¿ìè¶ÌÊÓƵ

How should the world’s living organisms be classified? Into how many kingdoms should they be grouped? ¿ìè¶ÌÊÓƵs have been grappling with these questions since the time of Aristotle, drawing on a broad base of biological characteristics for clues. The fossil record, visible traits of living organisms and, more recently, results from cell biology have all shaped theories of biological classification. But last year a new and controversial concept emerged: a classification of life based solely on molecular traits.

The focal point of the controversy is a tree of life, or ‘phylogeny’, devised by Carl Woese of the University of Illinois, Otto Kandler of the University of Munich and Mark Wheelis of the University of California. The tree is unusual because, unlike all previous schemes, it is constructed solely from biochemical data such as DNA sequences rather than a range of different organism characteristics. But that is not all. The scheme also challenges the idea that life on Earth is best divided into five kingdoms, with the main split being between bacteria and all other organisms. Woese and his colleagues create three main groupings by dividing the bacteria in two and unifying all other organisms.

Does Woese’s scheme help us understand how organisms are related or does it obscure these relationships? To answer this, one needs to know something about the aims of biological classification.

Many billions of types of organisms have lived at one time or another on the Earth, of which at least 30 million survive today (though no one knows exactly how many). Classifying such a bewildering diversity is no simple matter. Generations of natural historians have struggled with the task since the time of Linnaeus, the 18th-century Swedish biologist who classified as many as 10 000 living things. At first the aim was to identify, name and group organisms according to their visible characteristics, a practice which grew into the discipline of taxonomy. After Darwin, though, classification acquired a new dimension. Biologists drew ‘family trees’ or ‘phylogenies’, with the aim of grouping organisms according to their evolutionary histories.

Taxonomies such as those of Linnaeus are primarily information retrieval systems. Like a good library catalogue, their purpose is to convey as much as possible about a given organism-its genetic make-up, structure, physiology, behaviour and ancestry-by its place in the scheme. Taxonomies can be purely utilitarian or even arbitrary, such as the classification of flowers by colour. Phylogenies are quite different; their purpose is to depict evolutionary lines of descent.

Phylogenetic schemes attempt to reconstruct the history of life on Earth, a history stretching back 3400 million years. To devise them, natural historians decipher clues from both living species and the fossil record. The search for clues has traditionally embraced as much information about living organisms as possible: their appearance, anatomical organisation, development, mode of nutrition, metabolic pathways, gas emissions, pigments, the sequence of subunits making up their DNA, RNA and protein, behavioural interactions and fossil history.

But is such a vast amount of information really needed? Could biologists get by using just one key characteristic? In our opinion, the meaningfulness of any phylogenetic tree as a guide to evolutionary history depends critically on what, and how many, characteristics were used to construct it. In general, the more the better. Woese and his group, however, argue that molecular structures render all other organism characteristics, including visible traits, redundant.

Using the genetic sequence for a single type of RNA molecule, they have constructed a phylogenetic scheme in which all organisms are divided into three main groups rather than five. If it is correct, the scheme throws up a startling possibility. It implies that the world’s two main groups of bacteria are, in evolutionary terms, more distinct from each other than either is from all other life on Earth (¿ìè¶ÌÊÓƵ, Science, 11 August 1990).

But just how good a measure of evolutionary origin is the sequence of any gene or molecule taken alone? Here we are sceptical as criteria from genetics, cell biology and developmental biology still argue powerfully for a scheme with five kingdoms: Animalia, Plantae, Fungi, Protoctista (a kingdom comprising traditional protozoa, all algae, slime moulds and their allies) and Procaryotae (bacteria). An earlier version of this biological classification was first proposed in 1959 by Robert Whittaker of Cornell University, Ithaca.

A key aspect of the ‘five kingdom’ classification is that it highlights possession of more than a single genetic system, such as a cell nucleus, as an organism’s most striking characteristic. Its main division falls between eukaryotes, organisms with at least two genetic systems of different ancestral origins and prokaryotes, whose cells are single genetic systems. Eukaryotes, which include all animals, plants, fungi and protoctists, possess a membrane-bound nucleus as one of their genetic systems. Prokaryotes, which are all bacteria, do not.

In the revision of Whittaker’s original five-kingdom classification, published in 1988 by Karlene Schwartz and myself, we recognise the existence of two basic kinds of prokaryotes, archaeobacteria and eubacteria. Crucially, though, we regard them as members of a single kingdom. Here we differ radically from Woese and his group. For they argue that the microbes denoted as ‘archaea’ in their scheme are more distinct from eubacteria than they are from the ancestors of all other organisms.

In Woese’s classification, archaea and eubacteria acquire the status of superkingdoms-or ‘domains’. Their third domain comprises all eukaryotes, including plants, animals and other kingdoms. We disagree with this division of bacteria. In both external appearance and internal biological organisation, archaeobacteria and eubacteria resemble one another far more than they resemble any other type of organism. The standard reference volumes of bacteriology (Bergey’s Manual and The Prokaryotes) accept Procaryotae as a taxonomic group covering all bacteria.

History from genes

The molecular approach to phylogeny involves selecting a gene (or its RNA or protein product) in one organism and then examining the corresponding, or ‘homologous’, gene in a series of others. By comparing the chemical sequences of genes, researchers endeavour to trace a family tree showing how the genes have diverged during evolution. The precise position of an organism in the tree is fixed by the extent to which its gene sequence differs from those of other organisms. The underlying philosophy is that an organism’s evolutionary history is recorded much more clearly in its genetic sequences than anywhere else. In our view, however, the approach has serious limitations.

Like any other procedure, the molecular approach rests on a series of assumptions. First, it assumes that homologous genes-that is, genes that produce RNA molecules or proteins whose functions have not diverged in different organisms-can be unambiguously identified. A comparison of non-homologous genes will obviously give a spurious result. Secondly it assumes that the branches of the phylogenetic tree only diverge with time and never reunite. In other words, as two organisms become increasingly distinct, the sequences of all pairs of homologous genes in the two organisms drift further and further apart and the trend never reverses. Finally, it assumes that the major, or indeed only, source of evolutionary innovation is the gradual accumulation of gene mutations. All three propositions are questionable.

The molecular trait used in many analyses comes from the genes of ribosomal RNA. Ribosomes are the sites in cells where proteins are synthesised; they contain several strands of RNA bundled together with protein molecules. Woese and his team chose a stretch of one type of ribosomal RNA, the 16S RNA, and examined the sequences of its genes in different organisms.

A number of key properties of such sequences make them extremely useful as evolutionary markers. Not only is 16S RNA present in all organisms but it functions in the same way in protein synthesis in all organisms. Its sequences differ only in their finer features. For some 3400 million years, it seems, the sequences have been sheltered from the more violent effects of genetic reorganisation. This greatly increases the chances of identifying them correctly in different organisms.

The more subtle variation seen in 16S RNA sequences may be directly related to selective pressures, the way cells acquire symbionts or other differences in the ancestries of whole organisms. Yet interpreting the variation is far from straightforward. For dramatic evolutionary divergence might occur with only small accompanying changes in sequence, while minor evolutionary divergence might lead to large changes in molecular sequence.

Another worry is whether the sequences studied by molecular biologists are truly representative. Is it safe to assume that the molecules of a single member of a species yield information typical of the species? This is an especially vexing problem if the chosen sequence varies between different tissues of an organism, or different stages of its life cycle. And is it reasonable to assume that an analysis based on 16S RNA will be entirely consistent with analyses based on other molecules, such as 23S RNA?

Eukaryotes, which all carry two or more different genomes, each from a different ancestor, pose a particular problem: which genome should be probed for molecular traits? The usual assumption is that the genome found in the nucleus is intrinsically superior to all others for reconstructing the evolutionary history of the whole organism. Yet neither this nor any of the other assumptions can be fully verified.

In our modification of Whittaker’s ‘five kingdoms’ scheme, we incorporate the important results of molecular biologists, but also draw on the efforts of many other scientists. Our aim was to create the most useful, comprehensive and up-to-date way of handling the mass of data assembled by countless investigators over at least three centuries of biological science. Our scheme, like Whittaker’s, recognises two ‘superkingdoms’: the Procaryotae and Eucaryotae. All organisms in the Procaryotae (the bacteria) have one kind of genome. They consist of small-ribosome cells or are microbial multicellular organisms, each cell of which contains all the chemical components required for self-maintenance, growth and reproduction. Their hallmark is their distinctive molecular biology.

From a morphological and developmental point of view prokaryotes are simple, yet as a group they are by far the most diverse. The internal components of a prokaryotic cell experience only Brownian motion-passive movements caused by collisions with surrounding molecules-though some prokaryotes propel themselves by mysterious gliding or with the help of distinctive rotating flagella.

The prokaryotic domain in our scheme comprises just one kingdom, called Procaryotae or Monera, and its members are all the bacteria. It is divided into at least two subkingdoms, the archaeobacteria and eubacteria. Woese and his group showed that these can be distinguished by the structure of a small hairpin loop in 16S ribosomal RNA. In archaeobacteria the loop carries a side bulge made up of seven nucleotides; in eubacteria the bulge has six nucleotides. These differences-which correlate with differences in lipids, amino-acid sequences of some enzymes and cell wall structure-led Woese and his team to divide the bacterial kingdom.

Bacteria branching out

Most well-known bacteria are eubacteria. The archaeobacteria group has some particularly exotic members, though, such as the methanogens, which generate methane in sediments and the human gut, and a group of salt-loving organisms called halobacteria. Although the different types cannot be distinguished in the fossil record, the history of the Procaryotae extends from 3400 million years ago to the present day.

All members of our second superkingdom, Eukaryotae, are eukaryotes with more than one chromosome per nucleus and more than a single membrane-bounded organelle per cell. Eukaryotes originated from the symbiosis of different prokaryotic microbes, in which formerly free-living bacteria became discrete organelles working within the boundaries of cells. Because each type of bacteria came in with its own DNA, today’s eukaryotes carry more than one type of genome. The presence of multiple genomes in eukaryotic cells is a legacy of multiple prokaryotic ancestors.

Because of their prokaryotic ancestry, eukaryotic cells have at least two types of ribosome and the associated machinery for making proteins. The DNA of the cells’ chromosomes is surrounded by a membrane which cordons off the nucleus from the cytoplasm. A key difference between eukaryotes and prokaryotes is that inside all eukaryotic cells, Brownian motion is supplemented by directed movements of organelles caused by microtubules and other specialised motility proteins.

Within the eukaryotic domain of our scheme are three higher kingdoms. In order of appearance in the fossil record these are: Animalia, Fungi and Plantae . All larger eukaryotes can be classified unambiguously as animals, plants or fungi. Yet outside these three groupings there remain approximately 250 000 species of living things with nuclei which do not display animal, plant or fungal qualities. To accommodate them, biologists introduced another kingdom, the Protoctista-a term coined in 1861 by the Scotish naturalist John Hogg for ‘organisms neither animal nor plant’. The Protoctista includes protists, which have only a few cells or just one, as well as their much larger descendants.

In some schemes, the fungi do not enjoy the status of a separate kingdom, but instead find a niche in the Protoctista. The upshot is a division of the natural world into four kingdoms, as advocated in 1956 by Herbert Copeland, a biologist working at Harvard University. Because the fungi are a diverse group of 100 000 species that have a major ecological role as decomposers, especially on land, Whittaker raised their status to create a fifth kingdom. Copeland’s four kingdoms are acceptable to us, but on balance, we think that the fungi deserve the status of kingdom.

Blurred division

Researchers cannot resolve all debates about classification by recourse to evolutionary history. Even if we could all agree on phylogeny, numerous classification schemes could still be defended, all entirely consistent with it.

We agree with Woese and his colleagues over the evolutionary history of the prokaryotes but disagree about their classification. Specifically, we think Woese is wrong to split the prokaryotes-all of which have a single genome and a characteristic internal structure-into two superkingdoms, while simultaneously uniting all their descendants carrying more than a single genome into a third superkingdom. If Woese’s scheme were accepted, it would complicate the task of storing and retrieving biological information. It would also obscure the fundamental division of living things into organisms with a single ancestry and organisms with multiple genomes evolving by symbiosis. This division was first highlighted by the Russian biologist Konstantin Mereschkovsky, who in 1909 noted that there was no ‘middle ground’ between organisms with symbionts and organisms without. If an organism has a permanent symbiont such as a chloroplast, he argued, it belongs to the plant kingdom, but if it does not then it belongs to another kingdom.

In recent years molecular biology has become central to all modern taxonomy. It has provided new insights which will surely result in changes to the classification of certain groups of organisms, especially among the most diverse eukaryotes, the protoctists. But an awkward question remains: why should certain sets of molecular sequences take precedence over other sets? There is no reason, for instance, why ribosomal gene sequences from the nucleus of a plant cell should be considered any more representative of the organism’s evolutionary history than those associated with, say, its plastid genome.

Indeed the genomes of organelles can provide important clues. Plastids, for example, carry a gene whose product is the enzyme ribulose bisphosphocarboxylase and whose sequence reveals an unambiguous link between the organelle and certain bacteria. Links such as this underscore the crucial point: that eukaryotes are best likened to tightly integrated communities of different prokaryotes, not to uniform prokaryotes that simply grew large. The plant cell has at least three genomes and we need sets of sequences from all three in order to characterise its phylogeny. To force a choice between their ribosomal RNAs is irrational.

Finally, although we disagree with Woese and his colleagues on the prokaryote split, we do concur with their rejection of attempts to classify all living things as either plants or animals. These simplistic dichotomies are the most serious impediment to understanding. They are prejudiced and should be abandoned, not least because they underplay the importance of two of the richest strands of life on Earth.

Lynn Margulis is Distinguished University Professor of Botany at the University of Massachusetts, Amherst. Ricardo Guerrero is Professor of Microbiology at the University of Barcelona, Spain.

Further reading: Five Kingdoms: An Illustrated Guide to the Phyla on Earth, by L. Margulis and K. V. Schwartz, W. H. Freeman; Handbook of Protoctista, edited by L. Margulis, J. O. Corliss, M. Melkonian and DJ Chapman, Jones and Bartlett, 1990; The Fifth Kingdom, B. Kendrick, Waterloo.

* * *

Animal, vegetable or slime mould?

Animalia: All animals develop from balls of cells, or ‘blastula embryos’, and are thus multicellular. Their cells have at least two genomes derived from microbes. One is associated with the nucleocytoplasm, the other with mitochondria, energy-producing organelles. Their life cycles take a characteristic form. As diploid organisms-that is, organisms carrying two sets of chromosomes in the nucleus-they produce sex cells by the process of meiosis. Sex cells carry only one set of chromosomes each and are of two kinds, egg and sperm, which differ in shape and size. Embryonic development, marked by mitotic cell division, follows fertilisation.

Animal cells with undulipodia (the general term for moving organelles, such as cilia, made of nine pairs of microtubules) are incapable of cell division. The organisation of animal DNA into chromosomes allows researchers to define a karyotype (a description of the number and shape of the chromosomes) for each species, based on detailed examination of cells under the microscope. The fossil record of the Animalia extends from the late Proterozoic Eon to the present-a span of 700 million years.

Plantae: All plants develop from embryos and so are multicellular. Each plant cell has at least three microbial genomes, associated with nucleocytoplasm, mitochondria and plastids. In most plants plastids turn green to form chloroplasts, organelles which harness light energy. Plant life cycles feature two distinct phases and two periods of development. One phase culminates in the production of spores. These germinate and develop into small plants, ushering in the second phase, during which small male or female plants produce sex cells. Fertilisation results in the embryo being lodged in the female plant, whose growth and development completes the cycle. Undulipodia are restricted to certain male sperm. Plant chromosomes, like those of animals, can be observed and counted under the microscope. The fossil record of plants extends from the late Silurian Period (more than 400 million years ago). The major evolutionary changes establishing plant life on land occurred during the subsequent Devonian Period.

Fungi: Fungal cells have at least two genomes of microbial origins, one in their nucleocytoplasm and the other in their mitochondria. They develop from spores with only one set of chromosomes in the nucleus (the haploid condition), and form spherical or filament-shaped cells with walls of chitin. Sex, when it occurs, takes the form of conjugation, which results in cells with two nuclei. When the nuclei eventually fuse, spores are formed by the process of meiosis. But fungi also make spores without going through any sexual process. Either way, the spores are more resistant than growing tissues to drying and heat. Fungal cells never possess undulipodia. The fossil record of the group extends, like that of the plants, from the late Silurian or early Devonian to the present. Yeasts are single-celled fungi; other fungi are multicellular.

Protoctists: If an organism is neither animal nor plant nor fungus, yet has more than one membrane-bound genome, it is a protoctist. Protoctists that acquire ready-made food, such as ciliates, slime nets, slime moulds and tissue parasites, have at least two genomes of microbial origin, associated with nucleocytoplasm and mitochondria. Those that make their own food by harnessing light energy, such as algae, contain at least three-based in nucleocytoplasm, mitochondria and plastids.

Protoctists are distinguished from bacteria by the organisation of their DNA into chromosomes bound inside nuclei. Protoctists are in fact defined as nucleated microorganisms and their descendants, and include some large organisms such as seaweeds. There is much variation in sexual, genetic, developmental and molecular characteristics within the group. The number of chromosomes varies from 2 to about 16 000. Their fossil record extends from the middle Proterozoic Eon, about 1300 million years ago, to the present.