolozoa from
the Eozoa and group them together as a new
infrakingdom, Discicristata, which I place in the
subkingdom Neozoa.
Such a position is consistent with my earlier
argument that the triple enveloped chloroplast
envelope of dinoflagellates and euglenoids is a unique
shared derived character that links the two groups,
and which could have arisen through a primary
endosymbiosis, not a secondary endosymbiosis as
commonly assumed (Cavalier-Smith, 1982, 1995 a).
The presence of articulins in both alveolates (presently known only in ciliates) and euglenoids is
consistent with a closer relationship between Alveolata and Discicristata than is suggested by the 18S
rRNA tree. The fact that certain protein trees,
notably tubulins and EF 1- (Baldauf & Palmer,
1993) place Euglenozoa and Percolozoa as a single
clade near the green plants and not below the
neozoan radiation, is consistent with a primary
origin for the euglenoid chloroplast (Cavalier-Smith,
1982) and favours the view that 18S rRNA trees
place Discicristata too low, just as they do Microsporidia and Mycetozoa. Though the tetrakont
character of some Percolozoa and the absence of
Golgi dictyosomes were reasons for suggesting earlier
that they might be the most primitive mitochondrial
eukaryotes (Cavalier-Smith, 1993 c), the recent mitochondrial data for Reclinomonas suggest that histionids
may be more primitive. Detailed studies are needed
of mitochondrial genomes and a variety of nuclear
protein-coding genes in Percolozoa and a much
greater variety of tubulicristate flagellates in order to
test this and the present classification more
thoroughly.
If Discicristata are really part of the neozoan
radiation, as argued here, then the only really early

T. Cavalier-Smith
diverging eukaryote phyla are the Metamonada and
Trichozoa. To retain separate subkingdoms for these
two taxa (Archezoa and Eozoa) seems less useful
than to group them together in a single subkingdom,
as suggested above. Archezoa thus remains as the
basal paraphyletic protozoan and eukaryote subkingdom, as in the original 6-kingdom system
(Cavalier-Smith, 1983 a) ; but its phylogenetic definition is modified Archezoa comprise the two phyla
that diverged prior to the divergence of all the
mitochondria-containing groups. Contrary to what
was previously suggested (Cavalier-Smith, 1983 a),
the latest common ancestor of the Archezoa had
probably already started to incorporate at least some
of the genes from the symbiont that was eventually
converted into the mitochondrion, and therefore was
probably not a non-chimaeric eukaryote, as postulated earlier (Cavalier-Smith, 1983 b). Whether this
archezoan cenancestor had a fully developed mitochondrion or only a precursor of mitochondria is a
semantic rather than a factual issue. In my view
however it is proper to call the common ancestor of
the trichomonad hydrogenosome and the mitochondrion a mitochondrion since it must have been at
least facultatively aerobic with cytochromes and
must already have evolved an organelle-specific
protein import mechanism of the type shared
between the two organelles (Bradley et al., 1997), the
best demarcation criterion between an obligate
symbiont and an organelle. Therefore trichomonad
hydrogenosomes did evolve from mitochondria, as I
postulated (Cavalier-Smith, 1987 d). The suggestion
that the accepted common ancestor of these hydrogenosomes and mitochondrion might not be a
mitochondrion but merely a precursor of a mitochondrion (Bui, Bradley & Johnson, 1996) is a
distinction without substance since no properties are
suggested by which the non-mitochondrial precursor might be distinguished from a mitochondrion.
Even though one cannot (at least at present) deduce
whether or not the mitochondrial protein-import
mechanism had evolved prior to the divergence of
Metamonada and Trichozoa, it is probably best not
to continue to think of Archezoa as being primitively
amitochondrial. However, the reasons for postulating that the -proteobacterium was taken up by
an early amitochondrial and non-chimaeric eukaryote after the origin of the cytoskeleton, endomembrane system and phagocytosis (CavalierSmith, 1983 b, 1987 d) remain compelling, even if no
primitively amitochondrial descendants of this early
non-chimaeric phase of eukaryote evolution remain.
If Discicristata really occupy the position shown in

A revised six-kingdom system of life

Fig. 1, then it is necessary to redefine the term
neokaryotes, originally defined as the clade including
all eukaryotes that branch higher on rRNA trees
than Euglenozoa (Cavalier-Smith, 1993 b), because
this definition no longer defines a clade. I suggest
that neokaryote should now signify the clade
comprising Neozoa plus the four higher kingdoms of
life. Eukaryotes may therefor be divided simply into
tetrakont Archezoa, putatively without spliceosomal
introns, and the basically bikont neokaryotes, typically with spliceosomal introns (with the exception
of the euglenozoan Kinetoplastea, which probably
lost them. The present redefinition of both Archezoa
and neokaryote make the terms eokaryote (CavalierSmith 1993 b) and megakaryote (Cavalier-Smith,
1995 a) no longer necessary. Metakaryota (CavalierSmith, 1987 a) are not a taxon in the present system ;
but following the recognition that microsporidia and
Archamoebae are secondarily amitochondrial and
branch higher up molecular sequence trees than do
Parabasala, the name metakaryotes (CavalierSmith, 1987 a) must now include both microsporidia
and archamoebae and, therefore, now designates the
clade comprising all eukaryotes other than Metamonada.
(3) Demarcation between the two zoological
kingdoms : Protozoa and Animalia
Now that the phylogenetic position of Myxozoa has
been established by rRNA phylogeny (Smothers et
al., 1994 ; Schlegel et al. 1996), it is clear that their
vegetative unicellularity is secondarily derived as a
result of parasitism. It seems to be only a remote
possibility that their position on rRNA trees well
within the metazoan animals is a methodological
artefact of long-branch attraction. While other
molecular data are clearly needed, the present data
are convincing enough to make it highly probable
that Myxosporidia arose from a multicellular ancestor that possessed not only cell junctions, as do
their multicellular spores, but also collagenous
connective tissue, a nervous system and a gut.
Therefore Myxozoa must be excluded from the
kingdom Protozoa and placed within the kingdom
Animalia, as stated earlier (Cavalier-Smith,
1995 b, c). Since it is unclear whether they are
derived from Cnidaria, as is suggested by their
nematocyst-like extrusomes (Weill, 1935 ; Lom & De
Puytorac, 1965 ; Siddall et al., 1995) or are closer to
bilateral animals, as is weakly suggested by the
rRNA trees, at present Myxozoa are most appropriately excluded from both subkingdoms Radi-

231

ata and Bilateria and ranked as a third subkingdom

of the kingdom Animalia (Cavalier-Smith et al.,
1996 a). The great difference in phenotype between
Myxozoa and both Radiata and Bilateria also
justifies this high rank ; to place them in either
subkingdom would make it phenotypically too
heterogeneous. Long ago Lankester (1877) noted
that it was very hard to disprove the idea that many
of the Protozoa are not descended from Enterozoa
by degeneration : it appears that only the Myxozoa
have actually done so.
Recent molecular phylogenetic evidence that
dicyemid Mesozoa (Katayama et al., 1995) and
orthonectid Mesozoa (Hanelt et al., 1996) are related
to bilaterians means that they also have lost a
nervous system, gut and collagenous connective
tissue, another remarkable example of degeneration
raised as a possibility by Lankester. Though the
molecular data suggest that the two mesozoan classes
may not be directly related to each other (Hanelt et
al., 1996) the trees lack strong resolution, so the
closest relatives of both groups are uncertain and it
remains possible that Mesozoa are monophyletic ;
therefore I retain the phylum Mesozoa until such
time as its monophyly is more convincingly disproved, but in view of the radical differences from
Bilateria continue to place it in a distinct subkingdom Mesozoa (Cavalier-Smith, 1983 a, Cavalier-Smith et al., 1996 a). Whether Mesozoa are
monophyletic or not, it is now clear that dicyemids,
orthonectids and Myxozoa all arose by the loss of the
nervous system and of obvious collagenous tissue ;
thus it is no longer justifiable to use the absence of
collagenous connective tissue as a reason for excluding Mesozoa from Animalia (Cavalier-Smith,
1983 a, 1993 a). The complex life cycles of the
Mesozoa and their copulatory sex have long been
reasons for considering that they may have evolved
from a flatworm-like ancestor by the loss of the
nervous system (Stunkard, 1954 ; 1972) ; unfortunately the base of the bilaterian rRNA tree is a bushlike radiation in which no branching orders are
clearly resolved, which is consistent with the explosiveness of the early Cambrian radiation seen in the
fossil record. The fact that their mitochondrial
cristae are tubular not flat, evidence previously cited
in support of a protozoan character (Cavalier-Smith,
1983 a), is not sufficient reason for excluding them
from Animalia, since the cristae of flatworms can
also be tubular rather than the flat cristae of most
animals other than Ctenophora.
The generalization that animals have flat cristae
(Taylor, 1978) is clearly an oversimplification. It has

232

T. Cavalier-Smith

Table 3. Classification of the kingdom Protozoa and its 13 phyla

Subkingdom 1. Archezoa Cavalier-Smith 1983 em.
Phylum 1. Metamonada Grasse! 1952 stat. nov. et em. Cavalier-Smith 1981.
Subphylum 1. Eopharyngia Cavalier-Smith 1993 (e.g. Giardia, Hexamita, Trepomonas, Chilomastix).
Subphylum 2. Axostylaria Grasse! 1952 stat. nov. em. Cavalier-Smith 1993 (e.g. Oxymonas, Pyrsonympha).
Phylum 2. Trichozoa Cavalier-Smith 1997.
Subphylum 1. Anaeromonada Cavalier-Smith 1997 (Trimastix).
Subphylum 2. Parabasala Honigberg 1973 stat. nov. Cavalier-Smith 1997 (e.g. Trichomonas, Trichonympha).
Subkingdom 2. Neozoa Cavalier-Smith 1993 stat. nov. 1997 em.
Infrakingdom 1. Sarcomastigota* Cavalier-Smith 1983 stat. nov. em. [diagnosis : typically sarcodines, flagellates, or
amoeboflagellates ; usually free-living heterotrophs, rarely chimaeric photophagotrophs with nucleomorphs and periplastid
membranes ; pseudopods varied, but seldom eruptive ; typically aerobes with tubular, less often flat or rarely discoid,
mitochondrial cristae and prominent Golgi dictyosomes ; cortical alveoli absent ; axopodia usually absent (if present with
quincunx arrangement) ; amitochondrial species lack hydrogenosomes and well-developed dictyosomes].
Phylum 1. Neomonada* Cavalier-Smith 1997.
Subphylum 1. Apusozoa Cavalier-Smith 1997 (e.g. Apusomonas, Ancyromonas, Jakoba, Reclinomonas Ebria).
Subphylum 2. Isomita Cavalier-Smith 1997 (e.g. Phalansterium, Cyathobodo, Kathablepharis).
Subphylum 3. Choanozoa* Cavalier-Smith 1981 em. 1983 stat. nov. (e.g. Monosiga, Diaphanoeca, Corallochytrium,
Psorospermium).
Phylum 2. Cercozoa phyl. nov. (syn. Rhizopoda Von Siebold 1845 stat. nov. Haeckel 1886 as emended by Cavalier-Smith
1995 b, 1997 a ; present phylum emended by addition of Spongomonadida and transfer of Cristidiscoidia to Choanozoa :
Cavalier-Smith, 1997 b.) (diagnosis : unicellular phagotrophic heterotrophs or else photosynthetic algae with green chloroplasts
and nucleomorphs within a periplastid membrane located inside a fourth smooth membrane ; typically free-living aerobes
having peroxisomes and mitochondria with tubular (or very rarely flat or vesicular) cristae ; flagellates with two usually
anisokont cilia or a single cilium or non-flagellates (usually rhizopods) with a test and}or filose or reticulose pseudopodia or
with a green plastid and nucleomorph ; cilia without lateral flanges, paraxial rods, transition helix or tubular hairs ; cortical
alveoli and axopodia absent ; heterotrophs have a flexible cell surface without a rigid dense protein layer inside or outside the
plasma membrane ; distinct cytopharynx absent ; silica scales sometimes present but internal silica skeleton absent ; extrusomes,
if present, isodiammetric or a complex Stachel ; often with walled cysts).
Subphylum 1. Phytomyxa Cavalier-Smith 1997 (e.g. Plasmodiophora).
Subphylum 2. Reticulofilosa Cavalier-Smith 1997 (e.g. Chlorarachnion, Cryptochlora).
Subphylum 3. Monadofilosa Cavalier-Smith 1997 (e.g. Cercomonas, Gymnophrys, Euglypha, Spongomonas).
Phylum 3. Foraminifera (DOrbigny 1826) Eichwald 1830 stat. nov. Margulis 1974 (e.g. Allogromia, Ammonia).
Phylum 4. Amoebozoa Lu$ he 1913 stat. nov. em. [emended diagnosis : solitary or aggregative amoebae with usually noneruptive lobose pseudopods, not uncommonly with finer pointed subpseudopodia ; mitochondria with tubular cristae, or
sometimes absent ; Golgi dictyosomes well-developed except in Archamoebae ; sometimes with transient or permanent cilia ;
usually one cilium (rarely two) per kinetid and one kinetid (rarely many) per cell].
Subphylum 1. Lobosa Carpenter 1861 stat. nov. Cavalier-Smith 1997 em. (e.g. Amoeba, Acanthamoeba, Arcella, Difflugia,
Multicilia).
Subphylum 2. Conosa subphyl. nov. (diagnosis : kinetid, when present, with a cone of microtubules which typically
subtends the nucleus at its broader end, and with a lateral microtubular ribbon closer to the cell surface ; aggregative aerobes
with mitochondria or solitary anaerobes lacking mitochondria and peroxisomes).
Infraphylum 1. Archamoebae Cavalier-Smith 1983 stat. nov. (e.g. Pelomyxa, Mastigamoeba, Phreatamoeba, Entamoeba).
Infraphylum 2. Mycetozoa De Bary 1859 stat. nov.
Superclass 1. Eumyxa* Cavalier-Smith 1993 stat. nov. (e.g. Protostelium, Physarum).
Superclass 2. Dictyostelia Lister 1909 stat. nov. (e.g. Dictyostelium).
Infrakingdom 2. Discicristata infraking. nov. (diagnosis : mitochondrial cristae typically discoid ; pseudopods if present
are eruptive lobes ; kinetid typically with two centrioles, rarely four or absent).
Phylum 1. Percolozoa Cavalier-Smith 1991.
Subphylum 1. Tetramitia Cavalier-Smith 1993 (e.g. Percolomonas, Naegleria).
Subphylum 2. Pseudociliata Cavalier-Smith 1993 (Stephanopogon).
Phylum 2. Euglenozoa Cavalier-Smith 1981.
Subphylum 1. Plicostoma subphyl. nov. (diagnosis : with pellicular strips and}or a feeding apparatus comprising two or
three supporting rods and four or five curved or plicate vanes. (This formalizes the name plicostome, first used by Patterson
(1988) for this taxon, even though it is not ideal as the vanes are often not plicate and are absent from petalomonads)
[superclasses Diplonemia Cavalier-Smith 1993 stat. nov. (pellicular strips absent, e.g. Diplonema) and Euglenoida Bu$ tschli 1884
stat. nov. Cavalier-Smith 1997 (pellicular strips present, e.g. Euglena, Petalomonas, Peranema)].
Subphylum 2. Saccostoma subphyl. nov. (diagnosis : subapical cytostome and cytophyarynx reinforced on one side by
microtubules (the so-called microtubular root (MTR)}pocket feeding appararatus : hence the name from latin saccus, bag and
stoma mouth), lacking dense reinforcing rods or vanes ; without pellicular strips). Classes Kinetoplastea Honigberg 1963 stat.
nov. Margulis 1974 (e.g. Bodo, Trypanosoma, Leishmania) and Postgaardea cl. nov. (diagnosis : mitochondria without kinetoplast or
cristae : sole genus Postgaardi : Simpson et al., 1997).

A revised six-kingdom system of life

233

Table 3. (cont.)
Infrakingdom 3. Alveolata Cavalier-Smith 1991 em.
Superphylum 1. Miozoa Cavalier-Smith 1987.
Phylum 1. Dinozoa Cavalier-Smith 1981 em.
Subphylum 1. Protalveolata* Cavalier-Smith 1991 em. (e.g. Colponema, Ellobiopsis, Spironema, Hemimastix, Colpodella,
Perkinsus).
Subphylum 2. Dinoflagellata Bu$ tschli 1885 stat. nov. Cavalier-Smith 1991 (e.g. Noctiluca, Crypthecodinium,
Amphidinium).
Phylum 2. Sporozoa Leuckart 1879 stat. nov. Cavalier-Smith 1981 (syns Telosporidia Schaudinn 1900 ; Euspora Levine
1961 ; Polannulifera Levine 1969 ; Apicomplexa Levine 1970 pro parte).
Subphylum 1. Gregarinae Haeckel 1866 stat. nov. (e.g. Monocystis).
Subphylum 2. Coccidiomorpha Doflein 1901 [Superclasses Coccidia Leuckart 1879 stat. nov. Cavalier-Smith 1993
(e.g. Sarcocystis, Toxoplasma, Eimeria), Ascetospora Sprague 1979 stat. nov. (e.g. Haplosporidium, Paramyxa) and Hematozoa Vivier
1982 stat. nov. (e.g. Plasmodium, Babesia)].
Subphylum 3. Manubrispora subphyl. nov. (Diagnosis : plasmodial intracellular parasites of gregarines with inner
membrane complex in vegetative cells, but no apical complex, mitochondria, peroxisomes, centrioles or cilia ; uninucleate spores
usually spherical, with separate membrane-bounded polar body and manubrium}lamellar complex, a thin dense wall ; spores
formed within walled cyst by plasmotomy ; Golgi complex an aggregate of vesicles, associated with centrosomal plaque during
closed mitosis ; spindle intranuclear : sole class Metchnikovellea Weiser 1977 emend. Cavalier-Smith 1993 to exclude
Minisporea, e.g. Metchnikovella).
Superphylum 2. Heterokaryota Hickson 1903 stat. nov. Cavalier-Smith 1993.
Phylum Ciliophora Doflein 1901 stat. nov. Copeland 1956 em. auct.
Subphylum 1. Tubulicorticata de Puytorac et al. 1992 (e.g. Loxodes, Stylonychia, Colpoda).
Subphylum 2. Epiplasmata de Puytorac et al. 1992 (e.g. Tetrahymena, Paramecium, Vorticella).
Subphylum 3. Filocorticata de Puytorac et al. 1992 (e.g. Spathidium).
Infrakingdom 4. Actinopoda Calkins 1902 stat. nov. Cavalier-Smith 1997.
Phylum 1. Heliozoa Haeckel 1886 stat. nov. Margulis 1974 (e.g. Actinophrys, Acanthocystis).
Phylum 2. Radiozoa Cavalier-Smith 1987.
Subphylum 1. Spasmaria Cavalier-Smith 1993 (Sticholonche, acantharians, e.g. Acanthometra).
Subphylum 2. Radiolaria Mu$ ller 1858 emend. stat. nov. Cavalier-Smith 1993 (e.g. Thallassicolla, Aulacantha).
* Probably paraphyletic.
Almost certainly paraphyletic.
For a more detailed classification of protozoa to the level of subclass and discussion of its phylogenetic basis see CavalierSmith (1993 a, 1995 b, 1997 a, b) and Corliss (1994).
The name Archezoa (Cavalier-Smith 1983 a) is derived from the Greek arche meaning beginning or first (as is archetype),
not from archaios (meaning ancient as in archaeology) ; the spelling Archaezoa sometimes seen (e.g. Maynard Smith &
Szathmary, 1995) is incorrect. When I created the taxon I was perfectly aware that Perty (1852) had used Archezoa in a
broader sense ; but as nobody since Haeckel (1868 and subsequent editions) had used it at all, I considered that no confusion
would arise. In any case, it is often desirable to refine old names rather than to invent totally new ones ; any temporary confusion soon fades nobody now complains that the name Mollusca scarcely overlaps at all with Linnaeuss melange grouped
under that name, or that Insecta, Crustacea and Chordata no longer have the same circumscription as they once did, or that
Linnaeus included bats among the Primates, and Reptilia and many fish in his Amphibia, whilst his Reptilia included frogs
and tortoises but excluded snakes. If we followed those pedantic nomenclaturists who think that all changes in circumscription
of higher taxa necessarily require them to be renamed, we would have to change all these and many other established names
thus causing undue confusion. As in the present system, Haeckel defined Archezoa as the most primitive Protozoa : but he
thought that amoebae were the most primitive organisms and thus, unlike Cavalier-Smith (1983 a), excluded flagellates from
Archezoa. It is now clear that all amoebae are derived from flagellates, contrary to Haeckels view, so flagellates are more
primitive than amoebae, and all non flagellates except Dientamoeba are excluded from the present Archezoa.

long been known that those of Ctenophora are

distinctly tubular as are those of vertebrate adrenal
glands. Clearly, tubular cristae have evolved from
the ancestral animal condition of flat cristae several
times within the kingdom Animalia. The cristae of
myxosporidia are sometimes referred to as flat
(Corliss, 1984) and sometimes as tubular (Taylor,
1978), but they are usually not clearly either. They
are so different from the flat cristae of choano-

flagellates or the tubular cristae of most other

neozoan protozoa that it would be more accurate
simply to call them irregular. The origin of the
irregular cristae of myxozoa from the more typical
flat cristae of most lower invertebrates might have
been contemporaneous with the secondary origin of
their amoeboid plasmodial vegetative state ; according to the molecular coevolutionary theory of
the origin of mitochondrial cristae, a marked change

234

in cristal morphology is often a pleiotropic neutral

consequence of adaptive changes in the cell surface
(Cavalier-Smith, 1997 a).
The inclusion of Archezoa and the exclusion of
Myxozoa and Mesozoa from the kingdom Protozoa
means that the kingdom now includes all primitively
unicellular
phagotrophic
non-photosynthetic
eukaryotes that have not evolved by the secondary
loss of chloroplasts, and is predominantly made up of
such organisms. It is thus a relatively homogeneous
grade of organisation that will continue to be a very
useful major unit of classification. The fact that it is
obviously paraphyletic, in contrast to the holophyletic kingdom Animalia, does not lessen its utility.
The division of the phagotrophic zoological world
into two distinct kingdoms, the unicellular Protozoa
and the ancestrally triploblastic Animalia, appropriately recognizes the fundamental differences in
body plan between the two zoological kingdoms.
The higher level classification of each is summarized
in Tables 3 and 4. Whether one adopts some variant
of the five-kingdom system (Whittaker, 1969) or the
six-kingdom system (Cavalier-Smith, 1981 a, 1983 a),
the exclusion of Protozoa from the kingdom Animalia means that Haeckels term Metazoa is now an
unnecessary synonym of Animalia. It is desirable to
harmonize the circumscription of the vernacular
term animal with that of the formal term Animalia
and not to include protozoans within its ambit. Even
in the most elementary teaching, it is best not to refer
to protozoa as unicellular animals but to encourage
the view of protozoa as a different and more
primitive form of life than animals, plants or fungi.
Calling protozoa unicellular animals not only downplays the great significance of the differences in body
plan between the two kingdoms, but helps to
condemn the study of protozoa to a position of
secondary importance within zoology, which their
status as a separate kingdom belies. Zoology is the
study of animals and protozoa, each of which have
important but distinct roles in the biosphere. The
phrase unicellular animals is also particularly
inappropriate for the three groups of protozoan
algae : Euglenia, Chlorarachnea and photosynthetic
dinoflagellates. For the reasons discussed previously
(Cavalier-Smith, 1993 a, 1995 a), these three photosynthetic groups are too closely related to nonphotosynthetic protozoa to be excluded from the
kingdom. Since the term algae has long since ceased
to refer to a botanical taxon, there is no contradiction
in referring to these three groups as protozoan algae
or in suggesting that, like all other Protozoa, they
should be subject to the Zoological, not to the

T. Cavalier-Smith
Botanical Code of Nomenclature (Cavalier-Smith,
1981 a).
(4) Classification of the Neozoa
(a) The infrakingdom Sarcomastigota
Neozoan classification is still in a state of flux.
Ribosomal RNA sequence studies on a wide variety
of zooflagellates currently in progress in our laboratory indicate that the distinction between the
infrakingdoms Sarcodina and Neomonada is less
clear than it seemed earlier (Cavalier-Smith, 1997 a).
It therefore seems sensible to merge them into a
single infrakingdom, for which I adopt the name
Sarcomastigota, used earlier (Cavalier-Smith,
1983 a) for a similar assemblage of flagellates and
sarcodines. Unlike my earlier subkingdom Sarcomastigota (Cavalier-Smith, 1983 a), which was not
generally adopted, the present infrakingdom Sarcomastigota includes Choanozoa since the distinction
between tubular and flat cristae is less fundamental
than was earlier thought (for the reasons, see
Cavalier-Smith, 1997 a) ; it also excludes the Alveolata and Actinopoda, which are treated as separate
infrakingdoms. Here I follow Siddall, Stokes &
Burresson (1995) in placing Haplosporidia in the
Alveolata ; even though this assignment is most
uncertain (Cavalier-Smith, 1997 a), excluding them
from the infrakingdom Sarcomastigota makes it
phenotypically more uniform.
In the present system Mycetozoa are no longer
treated as a separate phylum ; instead they are
placed within the phylum Amoebozoa, and grouped
with the Archamoebae as a new subphylum Conosa,
characterized in the flagellate members of the group
by a cone of microtubules emanating from the often
single centriole and subtending the nucleus. It
appears that rRNA trees usually place the Conosa
much too low ; protein trees place them much higher
up near the opisthokonta (animals, fungi, Choanozoa) and in the case of actin group them together.
Only two subphyla are recognised in the Amoebozoa : Conosa and Lobosa. The Lobosa comprise the
three classes Amoebaea, Testacealobosea and Holomastigea (with sole genus Multicilia, which is now
placed in the Lobosa as it has cell surface glycostyles
similar to those of some Amoebaea : Mikryukov &
Mylnikov, 1996 a ). Following these simplifications
there are now only four sarcomastigote phyla : two
essentially sarcodine (Foraminifera, Amoebozoa)
and two predominantly flagellate but with a significant sprinkling of rhizopods or other non-flagellates
(Neomonada and Cercozoa).

A revised six-kingdom system of life

235

Table 4. Classification of the kingdom Animalia and its 23 phyla

Subkingdom 1. Radiata Linnaeus 1758 em. stat. nov. Cavalier-Smith 1983 (multicllellular ; radial or biradial
symmetry ; no anus).
Infrakingdom 1. Spongiaria* De Blainville 1816.
Phylum Porifera Grant 1836 (sponges).
Subphylum 1. Hyalospongiae Vosmaer 1886 stat. nov. em. [typically with silicious spicules ; sometimes also
calcified (i.e. pharetronids ; sphinctozoans are probably secondarily without spicules) but without calcareous
spicules ; mainly demosponges, e.g. Axinella, and hexactinellids].
Subphylum 2. Calcispongiae Blainville 1834 stat. nov. (syn. Calcarea Bowerbank 1864 e.g. Clathrina, Sycon).
Subphylum 3. Archaeocyatha Vologdin 1937 (extinct).
Infrakingdom 2. Coelenterata* Leuckart 1847 em. auct.
Phylum 1. Cnidaria* (corals, sea anemones ; jellyfish, hydroids).
Subphylum 1. Anthozoa* Ehrenberg 1831 stat. nov. (e.g. Sarcophyton, Actinia).
Subphylum 2. Medusozoa Petersen 1979 (e.g. Hydra, Physalia, TripedaU lia, Aurelia).
Phylum 2. Ctenophora Eschscholtz 1829 (comb jellies, e.g. Beroe, Pleurobrachia).
Infrakingdom 3. Placozoa infrak. nov. (without body cavity, gut or nervous system).
Phylum Placozoa Grell 1971 (Trichoplax).
Subkingdom 2. Myxozoa Grasse! 1970 stat. nov. Cavalier-Smith 1996 (unicellular non-ciliate parasites with
multicellular spores).
Phylum Myxosporidia Bu$ tschli 1881 stat. nov. Grasse! 1970 (e.g. Myxidium).
Subkingdom 3. Bilateria* Hatschek 1888 stat. nov. Cavalier-Smith 1983 (bilateral animals, primitively with an
anus and probably coelom).
Branch 1. Protostomia* Grobben 1908.
Infrakingdom 1. Lophozoa* infrak. nov. (diagnosis : primitively sessile with U-shaped gut and ciliated oral
tentacles with coelomic extensions ; early ciliated larvae trochophores, later often bivalved).
Superphylum 1. Polyzoa Thompson 1830 stat. nov. em. (diagnosis : no vascular system or longitudinal nerve
cords ; adult without shell ; zooids able to multiply by vegetative budding, often colonially ; larval brain disappears
at metamorphosis).
Phylum 1. Bryozoa* Ehrenberg 1831 (unnecessary junior synonym : Ectoprocta Nitsche 1870).
Subphylum 1. Gymnolaemata Allman 1856 stat. nov. (e.g. Bugula, Flustra, Membranipora).
Subphylum 2. Lophopoda Dumortier 1835 (syn. Phylactolaemata Allman 1856, e.g. Plumatella).
Phylum 2. Kamptozoa Cori 1929.
Subphylum 1. Entoprocta* Nitsche 1870 (e.g. Loxosoma, Pedicellina, Urnatella).
Subphylum 2. Cycliophora Funch & Kristensen 1995 stat. nov. (Symbion).
Superphylum 2. Conchozoa superphyl. nov. (diagnosis : vascular system ; ancestrally with calcareous shell,
primitively bivalved and unhinged).
Phylum 1. Mollusca Linnaeus 1758 em. Lamarck.
Subphylum 1. Bivalvia* Linnaeus 1758 stat. nov. em. auct. (bivalves, e.g. Mytilus, Pecten : unneccessary
junior synonyms Acephala Cuvier ; Lipocephala Lankester 1889).
Subphylum 2. Glossophora Lankester 1889 stat. nov. (with radula and head).
Infraphylum 1. Univalvia Linnaeus 1758 stat. nov. em. (diagnosis : non-chambered shell in one piece ;
tentacles ; crystalline style : monoplacophorans, gastropods, scaphopods).
Infraphylum 2. Spiculata infraphyl. nov. (diagnosis : calcareous spicules and}or multiple shell plates ;
without eyes, tentacles or crystalline style : aplacophorans, caudofoveates, polyplacophorans).
Infraphylum 3. Cephalopoda Cuvier 1797 (single multichambered shell, e.g. octopus, squid, nautilus,
cuttlefish).
Phylum 2. Brachiozoa phyl. nov. (diagnosis : with lophophore and vascular system).
Subphylum 1. Brachiopoda* Dume! ril 1806 (e.g. Lingula, Terebratula).
Subphylum 2. Phoronida Hatschek 1888 (Phoronis, Phoronopsis).
Superphylum 3. Sipuncula superphyl. nov. (U-shaped gut, ventral non-ganglionated nerve cord ; no vascular
system or shell ; pelagosphaera larva).
Phylum Sipuncula Raffinesque 1814 stat. nov. Sedgwick 1898 (e.g. Golfingia, Phascolosoma).
Superphylum 4. Vermizoa superphyl. nov. (diagnosis : coelomate worms with blood and straight gut with
anus ; ciliated larvae without bivalved shells ; two ventrolateral or one primitively paired ventral nerve cord).
[continued overleaf

236

T. Cavalier-Smith

Table 4. (cont.)
Phylum 1. Annelida Lamarck 1809 (unnecessary junior synonym Chaetopoda : ancestrally segmented with
coelom around gut ; chaetae ; ganglionated ventral nerve cord).
Subphylum 1. Polychaeta* Grube 1850.
Infraphylum 1. Operculata infraphyl. nov. (diagnosis : live in calcareous tubes with operculum ;
lacking muscular pharynx, e.g. Spirorbis, Serpula).
Infraphylum 2. Pharyngata infraphyl. nov. (diagnosis : with muscular pharynx ; if tube-dwellers
without calcareous operculum : bristleworms, e.g. Nereis, Arenicola, Sabella) .
Subphylum 2. Clitellata auct. (earthworms, leeches).
Subphylum 3. Echiura Sedgwick 1898 orth. em. Stephen 1965 stat. nov. (e.g. Urechis, Bonellia).
Subphylum 4. Pogonophora Johannson 1937 stat. nov. (Frenulata and Vestimentifera, e.g. Riftia).
Phylum 2. Nemertina Oersted 1844 em. Schultze 1850 (unnecessary syn. Rhynchocoela Schulze 1850)
(unsegmented coelom around only eversible proboscis : proboscis worms, e.g. Nemertes).
Infrakingdom 2. Chaetognathi Leuckart 1854 stat. nov.
Phylum Chaetognatha Hyman 1959 (arrowworms, e.g. Sagitta, Spadella).
Infrakingdom 3. Ecdysozoa infrak. nov. (diagnosis : with thick cuticle that is moulted ; no surface cilia or
ciliated larvae ; gut, if present, with anus ; coelom present only embryonically or absent ; ventral nerve cord ; usually
gonochoristic). Name suggested for this clade minus Loricifera by Aguinaldo et al. 1997.
Superphylum 1. Haemopoda superp. nov. (diagnosis : body segmented, with limbs on several segments ; adult
body cavity a haemocoel that extends into the limbs).
Phylum 1. Arthropoda von Siebold and Stannius 1848 (hard cuticle ; jointed limbs moved by muscles).
Subphylum 1. Cheliceromorpha Boudraux 1978.
Infraphylum 1. Pycnogonida Latreille 1810 (sea spiders , e.g. Nymphon).
Infraphylum 2. Chelicerata Heymons 1901 (arachnids, eurypterids, king crabs).
Subphylum 2. Trilobitomorpha Strmer 1944 stat. nov. (trilobites ; trilobitoids).
Subphylum 3. Mandibulata Snodgrass 1938.
Infraphylum 1. Crustacea* Pennant 1777 (e.g. barnacles, crabs, shrimps, copepods, ostracods).
Infraphylum 2. Myriapoda Leach 1814 (centipedes, millipedes, symphylans, pauropods).
Infraphylum 3. Insecta Linnaeus 1758 em. auct. (Unnecessary junior synonym Hexapoda : e.g.
springtails, silverfish, locusts, bees, termites, beetles, moths, Drosophila).
Phylum 2. Lobopoda phyl. nov. (soft cuticle ; unjointed limbs with terminal claws ; both muscles and
hydraulic pressure involved in locomotion).
Subphylum 1. Onychophora Grube 1853 (e.g. Peripatus).
Subphylum 2. Tardigrada Doye' re 1840 stat. nov. (water bears, e.g. Echiniscus).
Superphylum 2. Nemathelminthes superphyl. nov. (diagnosis : unsegmented worms with spiny mouthparts ;
vascular system and limbs absent ; coelom bounded by epithelium absent).
Phylum Nemathelminthes Gegenbaur 1859 em. (diagnosis as for the superphylum).
Subphylum 1. Scalidorhyncha subphyl. nov. (diagnosis : retractile head covered with circlets of spined
scalids).
Infraphylum 1. Priapozoa infraphyl. nov. (diagnosis : unsegmented ; larva or adult with cuticular
lorica of longitudinal plates into which it can retract : classes Priapula, Loricifera).
Infraphylum 2. Kinorhyncha Reinhard 1887 (superficially segmented ; without lorica).
Subphylum 2. Nematoida Rudolphi 1808 (as Nematoidea) orth. em. stat. nov.
Infraphylum 1. Nematoda Gegenbaur 1859 orthog. em. stat. nov. (e.g. Caenorhabditis, Ascaris).
Infraphylum 2. Nematomorpha Vejovsky 1886 stat. nov. (e.g. Gordius, Nectonema).
Infrakingdom 4. Platyzoa infraking. nov. [diagnosis : ciliated non-segmented acoelomates or pseudocoelomates
lacking vascular system ; gut (when present) straight, with or without anus] ; possibly neotenously derived from
loxosomatid-like entoproct larvae.
Phylum 1. Acanthognatha phyl. nov. (diagnosis : chitinous cephalic bristles or jaws ; gut if present with
anus or anal pore).
Subphylum 1. Trochata subphyl. nov. (diagnosis : syncytial epidermis with radial tubules penetrating
the cuticle ; often with paired lemnisci in neck region ; epidermis multiciliated and}or non-ciliated ; proboscis often
functions as introvert ; protonephridia flame bulbs ; pseudocoelomate ; gonochoristic).
Infraphylum 1. Rotifera* Cuvier 1798 stat. nov. (e.g. Collotheca, Asplancha).
Infraphylum 2. Acanthocephala Rudolphi 1809 stat. nov. (e.g. Moniliformis).
Subphylum 2. Monokonta subphyl. nov. (without lemnisci or eversible proboscis ; monociliated
epidermal cells ; protonephridia solenocytes ; acoelomate ; hermaphrodite : classes Gastrotricha, Gnathostomulida).

A revised six-kingdom system of life

237

Table 4. (cont.)
Phylum 2. Platyhelminthes Gegenbaur 1859 em. Minot 1876.
Subphylum 1. Turbellaria Ehrenberg 1831 em. Ehlers 1864 (usually multiciliated, non-syncytial
epidermis retained in adult).
Infraphylum 1. Mucorhabda* infraphyl. nov. (mucoid non-lamellate rhabdoids ; protonephridia
absent or with two cilia, e.g. catenulids, acoels).
Infraphylum 2. Rhabditophora Ehlers 1985 (unranked) stat. nov. (lamellate rhabdites ;
protonephridia with many cilia : macrostomids, polyclads, Neoophora).
Subphylum 2. Neodermata Ehlers 1985 (unranked) stat. nov. (larval epidermis shed ; replaced by
syncytial neodermis).
Infraphylum 1. Trematoda Rudolphi 1808 stat. nov.
Infraphylum 2. Cercomeromorpha Bychowsky 1937 stat. nov. (monogeneans, tapeworms).
Branch 2. Deuterostomia Grobben 1908.
Infrakingdom 1. Coelomopora Marcus 1958 (as superphylum) stat. nov.
Phylum 1. Hemichordata Bateson 1885 stat. nov. orthog. em. auct.
Subphylum 1. Pterobranchia Lankester 1878 stat. nov. (e.g. Cephalodiscus ; graptolites).
Subphylum 2. Enteropneusta Gegenbaur 1870 stat. nov. (e.g. Balanoglossus).
Phylum 2. Echinodermata De Brugie' re 1789.
Subphylum 1. Homalozoa* Whitehouse 1941.
Subphylum 2. Pelmatozoa Leuckart 1848.
Infraphylum 1. Blastozoa Sprinkle 1973 (e.g. blastoids, cystoids).
Infraphylum 2. Crinozoa Matsumoto 1929 (sea lilies, feather stars).
Subphylum 3. Eleutherozoa Bell 1891.
Infraphylum 1. Asterozoa Von Zittel 1895 (starfish, brittle stars, concentricycloids).
Infraphylum 2. Echinozoa Haeckel in Von Zittell 1895 (sea urchins ; sea cucumbers).
Infrakingdom 2. Chordonia Haeckel 1874 em. Hatschek 1888 stat. nov.
Phylum 1. Urochorda Lankester 1877 (urochordates).
Subphylum 1. Tunicata Lamarck 1816 (tunicates).
Infraphylum 1. Ascidiae Blainville 1824 (ascidians, sorberaceans).
Infraphylum 2. Thaliae auct. (salps).
Subphylum 2. Appendicularia Lahille 1890 stat. nov. (larvaceans).
Phylum 2. Chordata Bateson 1885 em.
Subphylum 1. Acraniata* Bleeker 1859.
Infraphylum 1. Cephalochordata Owen 1846 (amphioxus).
Infraphylum 2. Conodonta Eichenberg 1930 stat. nov. (extinct).
Subphylum 2. Vertebrata Cuvier 1812 (unnecessary synonyms : Craniota Haeckel, Craniata auct.).
Infraphylum 1. Agnatha auct. (lampreys, hagfishes).
Infraphylum 2. Gnathostomata auct. (jawed fish, tetrapods).
Subkingdom 4. Mesozoa subking. nov. (bilateral multicellular parasites with ciliated epithelium ; gut, nervous
and vascular systems absent).
Phylum Mesozoa van Beneden 1877 (dicyemids, orthonectids).
* Probably paraphyletic.
Almost certainly paraphyletic.
The present system primarily emphasises classical morphological and embryological data and phylogenetic ideas,
but has been revised to take into account recent molecular phylogenetic evidence (reviewed by Ueshima, 1995 ; see
also Cavalier-Smith et al., 1996 a ; Bridge et al., 1995 ; Winnepenninckx et al., 1995 ; Winnepenninckx, Backeljau and
De Wachter, 1995 ; Satoh, 1995 ; Mackey et al., 1996 ; Garey et al., 1996 a, b)

Cercozoa is a new name for the assemblage of

filose and reticulose amoebae and phylogenetically
related zooflagellates (including sarcomonads like
Cercomonas) for which the provisional name Rhizopoda was recently adopted (Cavalier-Smith

1995 a, b, 1997 a). Because rRNA sequence has shown

that flagellate and amoeboid taxa are phylogenetically intermingled, the names Sarcodina and
Rhizopoda are now both abandoned as formal
names for taxa. They will however remain useful as

T. Cavalier-Smith

238

non-phylogenetic designations of body form in

descriptive and ecological studies, like flagellate or
alga ; thus rhizopod can continue to be applied in
the traditional sense to any amoeba, irrespective of
its taxonomic affinity, to contrast it with a flagellate
or sporozoan. Neomonada is a rather broad paraphyletic assemblage with many phylogenetically
diverse lineages, some more closely related to one or
more of the four higher kingdoms of life than to each
other. As our knowledge of neozoan phylogeny (in
which neomonad diversification played a central
role) improves the classification and possibly also the
circumscription of the Neomonada will need to
change : see Cavalier-Smith (1997 b). Though Cercozoa form a major clade on rRNA trees (CavalierSmith & Chao, 1997 and unpublished), some of
them are not phenotypically very distinct from some
Neomonada, so the boundary between the two phyla
is not morphologically clear cut, and will probably
have to be reevaluated as our understanding of these
centrally important but much neglected taxa improves.
The class Athalamea was formerly grouped with
Foraminifera in the phylum Granuloreticulosa (Lee,
1990) or Reticulosa (Cavalier-Smith, 1993 a). Electron microscopy of Penardia (Mikrjukov & Mylnikov,
1995) and Gymnophrys (Mikryukov & Mylnikov,
1996 b) shows kinetocysts similar to those of some
Cercomonas and a Golgi associated with the nuclear
envelope as in several cercozoans and neomonads.
Unlike Foraminifera they have no tests, their
pseudopods do not show bidirectional streaming but
differ from cercomonad pseudopods primarily in
containing some microtubules, their trophic cells
have two subparallel centrioles with vestigial ciliary
stumps, and they are freshwater, not marine. In the
absence of any specific ultrastructural similarities
with Foraminifera, the taxon Granuloreticulosa or
Reticulosa appears unnatural. Because of their
similarities to cercozoan amoeboflagellates, Athalamea (sole order Biomyxida) are here transferred to
the subphylum Monadofilosa of the Cercozoa, even
though they have flat mitochondrial cristae whereas
most cercozoans possess tubular cristae. Foraminifera thus become a phylum in their own right,
characterized by granuloreticular pseudopods emanating from one or more pores in their tests and
tubular mitochondrial cristae ; Granuloreticulosa}
Reticulosa are discontinued.
(b) The infrakingdom Alveolata
The flagellates formerly grouped as the apicom-

plexan subphylum Apicomonada (Cavalier-Smith,

1993) are here transferred to the subphylum Protalveolata of the flagellate phylum Dinozoa, and the
name Apicomplexa is abandoned as an unnecessary
junior synonym of the traditional name Sporozoa
[Corliss (1971) trenchantly and sensibly criticized
the multidudinous unneccessary new names for the
Sporozoa sensu stricto]. Sporozoa in the present sense
are restricted to obligate non-flagellate parasites,
and unlike Apicomplexa sensu Cavalier-Smith
(1993 a) include three taxa that lack an apical
complex : Manubrispora (comprising the metchnikovellids, formerly regarded as microsporidians, but
here excluded from Microsporidia because they
appear to have a cortical alveolus like that of most
sporozoans), Paramyxea, Haplosporidia. Furthermore, as outlined elsewhere (Cavalier-Smith,
1998 a), the parasitic flagellate Perkinsus is probably
less closely related to Sporozoa than are some free
living flagellates without an apical complex ; it
appears that its similarities to Sporozoa, which were
the reason for grouping them together in the new
phylum Apicomplexa (Levine, 1970) are convergent. No molecular evidence is yet available as to the
correct phylogenetic position of Manubrispora and
Paramyxea, and that for the Haplosporidia is
ambiguous (Cavalier-Smith, 1997 a).

(c) The infrakingdom Actinopoda

Though the monophyly of Heliozoa is rather
dubious, it would be premature to abandon either
the taxon Actinopoda or the phylum Radiozoa. The
suggestion that Radiozoa are polyphyletic (Amaral
Zettler, Sogin & Caron, 1997) may be mistaken : the
failure of Acantharia and Radiolaria to group
together on their distance trees could simply be an
artefact of the especially high evolutionary rate of
rRNA in Radiolaria, as clearly indicated by their
long branches, which probably places them too low
in the tree, an artifact that is even more marked for
Mycetozoa and Microsporidia. My own unpublished
maximum likelihood analyses indicate that Radiolaria and Acantharia may actually be sister groups
and that Radiozoa is probably a valid phylum ; the
position of Radiolaria, however, is not robust and
with some taxon samples they have a tendency to
associate instead with other long branches, but not
always the same ones as observed by Amaral Zettler,
Sogin & Caron (1997).

A revised six-kingdom system of life

V. THE KINGDOM ANIMALIA AND ITS 23
PHYLA

The animal kingdom is here divided into only 23

phyla, mostly familiar but a few novel, grouped in
four morphologically very distinct subkingdoms
(Table 4).
(1) Radiata, the ancestral animal
subkingdom
I have used the more ancient term Radiata of
Lamarck and Cuvier rather than the more recent
Diploblastica (Haeckel, 1866) for the subkingdom
comprising the four most primitive animal phyla,
even though Radiata originally included also Echinodermata and often other taxa which are now
properly classified within Bilateria. The term Diploblastica is entirely inappropriate for this subkingdom
since the majority of Radiata are not diploblastic,
but triploblastic. Only the cnidarian class Hydrozoa
is truly diploblastic. The fact that the cnidarian
subphylum Anthozoa (i.e. the majority of Cnidaria)
is fundamentally triploblastic was emphasized by
Pantin (1960), while the mesogloea of Scyphozoa
also frequently contains cells as does that of
Ctenophora (Hyman, 1940). Sponges are fundamentally triploblastic, as their mesohyl contains
many cells. As sponges are almost certainly the first
animals and probably arose directly from choanoflagellate protozoa (Kent, 1881 ; Cavalier-Smith,
1981 a ; Wainright et al., 1993 ; Cavalier-Smith et al.,
1996 a), the kingdom Animalia is ancestrally triploblastic (Cavalier-Smith, 1983 a) and the diploblastic
condition of Hydrozoa alone is a derived simplification. Haeckels (1866) influential idea that triploblasty evolved from diploblasty is clearly wrong and,
to use Hymans (1959) trenchant phrase (when
referring to another phylogenetically incorrect concept of Gephyrea), must be obliterated from
zoology . Though the term diploblastic may properly be applied to Hydrozoa, radiates should no
longer be referred to as diploblasts (e.g. Christen et
al., 1991 ; Ueshima, 1995).
As the mesohyl of sponges is probably both
homologous with and ancestral to the mesogloea of
Cnidaria it would be sensible to call it mesogloea
also.
(2) The derived subkingdom Mesozoa
Since the kingdom Animalia almost certainly
evolved by a transition between a colonial choano-

239

flagellate and the first sponge, during which collagenous mesenchymal connective tissue and epithelia
first evolved, it is highly improbable that Mesozoa
are phyletic links between Protozoa and Animalia.
As discussed in section IV molecular evidence
indicates that mesozoans are derived from Bilateria
by the loss of nervous system and gut. Because of
these radical differences I continue to treat them as a
separate subkingdom, even though this makes
subkingdom Bilateria paraphyletic.
(3) The number of animal phyla
Hyman (1940) recognised only 21 animal phyla or
later 23 (1959). Many American college textbooks
now recognize about 35, and Mo$ hn (1984) had as
many as 38. This taxonomic inflation has mainly
come about not by the discovery of new groups or by
a clear demonstration that many of Hymans phyla
were polyphyletic and had to be split. Much of it
arose by the excessive splitting of the phyla Arthropoda and Aschelminthes of Hyman (1940) ; in the
case of the aschelminths every class has been treated
as a separate phylum, not because of convincing
evidence that they are phyletically unrelated but
merely because their phyletic relationships have
been unclear. In essence phylogenetic agnosticism,
rather than reasoned arguments, has led to an
unnecessary inflation in the number of phyla, which
obscures rather than clarifies their evolutionary
relationship and makes it more difficult for students
to appreciate the diversity of animals than would a
system with fewer phyla. In the present system I
have therefore taken the bull by the horns or
perhaps I should say the priapulid by the spines
and grouped the aschelminth classes into just two
phyla, each of which I think is monophyletic. As a
result of also adopting a broader definition of the
Annelida and merging brachiopods and phoronids
into a single new phylum, Brachiozoa, my present
system still has only 23 phyla, even though (unlike
Hyman) I accept the phylum Placozoa, treat
Urochorda as a separate phylum from Chordata,
and segregate tardigrades and onychophorans from
Arthropoda as the phylum Lobopoda. I do not
accept the recently discovered Cycliophora (Funch
& Kristensen, 1995) as a separate monogeneric
phylum ; Symbion is sufficiently similar in body plan
to Entoprocta to be grouped with them in an
emended phylum Kamptozoa, but in a separate
subphylum. Thirteen of the present animal phyla
are identical to those of Hyman, and seven are
identical with those of the 18 phylum system of

240

Lankester (1911), though I use the older name

Annelida in preference to his Chaetopoda.
(4) A broadened phylum Annelida
The phylum Annelida recognised here is that of
Sedgewick (1898) with the addition of the Pogonophora, which were then unknown. Because of their
anterior oligomerous coelom and the belief that they
lack chaetae, early workers on Pogonophora thought
they were unrelated to annelids and closer to
deuterostomes, and so deserved their own phylum
(Reisinger, 1938). Hyman (1959) accepted this, but
at that time it was still not known that early workers
were completely ignorant of the posterior segmented
chaetiferous portion of the animals body. Since this
was discovered it has been clear that Pogonophora
are not deuterostomes, but instead are related to
annelids. The closeness of this relationship has been
strongly confirmed by sequence data (Winnepenninckx, Backeljau & De Wachter, 1996). It is now
evident that Pogonophora are annelids that have
lost their gut as a result of the evolution of the
capacity to cultivate chemosynthetic bacteria within
their body and give up predatory feeding. They
probably arose from tubicolous polychaetes. Though
loss of the gut and evolution of the ability to cultivate
bacteria in a highly developed trophosome are
important innovations, Pogonophora retain such a
major part of the fundamental annelid body plan
(segmentation with chitinous chaetae, haemoglobincontaining vascular system and respiratory tentacles
like polychaetes) that there is no longer any
justification for treating them as a separate phylum,
still less in dividing them into two phyla as has been
done. Ranking Pogonophora as a subphylum (within
which typical pogomophorans and Vestimentifera
can be separate classes) is sufficient recognition of
their distinctiveness from Polychaeta and Clitellata,
both here ranked as subphyla of the Annelida.
Sedgewick (1898) regarded the class Echiuroidea
as obviously true Annelids . He was the first to
realize they were radically different from sipunculoids, with which they were previously grouped in
the unnatural phylum Gephyrea (which Hyman
rightly obliterated from zoology ) ; he first made
Sipunculoidea a separate phylum. Hyman (1940)
first treated Echiuroida as a separate phylum on the
grounds that they are unsegmented , whilst at the
same time admitted that they might alternatively be
appended to the Annelida, to which phylum they
are undoubtedly related (Hyman, 1959). Even
though Hyman never wrote a volume of her treatise

T. Cavalier-Smith
dealing with the echiuroids and annelids and
therefore gave no more detailed discussion than that,
her unsound decision has been widely followed by
textbooks. However, the fact that echiuroids have
well developed chaetae, typical trochosphere larvae,
and a vascular system and nephridia similar to those
of other annelids, leaves little doubt that they are
annelids in which the formerly discrete coelomic
cavities have become secondarily merged into one.
As Sedgewick pointed out, there are traces of
segmentation in young Echiurus. Loss of septa and
partial merger of coelomic cavities is very common
in polychaetes and virtually complete merger occurred in leeches (Clark 1964) ; the almost total
suppression of segmentation in adult echiuroids is
probably a locomotory adaptation to a relatively
sedentary burrowing life in which the proboscis took
on an even more important role than in analogous
burrowing polychaetes. These adaptations do not
constitute a fundamentally different body plan from
other annelids. Treatment of the small number of
Echiuroida as a subphylum of the Annelida is
greatly preferable to making them a phylum in their
own right, which obscures their true affinities, and
devalues the idea that separate animal phyla should
have fundamentally different body plans.
(5) The pseudocoelomate phyla
Nemathelminthes and Acanthognatha phyl.
nov.
The proper status of the pseudocoelomate aschelminths has always been problematic. But as, Lorenzen (1985) stressed, there are many synapomorphies
that can be used to establish real relationships
between some of the classes. Therefore it is highly
unsatisfactory to rank almost every class as a separate
phylum. But, like most zoologists, I do not think that
aschelminths as a whole are monophyletic. Instead
there seem to be two fundamentally different
phylogenetic lineages, which I treat as phyla, and a
larger number of goupings of related classes, which I
treat as subphyla and infraphyla. Some of these are
more strongly supported by existing data than
others.
The new subphylum Trochata (comprising rotifers and Acanthocephala) is very strongly supported
by the synapomorphies listed in Table 2 [see also
Lorenzen (1985) and Nielsen (1995)] as well as by
molecular trees (Winnepenninkx et al., 1995 a ; Garey
et al. 1996 b). Ranking Acanthocephala as a phylum
(Hyman, 1940) overemphasized the importance of
the secondary loss of the gut and cilia ; these

A revised six-kingdom system of life

degeneracies, the syncytial epidermis, and proboscis
hooks are adaptations to parasitism remarkably
convergent with those of cestodes, which are also not
placed in a separate phylum from their closest freeliving ciliated relatives, the Turbellaria. I rank both
Rotifera and Acanthocephala as infraphyla so that
Bdelloidea and Monogogonta can be classes, as
preferred by those who have treated Rotifera as a
phylum. Even though Acanthocephala probably
arose from rotifers by parasitic degeneration, I prefer
not to place them within Rotifera as suggested by
Garey et al. (1996 b), since the two groups are
phenotypically so different and it would be nomenclaturally confusing to call Acanthocephala rotifers
merely because they evolved from them.
The new subphylum Monokonta (comprising
gastrotrichs and gnathostomulids) is primarily based
on their monociliated epithelial cells (whence the
name) and needs testing by gene sequencing. The
grouping of Monokonta and Trochata as the new
phylum Acanthognatha also needs such testing, as it
is not clear whether the chitinous jaws of rotifers and
gnathostomulids are homologous or not.
The grouping of Nematoda and Nematomorpha
as subphylum Nematoida is an old idea, well
supported both by morphology (Lorenzen, 1985)
and gene sequences (Winnepenninkx et al., 1995 a).
The similarities between the lorica of longitudinal
plates in Loricifera and larval priapulids is the basis
for grouping them as the new infraphylum Priapozoa. Priapozoa are grouped with kinorhynchs as the
new subphylum Scalidorhyncha, since they all have
a retractile head covered with circlets of spined
scalids, which seems unlikely to be convergent.
Molecular testing is however needed. My grouping
of Scalidorhyncha and Nematoida as the Nemathelminthes (an old phylum name) might be
questioned on the ground that such a grouping was
not evident on a recent rRNA tree (Winnepenninkx
et al., 1995 a). However that tree does not specifically
relate either Nematoida or priapulids to any other
group, and thus lacks resolution as to their relationship and therefore does not clearly disprove the
monophyly of the Nemathelminthes postulated here.
(6) The new phyla Brachiozoa and Lobopoda
Recent rRNA phylogenies show that phoronids
probably arose from inarticulate brachiopods by the
loss of the shell (Cohen, Gawthrop & CavalierSmith, 1997). Since in other respects both groups
share a basically similar body plan, phoronids are to
brachiopods as slugs are to snails, and do not merit

241

their own phylum. Therefore I rank both groups as

subphyla within the new phylum Brachiozoa.
Though molecular data support the relationship to
arthropods of both Onychophora and Tardigrada
(Garey et al., 1996 a), I separate them from Arthropoda as a new phylum Lobopoda (a name proposed
informally by Manton : 1977) because unlike true
arthropods they do not have jointed limbs and a
rigid cuticle.

(7) Phylogenetic assumptions behind the

new bilaterian infrakingdoms and
superphyla
Even if we adopt a broad phylum Annelida and
recognize only two aschelminth phyla, there are still
17 separate bilaterian phyla, which fall naturally
into two branches : Protostomia (13 phyla) and
Deuterostomia (4 phyla). Deuterostomia are divisible into two long-standing groups, Coelomopora
(Hemichordata and Echinodermata) and Chordonia (Urochorda and Chordata), which I here rank
as infrakingdoms, though a rank of superphylum
would also be acceptable. For protostomes, however,
the number of phyla is so large that we need both
infrakingdoms and superphyla in order to group
them informatively. I divide them into four new
infrakingdoms. Ecdysozoa comprise the three phyla
(Arthropoda, Lobopoda, Nemathelminthes) with a
moultable cuticle and a very poorly developed or
absent coelom (Aguinaldo et al., 1997). As they are
undoubtedly more closely related to each other than
to the Nemathelminthes I group Lobopoda and
Arthropoda together as the new superphylum
Haemopoda, because of their shared segmented
body and haemocoel that extends into the limbs.
This superphylum enables one to indicate that
lobopods are closely related to, but not actually
arthropods. By adopting the name Lobopoda I do
not imply any connection with polychaetes, some of
which have been referred to as lobopodial (Sharov,
1966). I do not think there is any direct phylogenetic
connection whatever between Haemopoda and
Annelida ; haemopod limbs and polychaete parapodia have always seemed to me merely analogous,
not homologous as assumed by the annelid theory of
the origin of arthropods (Snodgrass, 1938 ; Sharov,
1966). The gulf between Ecdysozoa and Lophozoa,
the other major protostome infrakingdom, is immense and very difficult to cross in plausible
megaevolutionary steps, which is why I have given
both the high rank of infrakingdom.

242

Lophozoa, the largest protostome infrakingdom, comprise the six protostome phyla that
have well developed coelomic cavities plus the
Kamptozoa, which do not. Unlike most zoologists, I
do not accept Hymans (1951) view that the absence
of a coelom in entoprocts precludes a specific
relationship with Bryozoa. Like Nielsen (1995) I
think the two groups are related. I suspect that
Kamptozoa arose from an early bryozoan by the loss
of the coelom. Though I do not altogether rule out
the reverse possibility that the coelom first evolved in
Bryozoa from a non-coelomate entoproct-like ancestor, I prefer the view that Bilateria were ancestrally coelomate and that the coelom first arose
simultaneously with the bilaterian gut by partitioning the anthozoan coelenteron into two parts, as
proposed in my invited manuscript for the Systematics Association 1984 meeting on the relationships
of lower invertebrates that was rejected by the
editors because it was too controversial (Conway
Morris et al., 1985). My novel version of the
Archicoelomate theory (Ulrich, 1949) of the origin
and diversification of bilateral animals was clearly
unacceptable to zoologists raised on Hymans dogmatic view that all non-coelomates were ancestrally
so. My present grouping of both Bryozoa and
Kamptozoa in the superphylum Polyzoa may be
equally unacceptable to many. But I have yet to see
a more convincing explanation of the origin of
Kamptozoa than by coelomic reduction, possibly
from a lophopod-like bryozoan.
Like the resurrection of the Polyzoa, my grouping
of the Brachiozoa with the Mollusca as the superphylum Conchozoa may seem a retrograde step,
harking back to Cuviers inclusion of brachiopods in
the Mollusca. However, as I argued at the 1984
meeting, Hymans dogma that Polyzoa, Brachiozoa
and Bivalvia are decephalized is probably incorrect.
There is no phylogenetic evidence whatever that any
of these animals ever had heads. Hymans dogma
was expressed so forcefully as to inhibit dissenting
thought : lophophorates and deuterostomes as seen
today are all decephalized animals of sedentary or
sessile habits. It is inconceivable that such types
should originate the Bilateria. It appears self-evident
that only well-cephalized, active forms can originate
definitive bilateral symmetry . Her view was purely
intuitive, unsupported by any phylogenetic reasoning. The fact that the heads of radulate molluscs,
annelids, nemathelminths, arthropods and vertebrates are all so different from each other, and lack
any recognizable homology, argues strongly that
their common ancestor was not strongly cephalized

T. Cavalier-Smith
and that each of these groups became cephalized
independently. Given that both radiate phyla
(Cnidaria and Ctenophora) are uncephalized, there
is no phylogenetic reason to think that the ancestral
bilaterian was cephalized. I argue that all Lophozoa
are ancestrally non-cephalized and that the Bivalvia
among the molluscs and the tubicolous Operculata
among the annelids are closer to the ancestral state
than their more cephalized relatives. The bivalved
character of the polyzoan larvae, of brachiopods,
and bivalve molluscs suggests that the common
ancestor of Polyzoa and Conchozoa had a bivalved
larva and sedentary non-cephalized adults. On this
view the traditional creeping radulate archimollusc
(von Salvini-Plawen, 1990) was not the ancestor of
Mollusca as a whole, but just of the subphylum
Glossophora (i.e. the radulate molluscs).
The other novel assemblage, the superphylum
Vermizoa (a name I proposed at the 1984 meeting)
comprises Annelida and Nemertina, both of which
have well developed closed vascular systems. Though
it is sometimes denied that the rhynchocoel of the
nemertines is a true coelom, the presence of the
vascular system and gut with an anus makes them
much more similar to annelids than to the flatworms,
with which many authors have grouped them. Since
echiuroids testify to the likelihood that some annelids
can lose segmentation, whereas in others the coelom
can become occluded by parenchyma (Rieger,
1985), there are no strong morphological reasons
against grouping annelids and nemertines in the
same superphylum.
Ribosomal RNA trees show that all the lophozoan
phyla, except the Sipuncula, are so closely related to
each other that their branching order cannot be
readily determined. Because of their unique character Sipuncula are placed solitarily in their own
lophozoan phylum. The name Lophozoa was chosen
on the assumption that the ancestor had a retractile
pair of tentacles with an interior coelomic cavity.
This is generally agreed for the Bryozoa and
Brachiozoa, but the ctenidia of molluscs, the tentacles of Sipuncula, and the gills of operculate
polychaetes may, I suggest, also be homologues of
the lophophore. It is much more doubtful that the
nemertine proboscis has a similar origin, but the
possibility cannot be totally dismissed. If the entoprocts are derived from Bryozoa by coelomic
occlusion their tentacles also may be derived from
those of the lophophore.
The deep divergence between the Ecdysozoa and
the Lophozoa is supported by the rRNA trees, which
are also inconsistent with an annelid ancestry for

A revised six-kingdom system of life

arthropods and Cuviers old group Articulata, which
is clearly poylphyletic. Since arthropods and molluscs appear almost simultaneously at the beginning
of the early Cambrian bilaterian fossil record,
Ecdysozoa and Lophozoa probably diverged very
soon after the origin of the Bilateria. Any theory of
how they are related is necessarily highly speculative.
The speculation that I presented at the 1984
Systematics meeting, and which I still favour was
that haemopod limbs evolved from tentacles of an
early solitary polyzoan. I had in mind the solitary
loxosomatid entoprocts, which can locomote actively
on the tips of their tentacles (Hyman, 1951). The
pseudocoel of such an active entoproct could have
given rise directly to the haemocoel of the Haemopoda and the pseudocoel of Nemathelminthes. Such
an origin involves much less change than the view
that arthropods evolved from a coelomate legless
worm. Entoproct tentacles can already function
similarly to lobopodal limbs and are similarly
arranged in two lateral rows. Nemathelminthes
could have arisen from an early lobopod through the
loss of limbs after adopting a burrowing habit, like
snakes, apodans, and caecilians among vertebrates.
The third protostome infrakingdom is the noncoelomate Platyzoa, which contains only the phyla
Acanthognatha and Platyhelminthes. Both are ancestrally slender ciliated worms with no vascular
system, either acoelomate or pseudocoelomate.
Though Hyman (19401959) was utterly confident
that they represent the ancestral bilaterian condition, I am deeply sceptical of this, as I cannot
envisage how they might have evolved from a
radiate ancestor. It seems to me more likely that they
arose from neotenous entoproct larvae of the
creeping type found in Loxosoma : even Hyman (1951)
noted their remarkable similarity to a rotifer. If such
an origin is correct, then no separate coelomic loss
need be postulated. An alternative way of subdividing the Protostomia to that in Table 4 would be
to transfer Kamptozoa from the Lophozoa to the
Platyzoa. Though compatible with my phylogenetic
assumptions, I prefer to emphasize the life cycle and
trophic similarities of Polyzoa, rather than the loss of
the coelom. The rRNA trees are unfortunatly not
very helpful in understanding the affinities of the
Platyhelminthes ; their branches (like those of nematodes) are so long that one must suspect that longbranch artefacts may be placing them lower in the
tree than their correct position. When one has a
rapid, almost simultaneous radiation, as appears to
be the case, coupled with an elevated evolutionary
rate in some branches, such systematic biases may

243

easily swamp the true phylogenetic signal and yield

a positively incorrect answer.
I do not of course suggest that Ecdysozoa or
Platyzoa evolved directly from any extant polyzoan
group. The discovery of Symbion suggests that the
early radiation of Kamptozoa may have been much
more varied than we can now guess from the few
extant species. Unfortunately the fossil record of this
phylum of tiny animals is so poor that it will
contribute very little to testing my thesis that similar
organisms could have been ancestral to both Ecdysozoa and Platyzoa.
The fourth protostome infrakingdom is the
Chaetognathi, which are so different from all other
protostome animals that it does not seem possible to
group them with any other phylum. Their rRNA
sequences also leave them in an isolated position,
and give no support to Hymans view that they are
deuterostomes. The protostome infrakingdoms
might be further reduced to three by placing
chaetognaths in the Ecdysozoa, with which they
share some characters, but I prefer to emphasize
their uniqueness.
(8) New animal subphyla and infraphyla
Subphyla have long been used in the Arthropoda
and Chordata which have the largest number of
disparate classes. In other phyla I have made more
extensive use of them than is traditional, partly to
avoid unnecessary splitting into separate phyla and
partly to emphasize that some classes really are more
closely related to each other than to others. Most of
the subphylum groupings are well accepted and I
have often used traditional names ; but slight changes
in rank are frequent (downward from phylum and
upward from class or superclass). Only three
subphyla are new both in concept and name :
Scalidorhyncha (priapulids, loriciferans and kinorhynchs) ; Trochata (rotifers and acanthocephalans) ; Monokonta (gastrotrichs and gnathostomulids).
In phyla with a large number of classes I have
used infraphyla much more extensively than usual.
Again, some of these, like Cephalopoda, Agnatha
and Rotifera, are well known taxa given intermediate ranks in order to include more cladistic
information in the classification. Only six are new in
both concept and name : the radulate mollusc
infraphyla Monoconcha (those with a single shell)
and Spiculata (those with multiple shell plates
and}or calcareous spicules) ; the polychaete Operculata and Pharyngata ; Priapozoa (priapulids and

244

loriciferans) ; and Mucorhabda (turbellarians with

non-lamellate rhabdoids). Elementary texts would
benefit from using subphyla more extensively than
they do, but infraphyla may not be necessary for
them, though will be valuable in more specialized
works and databases.
VI. THE KINGDOM FUNGI AND ITS FOUR
PHYLA

(1) Circumscription of the Fungi

A kingdom Fungi entirely distinct from the kingdom
Plantae is now almost universally recognized (Whittaker, 1969 ; Carlile & Watkinson, 1994). Recently
this kingdom has sometimes been referred to instead
as Mycota, but this name has probably not been
validly published and is entirely unnecessary and
much less widely understood. It is increasingly clear
that Fungi are more closely related to animals than
to plants (Baldauf & Palmer, 1993 ; Wainright et al.,
1993) ; however, it is incorrect to speak of an
evolutionary link between fungi and animals
(Wainright et al., 1993) since the phylogenetic
linkage is indirect via the choanoflagellate protozoa.
As Cavalier-Smith (1987 b) first suggested, it seems
that Fungi evolved from choanoflagellate protozoan
ancestors as did animals, but independently.
Molecular evidence strongly supports the restriction of the kingdom Fungi to the taxa summarized in
Table 5, including (as discussed in section III above)
the unexpected inclusion of the Microsporidia. The
earlier circumscription that I advocated previously
excluding the Microsporidia (Cavalier-Smith,
1981 a, 1987 b) was supported by much rRNA
sequence evidence (Bruns, White & Taylor, 1991),
and accepted by the authoritative Dictionary of the
Fungi (Hawksworth et al., 1995). Several other taxa
have often been treated as Fungi, but are evolutionarily quite unrelated to them. Some recent mycology
texts (e.g. Alexopoulos, Mimms & Blackwell, 1996)
correctly exclude these elements from the kingdom
Fungi, whereas others (e.g. Carlile & Watkinson,
1994) still incorrectly place them in the same
kingdom.
The most important of the non-fungal groups
widely studied by mycologists and therefore formerly
classified as Fungi are the slime moulds, plasmodiophorids, oomycetes, hyphochytrids and thraustochytrids, which all properly belong either in the
kingdom Protozoa or in the third botanical kingdom,
Chromista, as is fully accepted by Hawksworth et al.
(1995). These organisms are not fungi but fungus-

T. Cavalier-Smith
mimics , protists that have convergently become
similar to fungi in one or more respects.
The protozoan groups (slime moulds and plasmodiophorids) remain vegetatively amoeboid but have
acquired aerial spores for dispersion convergently
with fungi. Slime moulds are polyphyletic, but the
majority of them form a monophyletic protozoan
infraphylum Mycetozoa (Table 2) (i.e. classes
Protostelea, Myxogastrea and Dictyostelea) ; though
the myxogastrean Physarum polycephalum and the
dictyostelid Dictyostelium discoideum most often do not
group together on 18S rRNA distance trees they
usually do so on 18S rRNA maximum-likelihood
trees (e.g. Cavalier-Smith, 1995 a, b), which appear
to be less misled by the very long Physarum branch.
Inspection of the alignment shows about seven
obvious molecular synapomorphies for the Mycetozoa, and Rusk, Spiegel & Lee (1994) have rRNA
sequence evidence that the protostelids are specifically related to dictyostelids and myxogastreans,
which is also supported by trees based on sequences
of the protein elongation factor EF 1 (Baldauf,
& Doolittle, 1997). Molecular evidence is not yet
available for the minor groups of non-mycetozoan
slime moulds, the acrasids, copromyxids and fonticulids. However, the morphology of their mitochondria and pseudopodia convincingly places
Acrasida within the class Heterolobosea (Page, 1987)
within the protozoan phylum Percolozoa (CavalierSmith, 1993 a, c). While their morphology is less
decisive, copromyxids are placed in the protozoan
class Lobosea of the phylum Amoebozoa, whilst
fonticulids are in the class Cristidiscoidea, now
within the subphylum Choanozoa of the protozoan
phylum Neomonada (Cavalier-Smith, 1998 a ; formerly Cristidiscoidea were in the Rhizopoda : Page,
1987 ; Cavalier-Smith, 1993 a, 1995 b, 1997 a) ; and
there is no reason whatever to group them with
Fungi. Plasmodiophorids have often been thought to
be related to Mycetozoa and therefore were treated
as fungi ; but molecular evidence clearly shows that
they are protozoa that belong in the Cercozoa
(formerly Rhizopoda : Cavalier-Smith & Chao,
1997).
In contrast to the foregoing protozoan taxa,
oomycetes, hyphochytrids and thraustochytrids all
belong in the kingdom Chromista (see section VIII
below). They are probably all secondarily nonphotosynthetic heterotrophs that have acquired
vegetative cell walls convergently with fungi.
The protozoan Corallochytrium limacisporum has
been erroneously treated as a fungus (RaghuKumar, 1987) ; however, rRNA phylogeny reveals

A revised six-kingdom system of life

that it acquired a cell wall and evolved osmotrophy
entirely independently of Fungi and heterotrophic
chromists, and is actually a protozoan related to
choanoflagellates (Cavalier-Smith & Allsopp, 1996).
Other Protozoa, notably parasitic Ichthyosporea
(now in subphylum Choanozoa : Cavalier-Smith,
1998 a) have also erroneously been treated as fungi in
the past, whereas some Fungi, notably Pneumocystis
(Edman et al., 1988), have mistakenly been thought
to be Protozoa.
(2) The trichomycete origin of microsporidia
The key step in the origin of microsporidia apart
from the losses of these two organelles was the origin
of intracellular parasitism and of the polar tube, the
organelle that uniquely distinguishes the phylum
from all others. The polar tube is formed within the
spore from membranes thought to be of Golgi
character, and is typically long and highly coiled. In
germinating spores it is extruded and its apex
attaches to host cells enabling the sporoplasm to
enter the host cell through the tube. Clearly the
origin of the polar tube and this unique method of
infection was both necessary and sufficient for the
origin of intracellular parasitism and the first
microsporidian. A necessary corollary of this type of
injection mechanism is the digestion of the spore wall
and the temporary nakedness of the infective stage.
Temporarily naked infective cells are common in
Chytridiomycetes, the most primitive fungal group.
Furthermore chitin walls have secondarily lost,
apparently independently in the vegetative cells of
the allomycete Coelomomyces and the trichomycete
Amoeboidium, which both parasitize animals. Thus
the secondary loss of chitin walls in fungal parasites
of animals is not unprecedented and could also have
occurred in the postulated fungal ancestor of
microsporidia and thus preadapted it to the origin of
the polar tube and an intracellular mode of life. The
intracellular stage of microsporidia is not always
entirely naked, but sometimes bears a surface coat,
which in Pleistophora is so thick as almost to resemble
a wall.
Elsewhere I propose that the microsporidian polar
tube may have evolved from apical spore bodies,
which are organelles of certain harpellalean
trichomycetes, probably extrusive, that are thought
to help the spore attach to the gut lining of their
arthropod host (Cavalier-Smith, 1998 b). The fact
that microsporidia mainly infect invertebrates, especially insects, and only very rarely infect protozoa,
and are unknown as parasites of Metamonada has

245

long caused some zoologists to be sceptical of the idea

that they are primitive eukaryotes. An origin from
one of the trichomycetes, which are all obligate
endoparasites of arthropods does not suffer from this
problem and makes sense ecologically as well as
phylogenetically and cytologically (Cavalier-Smith,
1998 b).
(3) Revision of higher fungal classification
The classification in Table 5 is based on that of
Cavalier-Smith (1987 b) but modified in several
important respects, including the subdivision of the
kingdom into the two subkingdoms Eomycota and
Neomycota. The new names of higher fungal taxa in
Cavalier-Smith (1987 b) were at the editors insistence not accompanied by proper diagnoses. This
is rectified in Table 5 which validates them and all
new fungal names introduced here for the first time.
Since many mycologists have in recent decades been
curiously reluctant to use higher taxa than orders I
have thought it desirable to make the present fungal
classification more comprehensive than for the other
five kingdoms. I therefore include superclasses,
classes, subclasses and superroders in order to bring
fungal megasystematics more into line with the
practice in other kingdoms.
The most important innovations are the creation
of the new hemiascomycete class Geomycetes and a
major revision of the phylum Archemycota, which
includes not only the fungi placed by Bruns, White
& Taylor (1991) and many other authors in the
nomenclaturally invalid phyla Chytridiomycota and
Zygomycota, but also the Laboulbeniales, which
were traditionally regarded as ascomycetes. I argued
previously (Cavalier-Smith, 1987 b) that Zygomycota do not differ sufficiently from Chytridiomycota to merit a separate phylum. I am unaware
of any mycologist who has seriously argued the case
for having two separate phyla : nonetheless this
taxonomic inflation has been widely adopted in the
past decade.
I suggest that traditional trichomycetes (Lichtwardt, 1986), which are all gut parasites of arthropods, are polyphyletic. Two orders (Eccrinales and
Amoebidiales) have Golgi dictyosomes, unlike all
other Zygomycotina, so I remove them from the
Trichomycetes and Zygomycotina as a new class
Enteromycetes. I had earlier (Cavalier-Smith,
1981 a) divided the traditional Chytridiomycotina
into two classes : Chytridiomycetes sensu stricto (with
Golgi dictyosomes) and Allomycetes (without dictyosomes), both of which I validate in Table 5. I group
Chytridiomycetes and Enteromycetes together to

246

T. Cavalier-Smith

Table 5. Classification of the kingdom Fungi, its four phyla and 20 classes
Subkingdom 1. Eomycota subking. nov. (hyphae, when present, usually lacking septa or rarely with imperforate
septa ; dikarya absent ; vegetative walls frequently absent in animal parasites ; mitochondria and peroxisomes often
absent, sometimes replaced by hydrogenosomes : cellulae non binucleatae ; septa usitate absens ; si praesens non
perforata).
Phylum 1. Archemycota phyl. nov. (diagnosis : mitochondria and peroxisomes usually present ; if absent then
possessing hydrogenosomes ; vegetative cell walls usually present : mitochondria aut hydrogenosomae praesentes.
Name introduced without Latin diagnosis by Cavalier-Smith, 1987 b).
Subphylum 1. Dictyomycotina subphyl. nov. (diagnosis : Golgi dictyosome of stacked cisternae : dictyosoma
praebens).
Class 1. Chytridiomycetes De Bary 1863 stat. nov. Sparrow 1958 em. Cavalier-Smith 1981 (emended
diagnosis : posteriorly ciliated zoospores : sporae ciliis posterioribus instructae).
Subclass 1. Rumpomycetidae subcl. nov. (diagnosis : zoospores with rumposome : zoospora rumposoma
instructa ; introduced as a class name without Latin diagnosis by Cavalier-Smith, 1987 b) (orders Chytridiales,
Monoblepharidales).
Subclass 2. Spizomycetidae subcl. nov. (diagnosis : rumposome absent : sine rumposoma ; introduced as
a class name without Latin diagnosis by Cavalier-Smith, 1987 b) (orders Spizellomycetales, Neocallimastigales).
Class 2. Enteromycetes cl. nov. (diagnosis : cilia absent ; parasites of arthropod gut : sine ciliis ; intestinum
arthropodis incolens) (orders Eccrinales, Amoebidiales).
Subphylum 2. Melanomycotina subphyl. nov. (diagnosis : Golgi cisternae unstacked ; melanin-pigmented resting
spores common ; mitochondria and peroxisomes present : dictyosoma non-praebens : cisternae disjunctae ; sporae
saepe fuscae).
Infraphylum 1. Allomycotina infraphy. nov. (diagnosis : with uniciliate zoospores. : zoospora cilio unico
instructa).
Class 1. Allomycetes cl. nov. (diagnosis : with uniciliate zoospores : zoospora cilio unico instructa ; name
introduced without Latin diagnosis by Cavalier-Smith, 1981 a) [orders Blastocladiales, Coelomomycetales ord. nov.
(diagnosis : trophic phase without walls : sine muris in statu pabulatorio ; sole family Coelomomycetaceae)].
Infraphylum 2. Zygomycotina infraphyl. nov. (diagnosis : cilia and zoospores absent : sine ciliis).
Superclass 1. Eozygomycetia supercl. nov. (diagnosis : without sporangiospores or aquatic spores ;
saprophytes or symbionts of vascular plants : sine sporangiosporis aut sporis aquaticis).
Class 1. Bolomycetes cl. nov. (diagnosis : saprophytes with 1112-singlet centriole ; single large propulsive
conidia borne on long unbranched conidiophores ; zygospores with adjacent beak-like projections ; not within a
sporocarp ; septate mycelia : conidophora instructa ; centriolum microtubulos continens) [Basidiobolales ord. nov.
(diagnosis as for Bolomycetes) (Basidiobolus)].
Class 2. Glomomycetes* cl. nov. (diagnosis : form vesicular arbuscular endomycorhizas with vascular
plants ; sclerotium-like sporocarps contain chlamydospores (Glomales) or laterally produced zygospores
(Endogonales) ; centrioles, conidia, aerial spores and stalked sporophores absent. Conidiophora et centriolum
absens ; in radices plantarum crescens ; sporae in sporocarpiis subterraniis).
Superclass 2. Neozygomycetia supercl. nov. (diagnosis : sporocarp and centrioles absent ; aerial or aquatic
aplanospores ; saprophytes or parasites of animals : sporangiospora aut sporae aquaticae instructa).
Class 1. Zygomycetes cl. nov. (diagnosis : zygospore wall modified from the gametangial wall, produced
between the gametangia ; passive aerial asexual spores borne on stalked sporophores ; usually saprophytes : murus
gametangii in murum zygosporae transiens ; sporae aeriae in sporophoro pedicellato).
Subclass 1. Mucoromycetidae Fries 1832 stat. nov. (diagnosis : coenocytic ; septa absent in young
mycelia ; with globose multispored sporangia and}or uni- or oligo- sporic sporangiola ; sporophore hypha-like)
[orders Mucorales : sporangium with columella ; zygospores lacking an investment of sterile hyphae ; Mortierellales
ord. nov. (diagnosis : sporangium without columella ; zygospores often invested by sterile hyphae ; chlamydospores :
sporangium sine columello ; hyphae steriles zygospora saepe investientes ; chlamydospora instructates)].
Subclass 2. Meromycetidae subcl. nov. [diagnosis : aerial spores formed in merosporangia or as
conidia ; globular sporangia absent ; mycelia septate with medial plug (Dimargaritales, Kickxellales) or not
(Piptocephalaceae, Cuninghamellales) ; sporophore often vesicular : sporae aeriae in merosporangiis aut conidiae :
sporangiae globulosae absens ; sporophorum saepe vesiculatum].
Class 2. Zoomycetes cl. nov. (diagnosis : parasites of animals or protozoa ; sporangiospores absent : in
animaliis aut protozois parasitici ; sporangiophora absens).
Subclass 1. Entomycetidae subcl. nov. (diagnosis : endoparasites of invertebrates or protozoa ; hyphae
syncytial or subdivided by imperforate septa into coenocytic segments ; spores propulsive conidia ; zygospores lateral

A revised six-kingdom system of life

247

Table 5. (cont.)
to gametangia : in animaliis aut protozois parasitici ; hyphae septatae aut non-septatae ; conidia se praecipitantes)
(orders Entomophorales, Zoopagales).
Subclass 2. Pedomycetidae subcl. nov. (diagnosis : symbionts of mandibulate arthropods, attached to
their cuticle by foot-like holdfasts ; imperforate plugged septa : in arthropodis mandibulatis parasiticae ; in cuticula
affixi haptero dilatato ; septa imperforata).
Superorder 1. Trichomycetalia superord. nov. (diagnosis : endocommensals within the gut ; spores
are trichospores, formed by budding both asexually and following conjugation or arthrospores formed by hyphal
fragmentation : intestinum incolens ; trichosporae aut arthrosporae instructae) (orders Harpellales and Asellariales).
Superorder 2. Pyxomycetalia superord. nov. (diagnosis : ectoparasites ; differentiated male and
female organs ; female organ contains endospores and develops from three cells enclosed by a multicellular
pseudoperithecium with an apical pore ; male organ an antheridium forming spermatia either endogenously or by
exogenous budding : ectoparasiticae ; pseudoperithecium endosporae continens ; antheridium spermatia continens)
(orders Laboulbeniales, e.g. Laboulbenia ; Pyxidiophorales).
Phylum 2. Microsporidia Balbiani 1882 stat. nov. Weiser 1977 (diagnosis : obligate intracellular parasites of
animals or rarely protozoa ; vegetative cell walls, mitochondria and peroxisomes absent ; spores with chitin walls and
an extrusive polar tube through which the sporoplasm enters the new host cell).
Class 1. Minisporea Cavalier-Smith 1993 (diagnosis : polar tube with honeycomb outer layer) (e.g.
Chytridiopsis, Buxtehudia, Hessea).
Class 2. Microsporea Levine & Corliss 1963 (diagnosis : polaroplast present : spores usually oval, rarely
rod-shaped or pyriform).
Subclass 1. Pleistophorea Cavalier-Smith 1993 (diagnosis : multiply by plasmotomy : one spore type :
Pleistophorida Stempell 1906, e.g. Pleistophora, Amblyospora, Glugea, Encephalitozoon).
Subclass 2. Disporea Cavalier-Smith 1993 (diagnosis : multiply by binary fission ; disporogenic, i.e. two
spore types, e.g. Nosema, Enterocytozoon, Spraguea, Caudospora).
Subkingdom 2. Neomycota subking. nov. (diagnosis : usually with a dikaryotic phase ; meiotic products form
basidiospores or endospores or divide mitotically once to form ascospores ; mitochondria and peroxisomes always
present : cellulae binucleatae plerumque praesentes ; endospora aut ascospora aut basidiospora instructa).
Phylum 1. Ascomycota Berkeley 1857 stat nov. (diagnosis : meiotic products or their daughters form endospores
by subdividing the cytoplasm by membrane, not by budding : sporae intracellulares).
Subphylum 1. Hemiascomycotina* Brefeldt 1891 stat. nov. Ainsworth 1966 (diagnosis : ascocarp absent ;
usually yeast-like : sine ascocarpo ; plerumque in forma fermenti).
Class 1. Taphrinomycetes cl. nov. (diagnosis : cell walls often lack chitin ; meiotic products yield four
endospores : muri saepe sine chitino endosporae quattuor ; name introduced without Latin diagnosis by CavalierSmith, 1987 b) (e.g. Taphrina, Schizosaccharomyces, Protomyces, Pneumocystis).
Class 2. Geomycetes cl. nov. (diagnosis : with intracellular cyanobacterial endosymbionts (Nostoc sp.) ;
hyphae coenocytic when young ; huge pale asexual spores form at hyphal tips ; sex unknown : hyphae juvenes sine
septis ; cyanobacteria intra hyphis incolentes) [Geosiphonales ord. nov. (diagnosis as for Geomycetes) and sole
family Geosiphonaceae (Geosiphon)].
Class 3. Endomycetes cl. nov. (diagnosis : budding yeasts ; chitin only in bud scars ; ascogenous hyphae
absent ; meiotic products four endospores formed by fusion of smooth cytoplasmic membranes around each nucleus
separately ; ascogenous hyphae and dikaryotic phase absent : fermenti gemmipares ; chitinum in muro inter cellulis
filiis, non in muri alteri ; endosporae quattuor ; validates the name introduced without Latin diagnosis by Von Arx,
1967).
Subclass 1. Dipomycetidae subcl. nov. (diagnosis) (e.g. Dipodascopsis).
Subclass 2. Saccharomycetidae de Bary 1866 stat. nov. (e.g. Saccharomyces, Candida, Endomyces).
Subphylum 2. Euascomycotina Engler 1897 stat. nov. Ainsworth 1966. (diagnosis : filamentous ; chitin
throughout the cell walls ; ascus vesicle surrounds all four meiotic products ; ascogenous hyphae with brief
dikaryophase ; ascocarp present in sexual forms : chitinum in muris omnibus ; ascocarpa plerumque instructae).
Class 1. Discomycetes Fries 1836 stat. nov. Ainsworth 1966 (ascocarp an apothecium).
Subclass 1. Calycomycetidae subcl. nov. (diagnosis) (mostly mazaedial lichens, e.g. Calicium,
Coniocybe).
Subclass 2. Lecomycetidae subcl. nov. (diagnosis : asci inoperculate : asci sine operculis) (most lichen
fungi, e.g. Usnea, Lecanora, Peltigera, Xanthoria).
Subclass 3. Pezomycetidae subcl. nov. (diagnosis : asci operculate : asci operculati) (e.g. Peziza, Tuber,
Morchella, Rhytisma).
[continued overleaf

248

T. Cavalier-Smith

Table 5. (cont.)
Class 2. Pyrenomycetes Fries 1821 stat. nov. Ainsworth 1966 (ascocarp a perithecium ; mycelia septate
when young).
Subclass 1. Verrucomycetidae subcl. nov. (diagnosis : pyrenocarpous lichen fungi : fungi lichenis
perithecia habentes) (e.g. Verrucaria).
Subclass 2. Ostiomycetidae subcl. nov. (diagnosis : non lichen fungi : non lichenes) (e.g. Neurospora,
Sordaria, Claviceps, Nectria, Xylaria, Daldinia, some fungi imperfecti).
Class 3. Loculomycetes cl. nov. (diagnosis : ascocarp not apothecial : ascocarpae non apotheciae).
Subclass 1. Dendromycetidae subcl. nov. (diagnosis : lichenized) (e.g. Arthonia, Lecanactis).
Subclass 2. Loculoascomycetidae Luttrell 1955 (ascocarps pseudothecial within a stroma ; asci
bitunicate) (e.g. Dothidea, Pleospora, Venturia).
Class 4. Plectomycetes Gwynne-Vaughan 1922 stat. nov. Ainsworth 1966 (ascocarp a cleistothecium ;
asci unitunicate) (e.g. Penecillium, Aspergillus, Eurotium, Erysiphe ; most fungi imperfecti).
Phylum 2. Basidiomycota de Bary 1866 em. auct. stat. nov. Moore 1980. [meiotic products form four
exospores (ancestrally ballistospores) by budding from the surface of the cell ; meiotic endospores absent ; dikarya
with clamp connections)].
Subphylum 1. Septomycotina* subphyl. nov. (diagnosis : uniporate septa without dolipores ; basidia divided
by transverse walls, but without long epibasidia ; centrosomes flat : septa sine doliporis ; muri transversi in basidiis
praebens, sine epibasidiis longis ; centrosomae complanatae).
Class 1. Septomycetes cl. nov. (diagnosis : centrosomes single layered : centrosomae monostromaticae :
name introduced without Latin diagnosis by Cavalier-Smith, 1987 b ; circumscription here narrowed by excluding
the non-septate groups now placed in Ustomycetes).
Subclass 1. Sporidiomycetidae Moore 1980 stat. nov. (Erythrobasidiales, Sporidiales).
Subclass 2. Uredomycetidae Brogniart 1824 stat. nov. (flat multilayered centrosomes) [rusts
(Uredinales), ballistosporous and some other exosporous yeasts (Septobasidiales)].
Subphylum 2. Orthomycotina subphyl. nov. (diagnosis : uniporate septa with dolipores ; centrosomes globular :
septa doliporis muri transversi in basidiis praebentes ; centrosomae globosae : name introduced without Latin
diagnosis by Cavalier-Smith, 1987 b).
Superclass 1. Hemibasidiomycetia Engler 1897 stat. nov.
Class Ustomycetes Moore 1980 (orders Ustilaginales, Tilletiales).
Superclass 2. Hymenomycetia Fries 1821 stat. nov. et em. (typically with complex basidiocarp ; ancestrally
with an exposed hymenium, but enclosed in polyphyletically derived subterranean fruiters ; soma rarely reduced to
yeast phase).
Class 1. Gelimycetes cl. nov. (diagnosis : basidia divided by vertical walls ; gelatinous basidiocarp : muri
longitudinali in basidiis praebens ; corpus fructorum gelatinosum : name introduced without Latin diagnosis by
Cavalier-Smith, 1987 b) (jelly fungi and related yeasts).
Subclass 1. Tremellomycetidae Fries 1821 stat. nov. Wells 1994 (syn. class Exidiomycetes Moore 1994)
subcl. nov. [diagnosis : muri longitudinali in basidiis praebens ; corpus fructorum gelatinosus ; basidia divided by
vertical septa ; saprobic jelly fungi (Tremellales) and related yeasts].
Subclass 2. Dacrymycetidae subcl. nov. [diagnosis : basidia furcate ; basidia furcata (Dacrymycetales).
Subclass 3. Auromycetidae basidia transversely septate with long epibasidia (Auriculariales) : basidia
transversaliter septata sed epibasidiiis longis instructa].
Class 2. Homobasidiomycetes Patouillard 1900 (basidia not subdivided ; basidiocarp usually nongelatinous ; no yeast phases).
Subclass 1. Clavomycetidae Fries 1821 stat. nov. (e.g. Clavaria, Tulasnella).
Subclass 2. Pileomycetidae Fries 1821 stat. nov. (e.g. Fomes, Agaricus, Coprinus, Boletus, Lycoperdon,
Cyathus ; include gasteromycetes, e.g. Phallus, Lycoperdon, Cyathus).
For morphological aspects of the phylogenetic basis of this classification see Cavalier-Smith (1987 b) ; for recent
molecular sequence evidence on neomycote phylogeny see Sugiyama & Nishida (1995). The nature of the kingdom
Fungi and the origin of Microsporidia are discussed in Cavalier-Smith (1998 b) Some evidence suggests that
subphylum Zygomycotina may be polyphyletic (Nagahama et al., 1995 ; Sugiyama, Nagahama & Nishida, 1996), so
it may need to be split in future. Although the names Zygomycotina, Zygomycetes, and Ascomycota have been
used widely for years, they appear not to have been validly published, so I validate them here.
* Probably paraphyletic.
Almost certainly paraphyletic.

A revised six-kingdom system of life

form a new subphylum Dictyomycotina, which
therefore includes all fungi with well developed
Golgi dictyosomes, the ancestral state. Allomycetes
are here grouped with the Zygomycotina (here
ranked as an infraphylum and emended by removal
of the Enteromycetes) in the new subphylum
Melanomycotina. The paraphyletic taxon Chytridiomycotina, which has been long used despite never
being validly published (for higher taxa mycologists
have often ignored the rule of the International
Code of Botanical Nomenclature that requires Latin
diagnoses for the valid publication of new names) is
now abandoned.
The remaining Trichomycetes (Harpellales and
Asellariales) are reduced in rank to a superorder
(Trichomycetalia : there is no standard mycological
suffix for a superorder), and placed in a new
zygomycote class exclusively made up of parasites of
animals or protozoa, which I call Zoomycetes, and
divide into two subclasses. One subclass (Pedomycetidae) includes the Trichomycetalia and a second
new superoder (Pyxomycetalia) created for the
Laboulbeniales and Pyxidiophorales (Blackwell,
1994), ectoparasites of arthropods formerly regarded
as ascomycetes because of the superficial resemblance
of their female organs to perithecia. Both Trichomycetalia and Pyxomycetalia attach to their host
cuticle by similar foot-like holdfasts (whence the
name Pedomycetidae) and have similar imperforate
plugged septa and a body form of discrete cells not
indefinite hyphae. The other new zoomycete subclass
is the Entomycetidae, comprising the orders Entomophorales and Zoopagales, tissue endoparasites of
protozoa or invertebrates, which have usually been
included in the Zygomycetes. Two other new classes
(the saprotrophic Bolomycetes and the endomycorrhizal Glomomycetes) have also been traditionally placed in the Zygomycetes, which are here
restricted to Zygomycotina with stalked sporophores
bearing passively dispersed aerial spores (e.g. the
well-known Mucorales). The present four zygomycote classes are all very much more homogeneous
than the traditional Zygomycetes and Trichomycetes. Though there are no sequences yet available for trichomycetes, those for other zygomycetes
show their great heterogeneity and justify their
subdivision into several classes (Nagahama et al.,
1995). The present system, however, emphasizes
homogeneity of morphology and does not slavishly
follow rRNA trees, since in Fungi these are highly
unclocklike, with numerous unusually long
branches, and so must in some respects be very
misleading.

249
VII. THE KINGDOM PLANTAE AND ITS FIVE
PHYLA

The plant kingdom sensu Cavalier-Smith (1981 a)

comprises all organisms possessing plastids with
double envelopes that are free in the cytoplasm.
Unlike chromists and protozoa, all plants are
obligately dependent on plastids and have never lost
them even when they have become secondarily nonphotosynthetic : this is probably because they are
essential for the synthesis of fatty acids, starch and
certain amino acids (Cavalier-Smith, 1993 e). As the
evidence for the single origin of the chloroplast from
a cyanobacterium and for the monophyly of Plantae
has recently been discussed in detail (CavalierSmith, 1995 a), and the circumscription of Plantae
has remained unchanged since Cavalier-Smith
(1981 a), I shall not discuss them further here, except
to stress that it is not the presence but the
morphology and location of the plastids, together
with their obligate nature, that define the kingdom.
Table 6 summarizes the classification of the kingdom ; from rRNA trees it is clear that the three major
lineages, Viridiplantae, Glaucophyta and Rhodophyta diverged almost simultaneously, so closely
following the origin of chloroplasts that it is hard to
determine their correct branching order or even to
corroborate the monophyly of the kingdom by gross
sequence similarity ; but there is other molecular
evidence for the monophyly of Plantae in the present
sense (Ragan & Gutell, 1995). The monophyly of
Plantae has been questioned on the basis of RNA
polymerase sequence trees (Stiller & Hall, 1997), but
there are too few taxa yet on these trees to have high
confidence in their branching order.
(1) New red algal subphyla
Currently red algae are divided into two classes : the
paraphyletic Bangiophyceae and holophyletic Florideophyceae. Bangiophyceae are ultrastructurally so
diverse that I divide them into two classes : Bangiophyceae sensu stricto (Bangiales, Rhodochaetales)
which have pit connections and Rhodellophyceae
(Porphyridiales, Cyanidiales, Compsopogonales)
which do not. In having pit connections, and on
rRNA trees, Bangiales are closer to Florideophyceae
(therefore I group them as the new subphylum
Macrorhodophtina) than to the Rhodellophyceae,
which I place alone in the new subphylum Rhodellophytina. The Porphyridiales are themselves very
diverse and should probably be subdivided into
more than one order.

250

T. Cavalier-Smith

Table 6. Classification of the kingdom Plantae and its five phyla

Subkingdom 1. Biliphyta* Cavalier-Smith 1981.
Infrakingdom 1. Glaucophyta infrak. nov. (diagnosis : peptidoglycan in plastid envelope : plastidae
peptidoglycanum instructae) (glaucophytes).
Phylum Glaucophyta Skuja 1954 (unnecessary syn. Glaucocystophyta Kies & Kremer 1986) (e.g. Cyanophora).
Infrakingdom 2. Rhodophyta infrak. nov. (diagnosis : plastid envelope lacks peptidoglycan : sine
peptidoglycano).
Phylum Rhodophyta Wettstein 1922 (red algae).
Subphylum 1. Rhodellophytina* subphyl. nov. (diagnosis : unicellular, e.g. Porphyridium, or undifferentiated
simple or branched uniseriate filaments with a basal disc, e.g. Stylonema ; pit connections absent : cellulae unicae
aut filamentae simplices uniseriatae cum disco basalo : sole class : Rhodellophyceae cl. nov. diagnosis as for
subphylum).
Subphylum 2. Macrorhodophytina subphyl. nov. (diagnosis : multicellular with extensive cell differentiation
and complex life histories ; pseudoparenchymatous or multiseriate filaments or flat sheets of cells ; pit connections
usually present : thallus multicelllularis ; saepe in forma pseudoparenchyma aut filamentae multiseriatae) (classes :
Bangiophyceae em. ; Floridiophyceae).
Subkingdom 2. Viridiplantae Cavalier-Smith 1981 (green plants).
Infrakingdom 1. Chlorophyta Cavalier-Smith 1993.
Phylum Chlorophyta auct.
Subphylum 1. Chlorophytina subphyl. nov. (diagnosis : cytokinesis without a phragmoplast ; multilayered
structure (MLS) ciliary roots absent except in Mesostigma, Pterosperma, Halosphaera sine phragmoplastis)
(chlorophyte green algae).
Infraphylum 1. Prasinophytae infraphyl. nov. (diagnosis : usually scaly flagellates with persistent
telophase spindle ; without phycoplast ; Mg 2, 4 divinyl porphyrin often used as a photosynthetic pigment : cellulae
cilio aut ciliis instructa, usitate squamatae ; fusus persistens in telophaso ; sine phycoplasto) [classes
Micromonadophyceae Mattox & Stewart 1984 em. with cruciate ciliary roots : orders Mamiellales Moestrup 1984
em., Pyramimonadales em., Mesostigmatales ord. nov. (diagnosis : biflagellates ; cruciate ciliary roots with one
MLS ; tubular ciliary hairs absent : cilia dua ; radices microtubulorum in forma crucis cum structura laminarum
multarum ; cilia sine pilis tubulatis ; formerly included in Pyramimonadales) ; Nephrophyceae Cavalier-Smith 1993,
with asymmetric microtubular roots, e.g. Nephroselmis, Pseudoscourfieldia].
Infraphylum 2. Tetraphytae infraphyl. nov. (diagnosis : cruciate microtubular roots without MLS ;
telophase spindle collapses ; cytokinesis involves a phycoplast : radices microtubulorum in forma crucis ; sine
structura laminis multis) (e.g. Ulva, Chlamydomonas, Chlorella ; type genus Tetraselmis).
Subphylum 2. Phragmophytina subphyl. nov. (diagnosis : vegetative cells non-motile ; zoospores or sperm
with asymmetric ciliary roots typically with MLS ; cytokinesis usually involves a phragmoplast) (to avoid ambiguity
it is best to call these plants phragmophytes informally, and to reserve the term charophyte strictly for the
infraphylum Charophytae).
Infraphylum 1. Charophytae Engler 1887 stat. nov. (haploid multicellular macrophytes with main axis
corticated with whorls of lateral branches ; with oogonia and antheridia surrounded by sterile cells) (class
Charophyceae sensu stricto : order Charales : stoneworts the traditional charophytes).
Infraphylum 2. Rudophytae infraphyl. nov. [diagnosis : antheridia absent or without investing sterile cells :
antheridia non cellulis sterilis circumcincta : (descriptive name : suggested vernacular term : rudophytes, meaning
simple Latin rudis plants)][classes Eophyceae cl. nov. (diagnosis : with biciliate zoospores or sperm : zoosporae
aut gametae masculinae ciliis instructae : the name eophyte is chosen because the Chaetophycidae may be ancestral
to land plants : Graham, 1993) subclasses Stichophycidae subcl. nov. (diagnosis : sarcinoid or unbranched filaments :
cells lack sheathed hairs : thallus non ramosus ; cellulae sine setis vaginatis ; descriptive name to indicate that the
most complex forms have only a single row stichus of cells) (orders Chlorokybales, Klebsormidiales) ;
Chaetophycidae cl. nov. (diagnosis : many cells have protective sheathed hairs (chaetae) : cellulae setis vaginatis
munitae) branched filaments or parenchymatous with sheathed hairs (Coleochaetales) ; Conjugophyceae auct. (all
cells, including gametes, lack cilia) (Desmidiales, Zygnematales)].
Infrakingdom 2. Cormophyta Endlicher 1836 (syn. Embryophyta auct.) stat. nov.
Phylum 1. Bryophyta Braun 1864 em. Eichler 1883 (hornworts, liverworts, mosses).
Subphylum 1. Hepaticae auct. stat. nov. (liverworts, e.g. Riccia, Marchantia, Lophocolea).
Subphylum 2. Anthocerotae auct. stat. nov. (e.g. Anthoceros).
Subphylum 2. Musci Linnaeus 1753 stat. nov.
Infraphylum 1. Sphagneae auct. stat. nov. (leaves with large porous dead cells as well as living ones ;
Sphagnum).

A revised six-kingdom system of life

251

Table 6. (cont.)
Infraphylum 2. Bryatae infraphyl. nov. (diagnosis : cells in leaves all alive with chloroplasts : cellulae omnis
in laminis chloroplastis instructae) (e.g. Andreaea, Bryum, Funaria ).
Phylum 2. Tracheophyta phyl. nov. (diagnosis : diploid phase with xylem and phloem : facies diploida xylem et
phloem instructa : name introduced without Latin diagnosis by Sinnott, 1935).
Subphylum 1. Pteridophytina Eichler 1883 stat. nov. (pteridophytes).
Infraphylum 1. Psilophytae infraphl. nov. (sine radicibus ; without roots, e.g. Psilophyton, Psilotum).
Infraphylum 2. Lycophytae auct. stat. nov. (e.g. Lepidodendron, lycopods, Selaginella, Isoetes).
Infraphylum 3. Sphenophytae auct. stat. nov. (e.g. Sphenophyllum, horsetails).
Infraphylum 3. Filices* Linnaeus 1753 stat. nov. (ferns).
Subphylum 2. Spermatophytina auct. stat. nov. (seed plants).
Infraphylum 1. Gymnospermae auct. ( seed ferns , cycads, conifers, gnetophytes).
Infraphylum 2. Angiospermae auct. (dicot and monocot flowering plants).

(2) New chlorophyte infraphyla

It has long been accepted that the ancestral green
plants are the scaly prasinophyte algae (Mattox &
Stewart, 1984), and that there was a basic bifurcation within green algae between the Chlorophytina and Charophytina. Because cormophytes
evolved from a charophytine algae, Charophytina
have often been treated as a separate phylum
(Jeffrey, 1971 ; Cavalier-Smith, 1993 e, 1995 a) to
enable Charophyta and Cormophyta to be grouped
together as an infrakingdom Streptophyta. It now
appears that the prasinophyte Mesostigma is cladistically closer to the traditional charophytes than to
the other prasinophytes (Melkonian, Marin &
Surek, 1995). Because of this and the fact that it is
the only prasinophyte with a multilayed structure, it
was recently transferred from the Chlorophyta to the
Charophyta (Cavalier-Smith, 1995 a). However, this
further diminishes the phenotypic difference between these two major green algal taxa ; therefore
though the recognition of two green algal phyla is
cladistically attractive it creates a boundary at a
point where the phenotypic difference seems too
slender to justify separate phyla. Therefore I revert
to the more traditional simpler system with a single
green algal phylum Chlorophyta divided into two
new subphyla. Streptophyta remains as a clade
name only.
As before (Cavalier-Smith, 1993 e), Chlorophytina
are divided into two major taxa (now ranked as
infraphyla) according to their cell division mechanism : Prasinophytae (biciliate or uniciliate with
persistent spindle) and Tetraphytae with collapsing
spindle and phycoplast. Charophytina are also subdivided into two infraphyla : Charophytae and
Coleophytae.

VIII. CHROMISTA, THE THIRD BOTANICAL

KINGDOM, AND ITS FIVE PHYLA

Most chromists are algae with chloroplasts containing chlorophylls a and c which are located not in
the cytosol (as in plants or in protozoan algae) but
within the lumen of the rough endoplasmic reticulum. In addition to their double chloroplast envelopes, chromistan plastids are surrounded by an
additional smooth membrane, the periplastid membrane (Cavalier-Smith, 1989 b). All chromistan algae
are evolutionary chimaeras between a eukaryotic
host and a eukaryotic (probably red algal) symbiont ;
the periplastid membrane is the relic of the red algal
plasma membrane. The unique location of plastid
plus periplastid membrane within the rough endoplasmic reticulum arose by the fusion of the
former phagosomal membrane that first enclosed the
endosymbiont with the nuclear envelope (Whatley,
John & Whatley, 1979 ; Cavalier- Smith, 1982,
1986 a). The evidence for the monophyly of the
Chromista is discussed in detail elsewhere (CavalierSmith, 1995 a). Recent molecular evidence (Cavalier-Smith, Allsopp & Chao 1994 a ; Cavalier-Smith
& Chao, 1997 a) makes it clear that chlorarachnean
algae, originally excluded from Chromista but
temporarily transferred into this kingdom because of
misleading rRNA trees (Cavalier-Smith 1993 a,
1994) are actually Protozoa, as considered previously
(Cavalier-Smith 1986 a).
Chromist monophyly remains controversial for
two reasons ; one is that the three major chromist
groups (Cryptista, Heterokonta and Haptophyta ;
see Table 7) diverged from each other almost
simultaneously with the three major plant lineages,
and the six lineages can appear to diverge in almost

T. Cavalier-Smith

252

Table 7. Classification of the kingdom Chromista and its five phyla

Subkingdom 1. Cryptista Cavalier-Smith 1989.
Phylum Cryptophyta Cavalier-Smith 1986 (e.g. Cryptomonas, Goniomonas).
Subkingdom 2. Chromobiota Cavalier-Smith 1991.
Infrakingdom 2. Heterokonta Cavalier-Smith 1986 stat. nov. 1995 em.
Superphylum 1. Sagenista superphyl. nov. (Diagnosis as for phylum Sagenista Cavalier-Smith 1995 b p. 1014).
Phylum Sagenista Cavalier-Smith 1995.
Subphylum 1. Bicoecia Cavalier-Smith 1989 [orders Bicoecales Grasse! & Deflandre 1952 (Bicoeca),
Anoecales Cavalier-Smith 1997 (e.g. Cafeteria) ; Pirsoniales ord. nov. [diagnosis : syncytial non-motile trophic phase
that feeds on diatoms ; with trophosome inside and auxosome outside the frustule ; zoospores with non-tubular hairs
on cell body : diatomas devorant ; in statu pabulatorio cellulae nuclei plures instructae ; trophosoma sine nucleis
intra frustulo, auxosoma multinucleata extra frustulo ; zoosporae pili non-tubulati vestitae) (sole family Pirsoniaceae
fam. nov. diagnosis as for order Pirsoniales : type genus Pirsonia].
Subphylum 2. Labyrinthista Cavalier-Smith 1986 stat. nov. 1989 (e.g. Labyrinthula, Thraustochytrium).
Superphylum 2. Gyrista superphyl. nov. (diagnosis : ciliary transition region commonly with a helix or a
double helix}ring system : regio transitoria ciliorum plerumque helicem aut helices praebens).
Phylum 1. Ochrophyta* Cavalier-Smith 1986 (as Ochrista) stat. nov. 1995.
Subphylum 1. Phaeista Cavalier-Smith 1995.
Infraphylum 1. Hypogyrista Cavalier-Smith 1995 (e.g. pedinellids, silicoflagellates, pelagophytes).
Infraphylum 2. Chrysista Cavalier-Smith 1986 stat. nov. 1995 (superclasses Phaeistia Cavalier-Smith
1995, e.g. brown algae, xanthophytes, Chrysomeris ; Limnistia Cavalier-Smith 1996, e.g. chrysophytes, eustigs,
Oikomonas, raphidophytes).
Subphylum 2. Diatomeae Dumortier 1821 stat. nov. Cavalier-Smith 1995 (centric and pennate diatoms).
Phylum 2. Bigyra phyl. nov. (diagnosis : ciliary transition region with double helices or concertina-like rings ;
without plastids : regio transitoria ciliorum annuli in forma concertinae praebens ; sine plastidis).
Subphylum 1. Bigyromonada subphyl. nov. (diagnosis : biciliate free-living bacterivorous phagotrophs
without cell walls ; retronemes on anterior cilium ; double ciliary transition helix : cilia dua ; murus absens ;
mastigonemae tubulatae in cilium anterius ; pabulum cellulae prokaryotae est ; regio transitoria ciliorum annuli in
forma concertinae praebens) (Developayella Tong 1995).
Subphylum 2. Pseudofungi Cavalier-Smith 1986 em. 1989 (oomycetes, hyphochytrids).
Subphylum 3. Opalinata Wenyon 1926 stat. nov. em. Cavalier-Smith 1993, 1997 (diagnosis : retronemes
absent from cilia ; gut commensals without peroxisomes : sine mastigonemis tubulatis in ciliis) (Proteromonas,
Karotomorpha, opalinids).
Infrakingdom 2. Haptophyta Cavalier-Smith 1995.
Phylum Haptophyta Cavalier-Smith 1986 (e.g. Pavlova, Prymnesium).
For a more detailed classification of Ochrophyta and Haptophyta see Cavalier-Smith & Chao (1996 b) and
Cavalier-Smith et al. (1996 d) respectively.
* Probably paraphyletic.
Almost certainly paraphyletic.

any order on different molecular trees ; the other is

that all three of the major lineages contain some nonphotosynthetic species, and it has been hard to
establish whether these are primarily or secondarily
photosynthetic. Many authors, following Margulis
(1970) and Whatley et al. (1979), have assumed that
all are primarily non-photosynthetic and that there
were three (or even more) independent implantations of eukaryotic algae into unrelated protozoan
hosts. In contrast, I have argued that all are
secondarily non-photosynthetic and originated by
several independent losses of chloroplasts and that
all chromists had a common photophagotrophic
ancestor the first eukaryote-eukaryote chimaera

that rapidly diverged into the three major lineages

(Cavalier-Smith, 1982, 1986 a). Given our present
knowledge of chromist phylogeny, one has to
postulate a minimum of six independent losses of
chloroplasts within the kingdom.
(1) Multiple losses of chloroplasts by
chromists
Evidence for chloroplast loss by chromists is growing ;
it is strongest for the Heterokonta. We now have
strong molecular phylogenetic evidence for three
losses (two in pedinellids : Cavalier-Smith, Chao &
Allsopp, 1995 ; Cavalier-Smith & Chao, 1996 b ; and

A revised six-kingdom system of life

one in Oikomonas : Cavalier-Smith et al., 1996 b)
within the predominantly algal phylum Ochrophyta
(Cavalier-Smith & Chao, 1996 b). We have weaker
evidence for a third loss in the ancestor of the
Pseudofungi (which comprise the classes Oomycetes
and Hyphochytrea, traditionally thought of as
fungi), which may have evolved from an ochrophyte
algal ancestor by plastid loss (Cavalier-Smith et al.,
1996 b). Recent maximum likelihood trees for rRNA
also suggest that Goniomonas, the sole aplastidic genus
in the phylum Cryptophyta, may also have arisen by
plastid loss (Cavalier-Smith et al., 1996 c). Thus, we
have strong evidence from the internal phylogeny of
the Heterokonta and Cryptista for three, and some
evidence for five, of the six plastid losses required by
the theory of a photophagotrophic latest common
ancestor for chromists. We lack such internal
evidence only for one established chromist group,
the phylum Sagenista (bicoecids, thraustochytrids
and labyrinthulids), which diverges from other
heterokonts near the base of the heterokont clade on
rRNA trees, before the earliest divergences within
the mostly photosynthetic Ochrophyta. Since molecular trees (contrary to what has been implied by
some authors) do not refute the quite strong
morphological and chemical evidence for the monophyly of Chromobiota (discussed in detail by
Cavalier-Smith, 1994 ; Cavalier-Smith et al., 1996 d),
it is much more parsimonious to postulate that
Sagenista also have lost chloroplasts than to suggest
that Ochrophyta and Haptophyta acquired indistinguishable fucoxanthin-containing chloroplasts independently, as some authors (e.g. Leipe et al., 1994 ;
Bhattacharya & Medlin, 1995) assume.

(2) Transfer of Opalinata to Chromista : the

new heterokont phylum Bigyra
When the kingdom Chromista was first established
(Cavalier-Smith, 1981 a) and its origin and systematics discussed in detail (Cavalier-Smith, 1986 a)
it was explicitly recognized that some organisms not
then placed in the kingdom might actually belong
there if they had secondarily lost both the defining
synapomorphic characters that were used to define it
(plastids within periplastid membrane inside rough
endoplasmic reticulum and bipartite or tripartite
ciliary hairs). The possibility that proteromonads
were really chromists, not protozoa (Cavalier-Smith,
1981 a), was suggested by their bipartite tubular
body hairs and ciliary transition helix. At present,
proteromonads are included with Opalinea, which

253

lack tubular hairs, in the subphylum Opalinata

(Cavalier-Smith, 1997 a). For reasons discussed elsewhere, I have recently left this taxon outside the
kingdom Protozoa (Cavalier-Smith, 1997 a). I now
place this subphylum within the Chromista in the
infrakingdom Heterokonta. This change is made as
a result of the recent discovery of a distinctive new
type of heterotrophic heterokont, Developayella elegans
(Tong, 1995), and a reappraisal of the evolution of
the ciliary transition region and cell surface in
chromists.
Developayella elegans resembles both Pseudofungi
and Opalinata in having a double helix distal to the
transverse plate in the ciliary transitional region,
which somewhat resembles a concertina in longitudinal section. The structure of the ciliary transition
region has proved to be an excellent phylogenetic
marker in the past ; for example the nine-fold star
that characterizes the subkingdom Viridiplantae
(Manton, 1965 ; Cavalier-Smith, 1981 a). Recently,
I have given it considerable weight in revising the
higher level classification of the heterokont algal
phylum Ochrophyta (earlier spelled Ochrista)
(Cavalier-Smith, 1995 a). I have divided the ochrophyte subphylum Phaeista into two infraphyla :
Hypogyrista, characterized by a short single ciliary
transitional helix located below (proximal to) the
transition plate and Chrysista characterized by a
longer single helix above the transition plate. These
differences are congruent with other morphological
differences and with molecular phylogeny (CavalierSmith & Chao, 1996 b). It therefore seems likely that
the double transitional helix is also a strong
phylogenetic character. Accordingly, I place Developayella elegans, Pseudofungi and Opalinata together to form a new heterotrophic heterokont
phylum, the Bigyra ; the name refers to the two
helices or sets of rings (which interpretation is correct
is not entirely clear) in the ciliary transition region
(from Latin bi-, meaning twice , and gyrus, meaning
circle or ring ; pronounced Bi- ji-ra ). This
change accepts Pattersons (1989) view that the
tubular somatonemes of the genus Proteromonas
evolved from heterokont ciliary retronemes, not vice
versa, and that the absence of retronemes or
somatonemes in Opalinea is the result of secondary
loss.
I have long considered that the bipartite hairs of
the genus Proteromonas and chromists are homologous ; Opalinata must either be sisters of Chromista
or derived from them. They are unlikely to be
directly ancestral to them since they lack both
peroxisomes and phagotrophy and are all gut

254

commensals of tetrapod vertebrates. A free-living

heterotrophic flagellate with tubular body hairs like
proteromonads but with phagotrophy and peroxisomes would be a suitable host for the origin of
chromists as postulated previously (Cavalier-Smith,
1986 a, 1989 b) ; however, flagellates with this exact
combination of characters are not currently known
and I now accept Pattersons (1989) alternative
hypothesis that proteromonads are derived from
heterokonts by transfer of ciliary hairs to the cell
body. My original interpretation of chromist origins
implicitly viewed Proteromonadida as a sister group
to Chromista (Cavalier-Smith, 1986 a) ; however, I
overlooked the fact that the long-known presence of
a double transitional helix in Opalinata (Brugerolle
& Joyon, 1975) made this assumption inconsistent
with my proposal in the same paper that the differing
transition region structures of the Heterokonta,
Cryptista and Haptophyta arose independently
during the divergence of these three taxa from the
ancestral chromist. One of these two contradictory
phylogenetic hypotheses must be wrong, unless the
double helical structures of Opalinata and Heterokonta are convergent rather than homologous, which
is possible but seems unlikely. If Opalinata were in
fact sisters to, rather than derived from, the chromists
the ancestral chromist must have had a double
transition helix and this must have been replaced by
quite different structures on three separate occasions :
the double plates of cryptomonads and haptophytes
and the bell-shaped structure of Labyrinthulea.
Originally, I assumed that the bell-shaped structure of Labyrinthulea was derived from the single
transitional helix of Chrysista and was related to the
oomycete}hyphochytrid double helix, and therefore
placed Labyrinthulea within the subphylum Pseudofungi (Cavalier-Smith, 1986 a). Later (CavalierSmith, 1989 b), I became very sceptical of this
relationship and removed Labyrinthulea from
Pseudofungi as a separate subphylum Labyrinthista,
and suggested that Labyrinthista diverged very early
in heterokont evolution before the divergence of the
major classes of ochrophyte algae and the origin of
Pseudofungi from a xanthophyte-like ochrophyte by
the loss of photosynthesis. Recent molecular evidence
strongly supports this early divergence of the
Labyrinthista (Cavalier-Smith, Allsopp & Chao,
1994 b ; Leipe et al., 1994), and shows that
Pseudofungi are not specifically related to Labyrinthista and may indeed be secondarily derived by
plastid loss from a phaeistan algal ancestor
(Cavalier-Smith et al., 1996 b). This early divergence
of labyrinthists and later origin of pseudofungi

T. Cavalier-Smith
considerably strengthens Pattersons (1989) postulate that the ciliary double transitional helix is a
derived condition within the Heterokonta. As there
seem no obvious reasons for a replacement of an
ancestral double helix by the different structures of
the Cryptista, Haptophyta, labyrinthists and hypogyrists, I prefer to accept Pattersons interpretation
and to abandon the idea that retronemes evolved
from somatonemes.

(3) The origin of ciliary retronemes and the

nature of the ancestral chromist
The alternative interpretation for the origin of ciliary
retronemes is that they evolved from pre-existing
non-tubular ciliary hairs (Cavalier-Smith, 1989 b).
Originally I did not favour this view because one
could not envisage a gradual origin for them in this
way because of their thrust-reversing properties,
whereas a gradual origin for tubular somatonemes
was more plausible (for the detailed arguments see
Cavalier-Smith, 1986 a : 317320). However if one
accepts, as I think we now must, a more saltatory
origin by a single mutation converting a non-tubular
hair into a rigid tube, which is not mechanistically
impossible, one can envisage an origin in situ on the
ciliary surface for functional tubular retronemes.
Simple non-tubular hairs suitable as ancestors for
retronemes exist in both the Glaucophyta (in
Cyanophora) and in the dinoflagellate and protalveolate Dinozoa. It is therefore highly probable that
such hairs were present in the common ancestor of
Plantae and Alveolata in which chloroplasts are
postulated to have first evolved (Cavalier-Smith,
1982, 1993 e, 1995 a). This favours the idea that the
host for the symbiogenetic origin of chromists may
not have been a heterotrophic protozoan (Whatley et
al., 1979 ; Gibbs, 1981) but an early photosynthetic
transitional alga in the actual process of converting
the ancestral cyanobacterium into the first chloroplast, a possibility first suggested by Cavalier-Smith
(1986 a) and elaborated in detail by Ha$ uber et al.
(1994).
The usual tendency of the three major plant
groups and three major chromist groups to intermingle on rRNA trees strongly supports the view
(Cavalier-Smith, 1989 b) that the basal diversification of both plants and chromists was virtually
simultaneous (Cavalier-Smith, 1986 a, 1995 a). If the
ancestral host was an early plant that arose after the
initial divergence of glaucophytes, rhodophytes and
Viridaeplantae, then the kingdom Plantae would

A revised six-kingdom system of life

actually be paraphyletic rather than holophyletic. If
on the other hand the host was an early photosynthetic intermediate between dinoflagellates and
plants, then Plantae would be holophyletic ; in the
latter case, the host might have contributed the
chlorophyll c of chromists and the red algal symbiont
the phycobilins. Current molecular evidence cannot
decide between these two possibilities ; nor can it rule
out the third possibility that the host was a
heterotrophic non-alveolate zooflagellate protozoan,
since only a tiny proportion of such protozoa have
been placed on the rRNA trees. On many trees,
alveolates branch within the plant}chromist assemblage whereas on others they are an outgroup. It
is very important for our understanding of algal
evolution to determine which position is correct.
Unless future research does identify a non-alveolate
zooflagellate as an outgroup to chromists, the idea
that the host was either an early plant that had not
yet abandoned phagotrophy or an early algal
alveolate will remain an attractive working hypothesis.
A unique heterokont genus possibly significant in
relation to the origin of retronemes is Pirsonia. There
are several species of these little-known phagotrophic
aplastidic chromists ; all feed on diatoms by piercing
their frustule by means of a non-motile syncytial
trophic phase that is bipartite, having a trophosome
within the frustule and auxosomes that remain
outside : their biciliate zoospores have non-tubular
hairs on their cell body and both cilia, in addition to
the usual tripartite retronemes on the anterior cilium
only (Schnepf & Schweikert, 1997). One species also
has dissimilar unipartite tubular hairs on both cilia
(Schweikert & Schnepf, 1997) ; whether or not these
are related to retronemes is unclear, as is the
phylogenetic position of Pirsonia itself. It is sufficiently different from all other heterokonts that I
place it in a new order, Pirsoniales (Table 7).
Provisionally I include Pirsoniales as a third order of
the class Bicoecea of the purely heterotrophic
phylum Sagenista (Cavalier-Smith, 1997 c), on the
assumption that it is an early diverging heterokont.
However a more derived position within the predominantly photosynthetic phylum Ochrophyta
cannot yet be ruled out, even though Ochrophyta at
present contain no totally aplastidic heterotrophic
biciliate taxa. It is likely that a number of presently
ill-characterized heterotrophic protists will eventually be assignable to the Sagenista. Pirsoniids are yet
another example of the considerable diversity of
non-algal body forms within the kingdom Chromista.

255

(4) Mitochondrial and cell-surface evolution

in chromists
A second difficulty with the idea that Opalinata are
an outgroup to the Chromista concerns the evolution
of mitochondrial cristae, which are flattened tubules
in cryptomonads but rounded tubules in Opalinata
and Chromobiota. This problem was avoided originally by assuming that the origin of chromists took
place very shortly after the origin of mitochondria,
before the morphology of cristae was stabilized, and
that cristae developed independently to their modern
form in Opalinata, Chromobiota and Cryptista
(Cavalier-Smith, 1986 a). During the past decade,
however, molecular evidence has steadily grown for
the alternative view (Margulis, 1970 ; Taylor, 1974)
that mitochondria evolved substantially earlier than
chloroplasts. In particular, it is increasingly clear
that tubular cristae evolved substantially before the
origin of chromists and that the cryptist tubules must
have undergone secondary flattening (CavalierSmith, 1991 d). In itself, this phylogenetic conclusion
poses no problem for the idea that Opalinata are
sisters to chromists, since such secondary flattening
must have occurred irrespective of whether or not
Opalinata are sisters to chromists.
However, my recent mechanistic interpretation of
the reasons for this secondary flattening (CavalierSmith, 1997 a) make much more sense if Opalinata
are derived from typical heterokonts than if they are
the chromist outgroup. Based on recent rRNA
evidence that the common ancestor of the four
higher kingdoms of life was probably a somewhat
amoeboid zooflagellate with a relatively fluid cell
surface, I have postulated that cryptist proteinaceous
pellicular plates evolved to stabilize the cell surface
following the acquisition of a plastid and deemphasis of phagotrophy, and that mutations causing the associated changes in the plasma membrane
pleiotropically affected the mitochondrial cristae by
flattening them. If, however, the chromist ancestor
already had an extensive cortical investment of
microtubules as in Opalinata, there would have
been relatively little need for the origin of the
cryptist pellicular plates. Moreover, one might have
expected such a useful cortical skeleton to have been
retained more widely in Chromista if such plates
were the ancestral state, as they have been for
example in the Euglenozoa. Both the opalinate
microtubular bands and the cryptist plates are
therefore probably derived within Chromista.
Because my molecular coevolution theory of the
significance of the major changes in morphology of

256

mitochondrial cristae (Cavalier-Smith, 1997 a) is, at

present, only an initial hypothesis, not a wellcorroborated interpretation, and because it is still
unclear how much earlier mitochondria arose than
plastids, these considerations clearly have less weight
than those concerning the transition helix phylogeny
discussed above. Since, however, they point in the
same direction they add some additional strength to
the conclusion that Opalinata are probably heterokonts.
(5) Retroneme loss in Opalinata
I have argued that the loss of ciliary retronemes by
phagotrophic heterokonts is most unlikely, because it
would reverse the direction of their ciliary feeding
current and so cause starvation unless accompanied
by the evolution of a novel mode of feeding
(Cavalier-Smith, 1986 a). I previously postulated
such loss only for the origin of the Haptophyta, in
which the origin of the haptonema for feeding could
have allowed the loss of retronemes that I have
argued were present in the ancestral chromobiote
(Cavalier-Smith, 1994). Retroneme loss is also highly
unlikely in phototactic heterokont algae or in any
other heterokonts with well-developed tactic behaviour, since it would reverse the direction of
swimming in response to environmental stimuli.
Retroneme loss is therefore highly improbable in any
free-living heterokont flagellate.
I postulate that the conversion of retronemes to
somatonemes uniquely in Proteromonas was possible
as a direct consequence of their colonization of the
gut of tetrapods. This novel habitat involved two
major changes compared with free-living life : first,
being surrounded by an isotropic food supply, a
ciliary feeding current became irrelevant and the
flagellate evolved pinocytotic feeding over its whole
body surface, rather than phagocytosis near the
ciliary base as in the ancestral heterokont. Whether
or not it had already lost phagocytosis and become
a saprotroph like pseudofungi before becoming a gut
symbiont is less important than the fact that, at some
stage, the loss of phagotrophy removed one source of
stabilizing selection for the retention of ciliary
retronemes. The second change was that entry into
an isotropic milieu probably made tactic behaviour
redundant, thus removing the second major source
of selection for retroneme retention. Thus, selection
would no longer have opposed the total loss of
retronemes or their movement onto the body surface.
Why were retronemes moved onto the body
surface to become somatonemes, not simply lost ? In

T. Cavalier-Smith
other words what is their function ? I suggest that
they serve to keep coarse and sharp food particles in
the host gut that are too large for the flagellate to
ingest away from its cell surface where pinocytosis
takes place. This would have two significant advantages : possibly the most important would be to
prevent large inedible particles from adhering to the
cell surface or simply blocking the pinocytotic uptake
of small digestible particles by their very close
proximity to the cell surface ; this could subtantially
increase the effective rate of feeding by pinocytosis.
Secondly, the hairs could reduce the chance of
damage to the delicate pinocytotic regions of the
plasma membrane by either the physical impact or
chemically harmful character of large inedible
particles or by the attempts of the cell to ingest larger
particles than it could handle.
If somatonemes have these advantages to Proteromonas, how is it that they were lost by Opalinea ? I
suggest that their function was replaced by the deep
cortical folds that characterise the class Opalinea
(both opalinids and the quadriciliate genus Karotomorpha). These cortical folds supported by a column
of microtubules are convergent with the actinsupported cortical folds of the apicomplexan gregarine protozoa, which are by far the largest
protozoan cells that inhabit animal guts and which
also feed pinocytotically. This common occurrence
of cortical folds in two such disparate groups strongly
suggests a similar function. The primary function in
both, I suggest, was to keep large inedible particles
away from, and to allow rapid access of small edible
particles to, the sites of pinocytosis ; in Opalinea
these are in the troughs of the cortical folds
(Patterson, 1985) as they also are in gregarines.
Clearly, they also have a skeletal function and have
been a mechanical preadaptation that has allowed
both opalinids and gregarines to grow to a huge size
compared with most protist cells. However, the fact
that the folds are also found in the relatively small
flagellate Karotomorpha bufonis suggests that their
primary function was, as postulated here, to prevent
the occlusion of and}or damage to the pinocytotic
sites by large inedible particles in the hosts gut. The
fact that the folds replaced the proteromonad
somatonemes suggest that they perform these functions more efficiently ; possibly they are more rigid
and less easily pushed aside by large particles in the
churning gut.
The cortical microtubules of Proteromonas to which
the somatonemes are attached were however a
preadaptation for the origin of the microtubular
columns that support the cortical folds. The likely

A revised six-kingdom system of life

absence of such cortical microtubules in the ancestor
of gregarines would explain why their folds are
instead supported by actin microfilaments. Likewise
the attachment of retronemes to ciliary microtubules
was a preadaptation to their subsequent attachment
to cortical microtubules.
The loss of somatonemes would in principle be
easier than the loss of retronemes as it would not
reverse the direction of ciliary hydrodynamic flow.
Thus the conversion of retronemes to the somatonemes of Proteromonas facilitated their subsequent
total loss.
After the present paper was submitted rRNA
sequence evidence was published (Silberman et al.,
1996) showing that Proteromonas lacertae-viridis is
specifically related to the heterokonts, and is therefore not an outgroup for chromists as a whole. This
adds further support to the arguments given above
for the view that retronemes were secondarily
transferred from the cilia to the cell surface of the
ancestral opalinate, and for transferring the Opalinata as a whole into the Heterokonta ; however,
although Proteromonas clearly clusters with the undoubted heterokonts, the published tree does not
actually group it with the Pseudofungi as would be
expected if the Bigyra are holophyletic, as postulated
here. Since however the bootstrap values for the
branching order at the base of the Heterokonta are
low, and since the deepest branches are rather long,
it is possible that some or even all of them are placed
too low in the tree, as may be often be true for the
Pseudofungi themselves (Cavalier-Smith et al.,
1996 b). Additional evidence is therefore needed to
test the monophyly of the Bigyra. Silberman et al.
(1996) also show that the non-ciliated gut symbiont
Blastocystis is related to Proteromonas, which was
entirely unexpected. This means that despite the
absence of cilia Blastocystis should be placed in the
Opalinata, and that it is not a fungus or protozoan
as previously supposed. As it differs from all other
Opalinata in the absence of cilia I place it in a new
class, Blastocystea (diagnosis : sine ciliis without
cilia). Blastocystis is the first chromist known to
parasitize humans. These new data are discussed in
more detail elsewhere (Cavalier-Smith, 1997 b).
(6) Are there other unidentified chromists
lurking within the kingdom Protozoa ?
For the reasons discussed above, the direct loss of
retronemes by phagotrophic heterokonts is so improbable that it is unlikely that any of the well
characterized zooflagellate protozoa are really hetero-

257

konts or cryptophytes that have directly lost retronemes, though such loss is not impossible if an
alternative mode of feeding and}or taxis evolved at
the same time. The remote possibility that foraminifera are heterokonts (Patterson, 1989) is now
excluded by rRNA phylogeny (Pawlowski et al.,
1994, 1996 ; Wray et al., 1995). Haptophytes are,
however, known to be able to lose the haptonema
entirely and some have lost the patelliform scales
found in most Prymnesiophyceae (Cavalier-Smith et
al., 1996 d) ; moreover a few are non-photosynthetic
(Marchant & Thomsen, 1994). A non-photosynthetic haptophyte that had lost both the haptonema and scales could easily be wrongly classified as
a protozoan. Although there is no positive reason to
think that any such misclassified haptophytes exist,
the case of the haptonema-less haptophyte Reticulosphaera japonensis that was misclassified as a heterokont (Grell, Heini & Schu$ ller, 1990) before molecular evidence for its true nature was found
(Cavalier-Smith et al., 1996 d) should alert one to the
possibility that some zooflagellates presently thought
of as protozoa might actually be drastically altered
haptophytes.
Though the loss of retronemes must be very rare,
it is, in principle, much easier for chromists to lose
their flagella totally. This has long been known to
have occurred in pennate diatoms and probably in
Chrysamoeba, and has recently been demonstrated for
Pelagococcus and Aureococcus, though as these are all
photosynthetic they have not been mistaken for
protozoa. If photosynthesis were also lost and the
organism had no other characters identifying it as a
chromist it could be wrongly treated as a nonflagellate protozoan, just as was done for Blastocystis.
It is conceivable that at least a few others of the
numerous parasitic protists of uncertain taxonomic
position listed by Patterson (1994) might actually be
chromists. Several hundred protist genera (those
listed in Table 5 of Patterson, 1994, minus a few like
Blastocystis, Dermocystidium and Trimastix, which have
now been put in phyla) are so poorly studied that
they cannot with confidence be placed in any of the
kingdoms or phyla of the present system, and mostly
not yet been placed in any suprageneric taxon.
Though the vast majority of these genera will
probably turn out to belong somewhere in the
infrakingdom Sarcomastigota of the subkingdom
Neozoa of the kingdom Protozoa (most I suspect in
the Cercozoa), a small minority might turn out to be
chromists, once they are studied by molecular
phylogenetics.
Patterson (1989) reasonably suggested that Diplo-

T. Cavalier-Smith

258

phrys and Sorodiplophrys may actually be non-flagellate thraustochytrids, but molecular evidence is
required to confirm or refute this. One other
organism (Corallochytrium limacisporum) initially
treated as a non-flagellate thraustochytrid (RaghuKumar, 1987) has recently been shown by rRNA
phylogeny to be a non-flagellate choanozoan protozoan instead (Cavalier-Smith & Allsopp, 1996).
Earlier Davidson (1982), Patterson & Fenchel
(1985) and Patterson (1986) proposed that actinophryid heliozoa were non-flagellate heterokonts ;
though Patterson (1994) appears no longer to
support this unconvincing view, the true position of
actinophryids still needs to be checked by rRNA
phylogeny. Pattersons (1994) informal unranked
group stramenopiles is identical in phylogenetic
concept to the infrakingdom Heterokonta. Apart
from the exclusion of Blastocystis, which prior to the
rRNA evidence mentioned above showed no obvious
signs of its heterokont affinities, the inclusion of
Reticulosphaera, which is now known to be an error,
and Commation, and the uncertainty about the
position of Diplophrys, Pattersons (1994) stramenopiles is also identical in composition to the infrakingdom Heterokonta as revised here. I do not
accept the inclusion of Commation in Heterokonta
(Thomsen & Larsen, 1993 ; Patterson, 1994), as I am
unconvinced that these protists have bipartite or
tripartite retronemes ; I have placed Commation
instead in its own order within the protozoan phylum
Neomonada in the class Kinetomonadea (CavalierSmith, 1997 a). In view of the ultrastructural
similarities with Heliomonadida, I have grouped
Heliomonadida and Commatiida together as the
subclass Ramicristia (Cavalier-Smith, 1997 a). The
reasons for not using the unnecessary new synonym
stramenopile in preference to the classical term
heterokont were expounded previously (CavalierSmith, 1993 a).
In sum, though the Chromista may, in future,
need to be augmented slightly by the addition of a
few misplaced heterotrophs, it is unlikely that the
boundary between Chromista and Protozoa will
change substantially, and even possible that both
kingdoms will be entirely stable in circumscription in
the future.

IX. ENVOI

In the present six-kingdom system the only changes

in the circumscription of the six kingdoms from that
of Cavalier-Smith (1983 a) are the transfer of

Opalinata from Protozoa to Chromista ; of Microsporidia from Protozoa to Fungi ; and of Myxozoa
and Mesozoa from Protozoa to Animalia. One can
expect the boundaries of the six kingdoms to be even
more stable in the future. While there remains a need
for a yet more thorough testing of the monophyly of
Plantae and Chromista, the present six-kingdom
system is phylogenetically and taxonomically
sounder, for the reasons discussed previously
(Cavalier-Smith, 1986 a, 1993 a), than the fivekingdom system found in its numerous mutually
contradictory variants in most textbooks. It is also
distinctly simpler, and so practically more convenient and easier for beginning students and general
users to comprehend, than the phylogenetically
congruent eight-kingdom system that I advocated
from 1987 to 1995. I hope that it will be widely
adopted.
The major advances over the past 15 years have
been in the phylogenetically sounder definition of
the phyla, subphyla and infraphyla, involving many
new creations, subdivisions and mergers of major
taxa, mainly in the bacteria, protozoa and heterokont Chromista, and in the definition of subkingdoms and infrakingdoms of all kingdoms. There
is probably still significant scope for further improvements in these respects, especially amongst the
neozoan protozoa, and a need for more detailed
testing of the recent changes, but it is likely that the
spate of new creations of higher level taxa that has
accompanied the recent extension of electron microscopy and molecular phylogeny into previously
unexplored territory will be much reduced in the
ensuing decades, and we can look forward fairly soon
to a period of consolidation and relative stability in
the classification and nomenclature of the major
groups of life.

X. CONCLUSIONS

1. An outline classification of a revised six-kingdom

system of life is presented, down to the level of
infraphylum. Intermediate very high level categories
(superkingdom, subkingdom, branch, infrakingdom,
superphylum, subphylum and infraphylum) are
extensively used to avoid splitting organisms into an
excessive number of phyla (only 60 being recognized) and kingdoms, and to achieve a balanced
system without unwarranted lumping.
2. The circumscription and high level classification of the zoological kingdoms Protozoa and
Animalia are modified in the light of recent

A revised six-kingdom system of life

molecular phylogenetic evidence that the protist
Myxozoa are actually Animalia, not Protozoa, and
that mesozoans are bilaterians. Mesozoa are removed from the kingdom Protozoa and placed as a
new infrakingdom within the subkingdom Bilateria
of the Animalia, which therefore are now subdivided
into three subkingdoms : Radiata (phyla Porifera,
Cnidaria, Placozoa, Ctenophora), Myxozoa, and
Bilateria (bilateral animals : all other phyla).
3. Microsporidia other than the metchnikovellids
are transferred from the kingdom Protozoa to the
kingdom Fungi.
4. I argue that the need for a simple and general
classification of the living world is now best met by
placing Archezoa as a subkingdom within the
Protozoa, as in my 1983 six-kingdom system, rather
than as a separate kingdom as I advocated from
1987 onwards. I group the 13 currently recognized
protozoan phyla into two subkingdoms, Archezoa
and Neozoa, and four neozoan infrakingdoms. The
reasons for these changes are discussed in detail in
relation to the principles of megasystematics, here
defined as systematics that concentrates on the
higher levels of classes, phyla and kingdoms.
5. These principles also make it desirable to rank
Archaebacteria as an infrakingdom of the kingdom
Bacteria, rather than as a separate kingdom.
Archaebacteria are here grouped with the infrakingdom Posibacteria to form a new subkingdom,
Unibacteria, comprising all bacteria bounded by a
single membrane. The existing bacterial subkingdom
Negibacteria, with separate cytoplasmic and outer
membranes, is here subdivided into two infrakingdoms, Lipobacteria, which lack lipopolysaccharide and have only phospholipids in the outer
membrane, and Glycobacteria, which have lipopolysaccharides in the outer leaflet of the outer membrane and phospholipids in the inner leaflet of its
bilayer. Thus, the primary grouping of the 10
bacterial phyla into two subkingdoms is based on the
number of cell envelope membranes, whilst their
secondary subdivision into four infrakingdoms emphasises their membrane chemistry ; the definition of
the negibacterial phyla, five of which are at least
partly photosynthetic, relies chiefly on photosynthetic mechanism and cell-envelope structure
and chemistry corroborated by rRNA phylogeny.
6. The circumscriptions of the kingdoms Protozoa
and Chromista are slightly changed by transferring
the subphylum Opalinata (classes Opalinea and
Proteromonadea) into the kingdom Chromista,
specifically to the infrakingdom Heterokonta, where
it is grouped with the subphylum Pseudofungi and

259

the recently discovered heterotrophic heterokont

Developayella elegans (placed here in the new subphylum Bigyromonada) to form a new purely
heterotrophic botanical phylum, Bigyra. Bigyra are
defined as heterotrophs with a double ciliary
transitional helix. This new grouping makes it
necessary to abandon the phylum name Opalozoa,
in which Opalinata were previously placed. A
detailed evolutionary theory is presented to account
for the loss of ciliary retronemes in Opalinata as a
consequence of their evolution of gut commensalism.
Recent phylogenetic evidence for multiple chloroplast losses within the kingdom Chromista strengthens the view that the ancestral chromist was a
photophagotroph that evolved by a single symbiogenetic event involving a phagotrophic biciliate host
and a red algal endosymbiont ; but the monophyly of
the Chromista still needs to be tested more.
7. No changes are made to the circumscription of
the botanical kingdom Plantae or the kingdom
Bacteria, which have both been stable since
Cavalier-Smith (1981 a). New plant subphyla and
infraphyla are created.
8. The two zoological kingdoms (Protozoa, Animalia) are subject to the Zooological Code of
Nomenclature, the single bacterial kingdom to the
Bacteriological Code of Nomenclature, and the three
botanical kingdoms (Plantae, Fungi, Chromista) to
the Botanical Code of Nomenclature.
XI. ACKNOWLEDGEMENTS
I thank Eric Harley and Mervyn Berman for hospitality
in their Department while this was being written, NSERC
for a research grant and the Canadian Institute for
Advanced Research Evolutionary Biology Programme for
Fellowship support. I thank E. Mayr, A. J. Roger and
P. Keeling for stimulating discussions and J. R. Garey for
sending me a paper in press. If any readers are aware of
any errors in citation, or of other kinds, in this paper I
would be most grateful if they would point them out to
me.
XII. REFERENCES
A, A. M. A., T, J. M.. L, L. S.,
R, M. C., G, J. R., R, R. A. & L, J. A.
(1997). Evidence for a clade of nematodes, arthropods and
other moulting animals. Nature (London) 387, 489493.
A, C. J., M, C. W. & B, M. (1996).
Introductory Mycology, 4th ed. Wiley, New York.
A, A. (1969). Phylogenetic relationships of the Procaryota
and the origin of the eukaryotic cell. New Phytologist 68,
591602.

260
A Z, L., S, M. L. & C, D. A. (1997).
Phylogenetic relationships between the Acantharea and the
Polycystinea : a molecular perspective on Haeckels Radiolaria.
Proceedings of the National Academy of Sciences, USA 94,
1141111416.
A, P. D. (1971). Monophyly and associated terms.
Systematic Zoology 20, 6369.
B, S. L. & P, J. D. (1993). Animals and fungi are
each others closest relatives, congruent evidence from
multiple proteins. Proceedings of the National Academy of Sciences,
USA 90, 1155811562.
B, S. L., P, J. D. & D, W. F. (1994). The
root of the universal tree and the origin of eukaryotes based on
elogation factor phylogeny. Proceedings of the National Academy
of Sciences, USA 93, 77497754.
B, S. L. & D, W. F. (1997). Origin and
evolution of the slime moulds. Proceedings of the National
Academy of Sciences, USA 94, 1200712012.
B, D. & M, L. (1995). The phylogeny of
plastids, a review based on comparisons of small-subunit
ribosomal RNA coding regions. Journal of Phycology 31,
489498.
B, S. M., D, C. F., P, J. D. & P, N. R.
(1996). Perspectives on archeal diversity, thermophily and
monophyly from environmental rRNA sequences. Proceedings
of the National Academy of Sciences, USA 93, 91889193.
B, M. (1994). Minute mycological mysteries : the
influence of arthropods on the lives of fungi. Mycologia 86,
117.
B, G. (1980). Intracellular protein topogenesis. Proceedings
of the National Academy of Sciences, USA 77, 14961500.
B, P. J., L, C. J., P, E. & J, P. J.
(1997). Targeting and translocation of proteins into the
hydrogenosome of the protist Trichomonas : similarities with
mitochondrial protein import. EMBO (European Molecular
Biology Organization) Journal 16, 34843493.
B, D., C, C. W., DS, R. & B, L. D.
W. (1995). Class-level relationships in the phylum Cnidaria :
molecular and morphological evidence. Molecular Biology and
Evolution 12, 679689.
B, G. & J, L. (1975). Etude cytologique
ultrastructurale des genres Proteromonas et Karotomorpha (Zoomastigophorea, Proteromonadida Grasse! 1952). Protistologica
11, 531546.
B, T. D., W, T. J. & T, J. W. (1991). Fungal
molecular systematics. Annual Review of Ecology 22, 525564.
B, E. T. N., B, P. J. & J, P. J. (1996). A
common evolutionary origin for mitochondria and hydrogenosomes. Proceedings of the National Academy of Sciences, USA
93, 96519656.
B, C. J., W, O., O, G. J., Z, L., F,
R. D., S, G. G., B, J. A., F, L. M.,
C, R. A., G, J. D., K, A. R.,
D, B. A., T, J. F., A, M. D., R, C. I.,
O, R., K, E. F., W, K. G., M,
J. M., G, A., S, J. L., G, N. S. M.,
W, J. F., F, J. L., N, D., U,
T. R., K, J. M., P, J. D., S, P. W.,
H, M. C., C, M. D., R, K. M., H, M.
A., K, B. P., B, M., K, H. P., F, C.
M., S, H. O., W, C. R. & V, J. C. (1996).
Complete genome sequence of the methanogenic archaeon,
Methanococcus jannaschii. Science 273, 10581073.

T. Cavalier-Smith
C, E. U. (1990). Phylum Microspora. In : Handbook of
Protoctista (ed. L. Margulis, J. O. Corliss, M. Melkonian and
D. J. Chapman), pp. 5372. Jones & Bartlett, Boston.
C, M. J. & W, S. C. (1994). The Fungi.
Academic Press, London.
C-S, T. (1978). The evolutionary origin and
phylogeny of microtubules, mitotic spindles and eukaryote
flagella. BioSystems 10, 93113.
C-S, T. (1980). Cell compartmentation and the
origin of eukaryotic membranous organelles. In Endocytobiology : Endosymbiosis and Cell Biology (ed. W. Schwemmler and
H. E. A. Schenk), pp. 831916. De Gruyter, Berlin.
C-S, T. (1981 a). Eukaryote kingdoms, seven or
nine ? BioSystems 14, 461481.
C-S, T. (1981 b). The origin and early evolution of
the eukaryotic cell. In Molecular and Cellular Aspects of Microbial
Evolution. Society for General Microbiology Symposium 32
(ed. M. J. Carlile, J. F. Collins and B. E. B. Moseley), pp.
3384. Cambridge University Press.
C-S, T. (1982). The origins of plastids. Biological
Journal of the Linnean Society 17, 289306.
C-S, T. (1983 a). A 6-kingdom classification and a
unified phylogeny. In Endocytobiology II (ed. H. E. A. Schenk
and W. Schwemmler), pp. 10271034. De Gruyter, Berlin.
C-S, T. (1983 b). Endosymbiotic origin of the
mitochondrial envelope. In Endocytobiology II (ed. W.
Schwemmler and H. E. A. Schenk), pp. 265279. De Gruyter,
Berlin.
C-S, T. (1986 a). The kingdom Chromista, origin
and systematics. In Progress in Phycological Research (ed. F. E.
Round and D. J. Chapman), vol. 4, pp. 309347. Biopress,
Bristol.
C-S, T. (l986 b). Cilia versus undulipodia. Bioscience
36, 293294.
C-S, T. (l986 c). The kingdoms of organisms.
Nature 324, 4l64l7.
C-S, T. (1987 a). The origin of cells, a symbiosis
between genes, catalysts and membranes. Cold Spring Harbor
Symposia on Quantitative Biology 52, 805824.
C-S, T. (1987 b). The origin of Fungi and pseudofungi. In Evolutionary Biology of the Fungi (ed. A. D. M. Rayner,
C. M. Brasier and D. M. Moore), (Symp. Br. Mycol. Soc. 13),
pp. 339353. Cambridge University Press.
C-S, T. (1987 c). The origin of eukaryote and
archaebacterial cells. Annals of the New York Academy of Sciences
503, 1754.
C-S, T. (1987 d). The simultaneous symbiotic
origin of mitochondria, chloroplasts, and microbodies. Annals
of the New York Academy of Sciences 503, 5571.
C-S, T. (1989 a). Systems of kingdoms. In McGraw
Hill Yearbook of Science and Technology, pp. 175179.
C-S, T. (1989 b). The kingdom Chromista. In The
Chromophyte Algae, Problems and Perspectives (ed. J. C. Green, B.
S. C. Leadbeater and W. C. Diver), pp. 381407. Clarendon
Press, Oxford.
C-S, T. (1990). Symbiotic origin of peroxisomes. In
Endocytobiology IV (ed. P. Nardon, V. Gianinazzi-Pearson, A.
M. Grenier, L. Margulis and D. C. Smith), pp. 515521.
Paris.
C-S, T. (1991 a). The evolution of cells. In Evolution
of Life (ed. S. Osawa and T. Honjo), pp. 271304. SpringerVerlag, Tokyo.

A revised six-kingdom system of life

C-S, T. (1991 b). Evolution of prokaryotic and
eukaryotic cells. In Foundations of Medical Cell Biology (ed. G.
E. Bittar), Vol. I, pp. 221278. J.A.I. Press, Greenwich, CT,
USA.
C-S, T. (1991 c). Intron phylogeny : a new hypothesis. Trends in Genetics 7, 145148.
C-S, T. (1991 d). Cell diversification in heterotrophic flagellates. In The Biology of Free-living Heterotrophic
Flagellates (ed. D. J. Patterson & J. Larsen), pp. 113131.
Oxford.
C-S, T. (1992 a). Origins of secondary metabolism.
In Secondary Metabolites, their Function and Evolution (ed. D. J.
Chadwick and J. Whelan), pp. 6487. Wiley, Chichester.
C-S, T. (1992 b). Bacteria and eukaryotes. Nature
356, 570.
C-S, T. (1993 a). Kingdom Protozoa and its 18
phyla. Microbiological Reviews 57, 953994.
C-S, T. (1993 b). Evolution of the eukaryotic
genome. In The Eukaryotic Genome, Organization and Regulation
(ed. P. M. A. Broda, S. G. Oliver & P. F. G. Sims ), pp.
333385. Cambridge University Press.
C-S, T. (1993 c). Percolozoa and the symbiotic
origin of the metakaryote cell. In Endocytobiology V (ed. H.
Ishikawa, M. Ishida, and S. Sato), pp. 399406. Tu$ bingen
University Press.
C-S T. (1993 d). The protozoan phylum Opalozoa.
Journal of Eukaryotic Microbiology 40, 609615.
C-S, T. (1993 e). The origin, losses and gains of
chloroplasts. In Origin of Plastids : Symbiogenesis, Prochlorophytes
and the Origins of Chloroplasts (ed. R. A. Lewin), pp. 291348.
Chapman & Hall, New York.
C-S, T. (1994). Origin and relationships of Haptophyta. In The Haptophyte Algae (ed. J. C. Green and B. S. C.
Leadbeater), pp. 413435. Clarendon Press, Oxford.
C-S, T. (1995 a). Membrane heredity, symbiogenesis, and the multiple origins of algae. In Biodiversity and
Evolution (ed. R. Arai, M. Kato and Y. Doi), pp. 75114.
Tokyo.
C-S, T. (1995 b). Cell cycles, diplokaryosis, and the
archezoan origin of sex. Archiv fuX r Protistenkunde 145, 189207.
C-S, T. (1995 c). Evolutionary protistology comes
of age. Archiv fuU r Protistenkunde 145, 145154.
C-S, T. (1995 d). Zooflagellate phylogeny and
classification. Cytology 37, 10101029.
C-S, T. (1997 a). Amoeboflagellates and mitochondrial cristae in eukaryote evolution, megasystematics of the
new protozoan subkingdoms Eozoa and Neozoa. Archiv fuX r
Protistenkunde 147, 237258.
C-S, T. (1997 a). Sagenista and Bigyra, two phyla
of heterotrophic heterokont chromists. Archiv fuX r Protistenkunde
148, 253267.
C-S, T. (1998 a). Neomonada and the origin of
animals and fungi. In Evolutionary Relationships Among Protozoa
(eds. G. H. Coombs, K. Vickerman, M. A. Sleigh and A.
Warren), pp. 375407. Chapman & Hall, London.
C-S, T. (1998 b). What are Fungi ? In : The Mycota
(ed. D. J. McLaughlin), Vol. VII. Springer-Verlag, in press.
C-S, T. & A, M. T. E. P. (1996). Corallochytrium, an enigmatic non-flagellate protozoan related to
choanoflagellates. European Journal of Protistology 32, 19.
C-S, T. & C, E. E. (1995). The opalozoan
Apusomonas is related to the common ancestor of animals,

261
fungi, and choanoflagellates. Proceedings of the Royal Society of
London B 261, 16.
C-S, T. & C, E. E. (1996 a). Molecular
phylogeny of the free-living archezoan Trepomonas agilis and
the nature of the first eukaryote. Journal of Molecular Evolution
43, 551562.
C-S, T. & C, E. E. (1996 b). 18S rRNA
sequence of Heterosigma carterae (Raphidophyceae), and the
phylogeny of heterokont algae (Ochrophyta). Phycologia 35,
500510.
C-S, T. & C, E. E. (1997). Sarcomonad
ribosomal RNA sequences, rhizopod phylogeny, and the
origin of euglyphid amoebae. Archiv fuX r Protistenkunde 147,
227236.
C-S, T., A, M. T. E. P. & C, E. E.
(1994 a). Chimeric conundra : are nucleomorphs and chromists monophyletic or polyphyletic ? Proceedings of the National
Academy of Sciences, USA 91, 1136811372.
C-S, T., A, M. T. E. P. & C, E. E.
(1994 b). Thraustochytrids are chromists, not Fungi : 18S
rRNA signatures of Heterokonta. Philosophical Transactions of
the Royal Society of London B 339, 139146.
C-S, T. C, E. E. & A, M. T. E. P.
(1995). Ribosomal RNA evidence for chloroplast loss within
Heterokonta, pedinellid relationships and a revised classification of ochristan algae. Archiv fuX r Protistenkunde 145, 209220.
C-S, T., A, M. T. E. P., C, E. E.,
B-E, N. & V, J. (1996 a). Sponge phylogeny, animal monophyly and the origin of the nervous
system : 18S rRNA evidence. Canadian Journal of Zoology 74,
20312045.
C-S, T., C, E. E., T, C. & H, S. (1996 b). Oikomonas, a distinctive zooflagellate related
to chrysomonads. Archiv fuX r Protistenkunde 146, 273279.
C-S, T., C , J. A., T, K. E.,
G, P., D, J., H, D. A. & MF, G. I.
(1996 c). Cryptomonad nuclear and nucleomorph 18S rRNA
phylogeny. European Journal of Phycology 31, 315328.
C-S, T., A, M. T. E. P., H$ , M. M.,
R, S. A., G, G., C, E. E., C, J. A. &
M, U-G. (1996 d). Chromobiote phylogeny : the enigmatic alga Reticulosphaera japonensis Grell is an aberrant
haptophyte, not a heterokont. European Journal of Phycology 31,
255263.
C, R., R, A., B, A., P, R. G, K.
G. & A, A. (1991). An analysis of the origin of
metazoans, using comparisons of partial sequences of the 28S
RNA, reveals an early emergence of triploblasts. EMBO
Journal 10, 499503.
C G. & R, A. (1995). Direct evidence for secondary
loss of mitochondria in Entamoeba histolytica. Proceedings of the
National Academy of Sciences, USA 92, 65186521.
C, R. B. (1964). Dynamics in Metazoan Evolution. Clarendon
Press, Oxford.
C, B. L., G, A. B. & C-S, T. (1998).
Molecular phylogeny of brachiopods and phoronids based on
nuclear-encoded small subunit ribosomal RNA gene sequences. Philosophical Transactions of the Royal Society, B, in
press.
C, F. (1875). Untersuchungen u$ ber Bacterien II. BeitraX ge zur
Biologie der Pflanzen 1, Heft 3, 141207.
C, S. T. (1839). Aids to Reflection xv 64.

262
C M, S., G, J. D., G, R. & P, H.
M. (1985). The Origins and Relationships of Lower Invertebrates.
Clarendon Press, Oxford.
C, H. F. (1956). The Classification of Lower Organisms.
Pacific Books, Palo Alto.
C, J. O. (1971). Common sense and courtesy in nomenclatural taxonomy. Systematic Zoology 20, 117122.
C, J. O. (1984). The kingdom Protista and its 45 phyla.
BioSystems 17, 87126.
C, J. O. (1994). An interim utilitarian ( user friendly )
hierarchical classification and characterization of the protists.
Acta Protozoologica 33, 151.
C, B. F. (1916). Studies on the Anoplura and Mallophaga, being a report upon a collection from the mammals
and birds in the societys gardens. Part 1 with a preface.
Proceedings of the Zoological Society of London 1916 Part 1,
253295.
D, L. A. (1982). Ultrastructure, behavior, and algal
flagellate affinities of the helioflagellate Ciliophrys marina, and
the classification of the helioflagellates (Protista, Actinopoda,
Heliozoea), Journal of Protozoology 29, 1929.
D D, C. (1982). Peroxisomes and related particles in
historical perspective. Annals of the New York Academy of Sciences
386, 14.
D I, N NN (1983). Ultrastucture of sporulation
in Minchinia dentali (Arvy), an haplosporean parasite of
Dentalium entale (Scaphopoda, Mollusca) : taxonomic implications. Protistologica 19, 453460.
DM, P., P, B., D-V, B. A., L,
J. & V, C. R. (1996). Characterization of the highly
divergent U2 homolog in the microsporidian Vairimopha
necatrix. Nucleic Acids Research 24, 515522.
D, A. (1875). Ursprung der Wirbelthiere und das Princip des
Functionwechsels : geneologischen skizzen. Engelmann, Leipzig.
D, W. F. & B, J. R. (1994). Tempo, mode, the
progenote, and the universal root. Proceedings of the National
Academy of Sciences, USA 91, 67216728.
E, P. & M, I. (1965). The relationship between bluegreen algae and bacteria. Biological Reviews 40, 143187.
E, D. R. & D, W. F. (1997). Archaea and the
origin(s) of DNA replication proteins. Cell 89, 995998.
E, T. D., L, J., V, G. S., V, M. H.,
ML, G. L. & K, S. K. (1996). Phylogenetic
analysis of beta tubulin sequences from amitochondrial
protozoa. Molecular Phylogeny and Evolution 5, 359367.
E, J. C., K, J. A., M, H., S, D. V., E,
H. J. & S, M. L. (1988). Ribosomal RNA sequence
shows Pneumocystis carinii to be a member of the fungi. Nature
334, 51922.
E, T. M., H, R. P. & W, D. M. (1995).
Biodiversity at the molecular level : the domains, kingdoms
and phyla of life. In Biodiversity : Measurement and Estimation
(ed. D. E. Hawksworth), pp. 2133. Chapman & Hall,
London.
E G. (1925). Bakterien-Cyclogenie, Berlin.
F, O. V., P, M. G., K, A. S.,
O, N. K. & T, V. Y. (1994). Protofilament as a
structural element of flagella of haloalkalophilic archaebacteria. Canadian Journal of Microbiology 40, 4553.
F, J. (1978). Cases in which parsimony or compatibility methods will be positively misleading. Systematic
Zoology 27, 401410.

T. Cavalier-Smith
F, R. D., A, M. D., W, O., C, R. A.,
K, E. F., K, A. R., B, C. J., T, J. F.,
D, B. A., M, J. M., M, K., S,
G., F, W., F, C., G, J. D., S, J.,
S, R., L, L. I., G, A., K, J. M.,
W, J. F., P, C. A., S, T., H, E.,
C, M. D., U, T. R., H, M. C., N,
D. T., S, D. M., B, R. C., F, L. D.,
F, J. L., F, J. L., G, N. S. M.,
G, C. L., M, L. A., S, K. V., F, C.
M., S, H. O. & V, J. C. (1995). Whole-genome
random sequencing and assembly of Haemophilus influenzae Rd.
Science 269, 496512.
F, P., B-L, N., C, F.,
D, M., E, C. & L, B. (1993). The nature of
the last universal ancestor and the root of the tree of life, still
open questions. BioSystems 28, 1523.
F, C. M., G, J. D., W, O., A, M. D.,
C, R. A., F, R. D., B, C. J.,
K, A. R., S, G., K, J. M., F,
J. L., W, J. F., S, K. V., S, M.,
F, J., N, D., U, T. R., S,
D. M., P, C. A., M, J. M., T, J-F.,
D, R. A., B, K. F., H, P-C., L, T. S.,
P, S. N., S, H. O., H III, C. A. &
V, J. C. (1995). The minimal gene complement of
Mycoplasma genitalium. Science 270, 397403.
F, P. & K, R. M. (1995). Cycliophora is a new
phylum with affinities to Entoprocta and Ectoprocta. Nature
(London) 378, 711714.
G, J. R., K, M., N, D. R. & B, J.
(1996 a). Molecular analysis supports a tardigrade-arthropod
association. Invertebrate Biology 115, 7988.
G, J. R., N, T. J. , N, M. R. & N, S.
A. (1996 b). Molecular evidence for Acanthocephala as a
subtaxon of Rotifera. Journal of Molecular Evolution 43,
287292.
G, A., P, H., & L G, H. (1996). Presence
of a mitochondrial-type 70-kDa heat shock protein in
Trichomonas vaginalis suggests a very early mitochondrial
endosymbiosis in eukaryotes. Proceedings of the National Academy
of Sciences, USA 93, 1461414617.
G, A., P, H. & L G, H. (1997). Evidence
for loss of mitochondria in Microsporidia from a mitochondrial-type HSP70 in Nosema locustae. Molecular and
Biochemical Parasitology 87, 159168.
G, S. P. (1981). The chloroplasts of some algal groups may
have evolved from endosymbiotic eukaryotic algae. Annals
New York Academy of Sciences 36, 1193208.
G R. & M, R. G. E. (1978). Proposal concerning
the higher taxa of bacteria. International Journal of Systematic
Bacteriology 28, 16.
G, S. P. (1981). Chloroplasts of some groups may have
evolved from endosymbiotic eukaryotic algae. Annals of the
New York Academy of Sciences 361, 193207.
G, J. P., K, H., D, P. T, L., B, B.
J., M, M. F., P, R. J., D, T., O, T.,
K, J., D, K. & O, M. (1989). Evolution of
the vacuolar H+-ATPase, implications for the origin of
eukaryotes. Proceedings of the National Academy of Sciences, USA
68, 66616685.

A revised six-kingdom system of life

G, G. A. (1817). Ueber die Entwicklungshefen der Thiere.
Nu$ rnberg, pp. 58.
G, J. & K (1996). Biological Science, sixth ed. Norton,
New York.
G, R. J. (1958). Surface films, their collapse on
compression, the shapes and sizes of cells and the origins of life.
In Surface Phenomena in Chemistry and Biology (ed. J. F. Danielli,
K. G. A. Parkhurst and A. C. Riddifin), pp. 278298.
Pergamon Press, New York.
G, L. E. (1993). Origin of Land Plants. Wiley, New York.
G, K. G., H, A. & S$ , S. (1990). The
ultrastructure of Reticulosphaera socialis Grell (Heterokontophyta). European Journal of Protistology 26, 3754.
H, E. (1866). Generelle Morphologie der Organismen. Reimer,
Berlin.
H, E. (1868). Naturliche SchoX pfungsgeschichte. Reimer,
Berlin.
H, B., S, D., A, C. M., L, L. A. &
L, E. S. (1996). The phylogenetic position of Rhopalura
ophiocomae (Orthonectida) based on 18S ribosomal DNA
sequence analysis. Molecular Biology and Evolution 13, 1187
1191.
H W. R. & D, D. W. (1978). Origin and early
evolution of bilayer membranes. In Light Transducing Membranes : Structure, Function and Evolution (ed. D. W. Deamer), pp.
2359. Academic Press, New York.
H$ , M. M., M$ , S. B., S, V. & M, U.-G.
(1994). How to evolve a complex plastid ? A hypothesis.
Botanica Acta 107, 383386.
H, D. E. (1995). Steps along the way to a
harmonised bionomenclature. Taxon 44, 447456.
H, D. E., K, P. M., S, B. C. & P D.
N. (1995). Ainsworth and Bisbys Dictionary of the Fungi, 8th ed.
CAB International, Wallingford.
H, W. (1966). Phylogenetic Systematics. University of Illinois
Press, Urbana.
H, D. J. (1979). The structure and phylogenetic significance of the flagellar transition region in the chlorophyll
c-containing algae. BioSystems 11, 243261.
H, D. M., H, J. P. & C, C. W.
(1994). Application and accuracy of molecular phylogenies.
Science 264, 671677.
H, G. L, D. D., N, T. A. & S M. L. (1994).
The unusually long small subunit ribosomal RNA of
Phreatamoeba balamuthi. Nucleic Acids Research 22, 465469.
H, H. T., I, T. & O, S. (1982). The phylogenetic
structure of the metabacteria. Zentralblatt fuX r Bakteriologie und
Hygiene [C] 3, 1830.
H, D. S., H, R. P., K, S., L, D. &
E, T. M. (1996). Molecular data suggest an early
acquisition of the mitochondrion endosymbiont. Proceedings of
the Royal Society of London B 263, 10531059.
H, J. S. (1957). The three types of evolutionary process.
Nature, 180, 454455.
H, J. S. (1959). Clades and grades. In Function and
Taxonomic Importance (ed A. J. Cain), pp. 2122. The Systematics Association, London.
H, L. H. (1940). The Invertebrates, Vol. 1. McGraw-Hill,
New York.
H, L. H. (1951). The Invertebrates, Vol. 3. McGraw-Hill,
New York.
H, L. H. (1959). The Invertebrates, Vol. 5. McGraw-Hill,
New York.

263
I, K., C, Y., H, M., O, N. & H, Y.
(1998). The origin of chlorarachniophyte plastids, as inferred
from phylogenetic comparisons of amino acid sequences of
EF-Tu. Journal of Molecular Evolution 45(6) in press.
I, N., K, K., H, M., O, S. & M, T.
(1989). Evolutionary relationship of Archaebacteria, Eubacteria, and eukaryotes inferred from phylogenetic trees of
duplicated genes. Proceedings of the National Academy of Sciences,
USA 86, 93559359.
J, C. (1971). Thallophytes and kingdoms a critique.
Kew Bulletin 25, 2919.
K, T. H., W, H., F, H, S, N. &
Y (1995). Phylogenetic position of the dicyemid
Mesozoa inferred from 18S rDNA sequences. Biological Bulletin
189, 8190.
K, P. & D, W. F. (1996). -Tubulins from early
diverging eukaryotic lineages : divergence and evolution of the
tubulin family. Molecular Biology and Evolution 13, 12971305.
K, P. & D, W. F. (1997). Evidence that
eukaryotic triosephosphate isomerase is of alpha-proteobacterial origin. Proceedings of the National Academy of Sciences,
USA 94, 12701275.
K, P., C, R. L. & D, W. F. (1994).
Archaebacterial genomes : eubacterial form, eukaryotic content. Current Opinion in Genetics and Development 4, 816822.
K, W. S. (18801882). A Manual of the Infusoria, Vols. 1 & 2.
Bogue, London.
K, M. (1983). The Neutral Theory of Molecular Evolution.
Cambridge University Press.
L, J. A. (1988). Origin of the eukaryotic nucleus determined
by rate invariant analysis of rRNA sequences. Nature 331,
184186.
L, J. A. (1989). Origin of eukaryotic nucleus : eukaryotes
and eocytes are genotypically related. Canadian Journal of
Microbiology 35, 109118.
L, J. B. (1810). Philosophie Zoologique. Paris.
L, B. F., B, G., OK, C. J., C, R.,
G, G. B., L C., S, D., T, M. &
G, M. W. (1997). An ancestral mitochondrial DNA
resembling a eubacterial genome in miniature. Nature (London)
387, 493497.
L, E. R. (1877). Notes on the embryology and
classification of the animal kingdom. Quarterly Journal of
Microscopical Science, 17, 395454.
L, E. R. (1880). Degeneration. London.
L, E. R. (1911). Zoology. In Encyclopedia Britannica,
11th ed. Vol. 28, pp. 10221039. Cambridge University Press.
L, J. J. (1990). Phylum Granuloreticulosa (Foraminifera). In
Handbook of Protoctista (ed. L. Margulis, J. O. Corliss, M.
Melkonian and D. J. Chapman), pp. 524548. Jones &
Bartlett, Boston.
L, G. F. (1974). How many are the kingdoms of
organisms ? Taxon 32, 261270.
L, A. (1675). Philosophical Transactions of the
Royal Society of London.
L, A. (1683). Philosophical Transactions of the
Royal Society of London.
L, D. D., W, P. O., G, J. H., P,
D., P, D. J., V, F., H, S. & S,
M. L. (1994). The stramenopiles from a molecular perspective, 16S-like rRNA sequences from Labyrinthuloides minuta
and Cafeteria roenbergensis. Phycologia 33, 369377.

264
L, N. (1675). Cours de Chymie contenant la maniere de faire les
operations qui sont en usage dans la medecine, par une methode facile
avec des raisonnements chaque operation, pour linstruction de ceux qui
veulent sappliquer a cette science. Lemery, Paris.
L, N. D. (1970). Taxonomy of the Sporozoa. Journal of
Parasitology 56 (sec. II,), 208209.
L, R. W. (1988). The Trichomycetes, Fungal Associates
of Arthropods. Springer-Verlag, New York.
L J., K, S. K., H, A., V, G. S. &
E, T. D. (1996). Tubulin genes from AIDS-associated
microsporidia and implications for phylogeny and benzimidazole sensitvity. Molecular and Biochemical Parasitology 78,
289295.
L, J. & D P, R. (1965). Studies on the myxosporidian ultrastructure and polar capsule development.
Protistologica 1, 5365.
L, S. (1985). Phylogenetic aspects of pseudocoelomate
evolution. In The Origins and Relationships of Lower Invertebrates
(ed. S. Conway Morris, J. D. George, R. Gibson and H. M.
Platt). pp. 210233. Clarendon Press, Oxford.
M, L. Y., W, B., D W, R.,
B, T., E, P. & G, J. R. (1996). 18S
rRNA suggests that Entoprocta are protostomes, unrelated to
Ectoprocta. Journal of Molecular Evolution 42, 552559.
M, I. (1965). Some phyletic implications of flagellar
structure in plants. Advances in Botanical Research 2, 134.
M, S. M. (1977). The Arthropoda : Habits, Functional
Morphology and Evolution. Oxford, Clarendon Press.
M, H. J. & T, H. (1994). Haptophytes in polar
waters. In The Biology of the Haptophytes (ed. J. C. Green and
B. Leadbeater), pp. 209228. Clarendon Press, Oxford.
M, L. (1970). Origin of Eukaryotic Cells. Yale University
Press, New Haven.
M, L. (1981). Symbiosis in Cell Evolution. Freeman, San
Francisco.
M, L. & S, K. V. (1988). Five Kingdoms, 2nd
ed. Freeman, New York.
M-S, F. D., K, M. N., V, M., G, J. C. & P, R. A. (1993). The hydrogenosomal
enzyme hydrogenase from the anaerobic fungus Neocallimastix
sp. L2 is recognised by antibodies, directed against the Cterminal microbody protein targeting signal SKL. European
Journal of Cell Biology 61, 8691.
M, K. R. & S, K. D. (1984). Classification of the
green algae : a concept based on comparative cytology. In
Sytematics of the Green Algae (ed. D. E. G. Irvine and D. M.
John), pp. 2972. Academic Press, London.
M S, J. & S, E. (1995). The Major
Transitions in Evolution. Freeman, Oxford.
M, E. (1990). A natural system of organisms. Nature 348,
491.
M, E. & A, P. D. (1991). Principles of Systematic
Zoology, 2nd edition. McGraw Hill, New York.
MF, G. I., W, R., R, M. E. & LU, N. (1997). Plastids in apicomplexan parasites. In
The Origins of Algae and their Plastids (ed. D. Bhattacharya).
Springer-Verlag, Berlin. In press.
M, M., M, B. & S, B. (1995). Phylogeny and
evolution of algae. In Biodiversity and Evolution (ed. R. Arai, M.
Kato and Y. Doi), pp. 153176. The National Science
Museum Foundation, Tokyo.

T. Cavalier-Smith
M, C. (1910). Theorie der zwei Plasmaarten als
Grundlage der Symbiogenesis, einer neuen Lehre von der
Entstehung der Organismen. Biologisches Centralblatt, 30,
278303, 321347, 353367.
M, K. A. & M, A. P. (1995). Fine structure of
an unusual rhizopod, Penardia cometa, containing extrusomes
and kinetosomes. European Journal of Protistology 31, 9096.
M, K. A. & M, A. P. (1996 a). Protist Multicilia marina Cienk. : flagellate or heliozoan ? Transactions of the
Russian Academy of Sciences 346, 136139. (in Russian)
M, K. A. & M, A. P. (1996 b). New information of structure and life cycle of Athalamida Amoebae
(Protista, Athalamida). Zoological Journal 75, 12831293.
M, A. (1995). Biological Systematics : The State of the Art.
Chapman & Hall, London.
M$ , E. (1984). System und Phylogenie der Lebewesen. Band 1.
Schwertzerbartsche Verlag, Stuttgart.
M-L, E. (1996). Fundamentals of the Fungi, 4th ed.
Prentice-Hall, Englewood Cliffs, NJ.
M, L. & M, J-P. (1996). Are Archamoebae true
Archezoa ? The phylogenetic position of Pelomyxa sp. as
inferred from large subunit ribosomal RNA sequencing.
European Journal of Protistology in press.
M$ , M. (1992). Energy metabolism of ancestral eukaryotes,
a hypothesis based on the biochemistry of amitochondriate
parasitic protists. BioSystems 28, 3340.
N, T., S, H., S, M. & S, J. (1995).
Phylogenetic divergence of the entomophthoralean fungi :
evidence from nuclear 18S ribosomal RNA sequences.
Mycologia 87, 203209.
N, N. J. (1783). TraiteU sur la mycitologie ou discours
historique sur les champignons en general, dans lequel on demontre leur
veritable origine et leur geU neU ration ; dou dependent les effets pernicieux
et funestes de ceux que lon mange avec les moyens de les eU viter.
Matthias Fontaine, Mannheim.
N, G. & P, N. (1981). Systematics and Biogeography :
Cladistics and Vicariance. Columbia University Press, New
York.
N, C. (1995). Animal Evolution : Interrelationships of the Living
Phyla. Oxford University Press.
O, G. J. & W, C. R. (1993). Ribosomal RNA, a key to
phylogeny. FASEB Journal 7, 113123.
O, G. J. & W, C. R. & O, R. (1994). The
winds of (evolutionary) change : breathing new life into
microbiology. Journal of Bacteriology 176, 16.
O, C. A. & K, N. K. (1996). Hypothesis : the
emergence of major cellular processes in evolution. FEBS
Letters 399, 119123.
O, R. (1858). Palaeontology. Encyclopedia Britannica (8th
edn) (ed. T. S. Traill), Vol. 17, 91176. Edinburgh.
P, F. C. (1987). The classification of naked amoebae
(Phylum Rhizopoda). Archiv fuX r Protistenkunde 133, 199217.
P, C. F. A. (1960). Diploblastic animals. Proceedings of the
Linnaean Society of London 171, 114.
P, D. J. (1985). The fine structure of Opalina ranarum
(family Opalinidae). Opalinid phylogeny and classification.
Protistologica 21, 413428.
P, D. J. (1986). The actinophryid heliozoa (Sarcodina,
Actinopoda) as chromophytes. In Chrysophytes : Aspects and
Problems (ed. R. A. Andersen and J. Kristiansen), pp. 4967.
Cambridge University Press.

A revised six-kingdom system of life

P, D. J. (1988). The evolution of Protozoa. Memoirs
Institute Oswaldo Cruz, Rio de Janiero. Suppl. 1. Vol. 83,
580600.
P, D. J. (1989). Stramenopiles, chromophytes from a
protistan perspective. In The Chromophyte Algae, Problems and
Perspectives (ed. J. C. Green, B. S. C. Leadbeater and W. L.
Diver), pp. 357379. Clarendon Press, Oxford.
P, D. J. (1994). Protozoa, evolution and systematics.
In Progress in Protozoology (ed. K. Hausmann and N.
Hulsmann), pp. 114. Fischer, Stuttgart.
P, D. J. & F, T. (1985). Insights into the
evolution of heliozoa (Protozoa, Sarcodina) as provided by
ultrastructural studies on a new species of flagellate from the
genus Pteridomonas. Biological Journal of the Linnean Society 34,
381403.
P, J., B, I., G-M, J. & G, M.
(1994). Phylogenetic position of Foraminifera inferred from
LSU rRNA gene sequences. Molecular Biology and Evolution 11,
929938.
P, J., B, I., F, J. F., C-S, T.
& G, M. (1996). Early divergence of Foraminifera
suggested by SSU rRNA gene sequences. Molecular Biology and
Evolution 13, 445450.
P, J. A. M. (1852). Zur Kenntnis der Kleinster Lebensformen.
Bern.
P, L. R. (1960). The Organization of Cells and Other
Organisms. Oxford University Press.
P, B. K. & O, J. M. (1989). Evolution of photosynthesis in anoxygenic photosynthetic procaryotes. In
Microbial Mats, Physiological Ecology of Benthic Microbial
Communities (ed. Y. Cohen and E. Rosenberg), pp. 402427.
American Society for Microbiology, Washington.
R, I. B. (1982). The Protozoan Nucleus, Morphology and
Evolution. Springer, Wien.
R-K, S. (1987). Occurrence of the thraustochytrid,
Corallochytrium limacisporum gen. et sp. nov. in the coral reef
lagoons of the Lakshadweep Islands in the Arabian Sea.
Botanica Marina 30, 8389.
R, M. A. & G, R. R. (1995). Are red algae plants ?
Botanical Journal of the Linnean Society 118, 81105.
R, P. H. (1970). A multiple origin for plastids and
mitochondria. Science 169, 641646.
R, E. (1938). Pogonofora. Fortschritte Zoologische 3, for
1937.
R, R. (1985). The phylogenetic status of the acoelomate
organization within the Bilateria ; a historical perspective. In
The Origins and Relationships of Lower Invertebrates (ed. S.
Conway Morris, J. D. George, R. Gibson and H. M. Platt).
pp. 101122. Clarendon Press, Oxford.
R, A. J. (1996). Studies on the Phylogeny and Gene Structure of
Early Branching Eukaryotes. Ph.D. thesis, Dalhousie University,
Canada.
R, A. J., G C, C., D, W. F. (1996). A
possible mitochondrial gene in the early-branching amitochondriate protist Trichomonas vaginalis. Proceedings of the
National Academy of Sciences, USA 93, 1461814622.
R, S., S, F. & L, S. (1994). Phylogenetic relationships of slime moulds inferred from ribosomal DNA. Abstracts
Fifth International Mycological Congress, Vancouver, 184.
S, N. (1995). Towards a molecular understanding of
developmental mechanisms underlying the origin and evol-

265
ution of chordates. In Biodiversity and Evolution (ed. R. Arai, M.
Kato and Y. Doi), pp. 267290. National Science Museum
Foundation, Tokyo.
S, E. & S, M. (1997). Pirsonia, phagotrophic
nanoflagellates incertae sedis, feeding on marine diatoms :
host recognition, attachment and fine structure. Archiv fuX r
Protistenkunde 147, 361371.
S, M. & S, E. (1997). Light and electron
microscopical observations on Pirsonia punctigerae sp. nov., a
nanoflagellate feeding on the marine centric diatom Thalassiosira punctigera. European Journal of Protistology 33, 168177.
S, J. W. (1993). Microfossils of the early Archean Apex
chert : new evidence of the antiquity of life. Science 260,
640646.
S, J. W. & K, C. (1992). The Proterozoic Biosphere : a
Multidisciplinary Study. Cambridge University Press.
S, A. (1898). A Students Text-book of Zoology. vol. 1.
Sonnenschein, London.
S, C. W., K, H. P., S, R. K., A, G., C,
C. C. Y., L, Q. Y., P, S. L., Y, F., S, M.
E., G, T., D, W. F., R, M. A. &
C, R. L. (1996). Organizational characteristics and
information content of an archaeal genome : 156 kb of
sequence from Sulfolobus solfataricus P2. Molecular Microbiology 22, 175191.
S, A. G. (1966). Basic Arthropodan Stock with Special Reference
to Insects. Pergamon, Oxford.
S, M. E., S, N. A. & B, E. M. (1995).
Molecular phylogenetic evidence that the phylum haplosporidia has an alveolate ancestry. Molecular Biology &
Evolution 12, 573581.
S, M. E., M, D. S. B, D., D, S. S. &
C, D. K. (1995). The demise of a phylum of protists :
Phylogeny of Myxozoa and other parasitic cnidaria. Journal of
Parasitology 81, 961967.
S, J. D., S, M. L., L, D. D. & C, C. G.
(1996). Human parasite finds taxonomic home. Nature 380,
398.
S, A. G. B., H, J., B, C., B, H.
R. & P, D. J. (1997). The ultrastructure and
systematic position of the euglenozoon Postgaardi mariagerensis.
Fenchel et al. Archiv fuX r Protistenkunde 147, 213225.
S, C. (1931). A Short History of Biology. Clarendon Press,
Oxford.
S, J. F., V D, C. D., S, L. H. J. &
S, R. O. (1994). Molecular evidence that the myxozoan
protists are metazoans. Science 265, 17191721.
S, B. J. & G, R. S. (1994). Presence and cellular
distribution of a 60 kDa protein related to mitochondrial
hsp60 in Giardia lamblia. Journal of Parasitology 80, 580590.
S, R. E. (1938). Evolution of the Annelida, Onychophora and Arthropoda. Smithsonian Miscellaneous Collection 97,
1159.
S, E. & W, C. R. (1981). The evolution of
prokaryotes. In Molecular and Cellular Aspects of Microbial
Evolution (ed. M. J. Carlile, J. F. Collins and B. E. B.
Moseley), pp. 131. Cambridge University Press.
S, R. Y. (1961). La place des bacteries dans le monde
vivant. Annales dInstitut Pasteur 101, 297312.
S, R. (1970). Some aspects of the biology of cells and their
possible evolutionary significance. Symposium for the Society of
Genernal Microbiology 20, 138.

266
S, R. Y., D, M. & A, E. A. (1963).
The Microbial World, 2nd ed. Prentice-Hall, Englewood Cliffs,
NJ.
S, R. Y. & V N, C. B. (1962). The concept of a
bacterium. Archiv fuX r Mikrobiologie 42, 1735.
S, J, W. & H, B. D. (1997). The origin of red algae :
implications for plastid evolution. Proceedings of the National
Academy of Sciences, USA 94, 45204525.
S, T. F. (1990). Plant Taxonomy. Columbia University
Press, New York.
S, H. W. (1954). The life history and systematic
relations of the Mesozoa. Quarterly Review of Biology 29,
230244.
S, H. W. (1972). Clarification of taxonomy in Mesozoa.
Systematic Zoology 21, 210214.
S J. & N, H. (1995). The higher Fungi : their
evolutioary relationships and implications for fungal systematics. In Biodiversity and Evolution (ed. R. Arai, M.Kato and Y.
Doi), pp. 177195. National Science Museum Foundation,
Tokyo.
S, J., N, T. & N, H. (1996). Fungal
diversity and phylogeny with emphasis on 18S ribosomal
DNA sequence divergence. In Microbial Diversity in Time and
Space (ed. R. Cowell, U. Simidu and K. Ohwada), pp.
150160. Plenum, New York.
S, D. L., O, G. J., W, & H, D. (1996).
Phylogenetic inference. In Molecular Systematics, 2nd ed. (ed.
D. M. Hillis and C. Moritz), pp. 40751. Sinauer, Sunderland, MA.
T, F. J. R. (1974). Implications and extensions of the
serial endosymbiosis theory of the origin of eukaryotes. Taxon
23, 229258.
T, F. J. R. (1978). Problems in the development of an
explicit hypothetical phylogeny of the lower eukaryotes.
BioSystems 10, 6789.
T, H. A. & L, J. (1993). The ultrastructure of
Commation gen. nov. (stramenopiles incertae sedis), a genus of
heterotrophic nanoplanktonic flagellates from Antarctic
waters. European Journal of Protistology 29, 462477.
T, S. M. (1995). Developayella elegans, nov. gen., nov. spec., a
new type of heterotrophic flagellate from marine plankton.
European Journal of Protistology 31, 2431.
T, S. (1998). Molecular systematics of oxygenic photosynthetic bacteria. In The Origin of Algae and their Plastids (ed.
D. Bhattacharya). Springer-Verlag, Berlin. in press.
U, R. (1995). Evolution and phylogeny of the Metazoa
with special reference to recent advances in molecular
phylogeny. In Biodiversity and Evolution (ed. R. Arai, M. Kato
and Y. Doi), pp. 197228. National Science Museum
Foundation, Tokyo.
V S, C. T. (1845). Lehrbuch der Vergleichende Anatomie der
Wirbellosen Thiere. Heft 1. pp. 1011. Viet Comp., Berlin.

T. Cavalier-Smith
V A, J. A. (1967). Pilzkunde. Cramer, Lehre.
W, X. & L, J. (1996). Fts Z ring : the eubacterial
division apparatus conserved in archaebacteria. Molecular
Microbiology 21, 313319.
W, P. O., H, G., S, M. L. & S, S. K.
(1993). Monophyletic origins of the Metazoa, an evolutionary
link with Fungi. Science 260, 340342.
W, R. (1938). Linterpre! tation des Cnidosporidies et la
valeur taxonomique de leur cnidome. Leur cycle compare! a' la
phase larvaire des Narcome! duses Cuninides. Travaux du Station
Zoologique Wimereux 13, 727744.
W J. M., J P. & W F. R. (1979). From
extracellular to intracellular : the establishment of mitochondria and chloroplasts. Proceedings of the Royal Society of
London Series B 204, 165187.
W, R. H. (1969). New concepts of kingdoms of
organisms. Science 163, 150160.
W, B., B, T., M, L., B, J.
M., D W, R. D., K, S. & G, J. R. (1995 a).
18S rRNA data indicate that Aschelminthes are polyphyletic
in origin and consist of at least three distinct clades. Molecular
Biology and Evolution 12, 11321137.
W, B., B, T. & D W, R. D.
(1995 b). Phylogeny of protostome worms derived from 18S
rRNA sequences. Molecular Biology and Evolution 12, 641649.
W, C. R. (1979). A proposal concerning the origin of life on
the planet earth. Journal of Molecular Evolution 12, 95100.
W, C. R. (1983). The primary lines of descent and the
universal ancestor. In Evolution from Molecules to Men (ed. D.
S. Bendall), pp. 209233. Cambridge University Press.
W, C. R. (1987). Bacterial evolution. Microbiological Reviews
51, 221271.
W, C. R. (1994). There must be a prokaryote somewhere :
microbiologys search for itself. Microbiological Reviews 58, 19.
W, C. R. & F, G. E. (1977 a). Phylogenetic structure of
the prokaryotic domain : the primary kingdoms. Proceedings of
the National Academy of Sciences, USA 74, 50885090.
W, C. R. & F, G. E. (1977 b). The concept of cellular
evolution. Journal of Molecular Evolution 10, 16.
W, C. R., K, O. & W, M. L. (1990). Towards
a natural system of organisms, proposal for the domains
Archaea, Bacteria, and Eucarya. Proceedings of the National
Academy of Sciences, USA 87, 45764579.
W, C. R. S, E. & L, W. (1985). What
are mycoplasmas : the relationship of tempo and mode in
bacterial evolution. Proceedings of the National Academy of
Sciences, USA 21, 305316.
W, C. G., L, M. R., D, R., L, J. J. & L,
J. H. (1995). Origin of the foraminifera. Proceedings of the
National Academy of Science, USA 92, 141145.