Rasnitsyn, 2008. Theoretical Foundations of Evolutionary Biology - 03
1.2.2. TAXONOMY. 1.2.2.1. Cladistics. 1.2.2.2. Phenetics. 1.2.2.3. Phyletics
1.2. METHODOLOGY OF PHYLOGENETICS, TAXONOMY, AND NOMENCLATURE 1.2.1. PHYLOGENETICS 1.2.1.1. Group Analysis 1.2.1.2. Trait Analysis 1.2.1.2.1. Difference Analysis 1.2.1.2.2. Similarity Analysis 1.2.1.3. Computer Cladistics
A.P.Rasnitsyn. Theoretical Foundations of Evolutionary Biology // V.V.Zherikhin, A.H.Ponomarenko, A.P.Rasnitsyn. Introduction to Paleoentomology. M.: KMK. 2008. 371 pp. 1.2.2. TAXONOMY 1.2.2.1. Cladistics 1.2.2.2. Phenetics 1.2.2.3. Phylistics
1.2.2. TAXONOMY Taxonomic classification organises biological diversity in a manner that facilitates our interaction with it. There exist many forms of classification created for specific purposes and specific categories of users (for example, a system of life forms, or a system of pests according to the nature of the damage they cause). These present no significant theoretical problems, and the discussion here will concern only that system which serves as an interdisciplinary language. We shall call it the general system of organisms. Taxa of the general system, in order to fulfil their functions, must constitute meaningful groupings from the standpoint of the broadest possible range of users, just as a bird represents a perfectly definite concept not only for the ornithologist, but also for the hunter, the cook, and the artist. To meet such a goal, the taxa of the general system must be maximally homogeneous within themselves and maximally distinct from one another across the broadest possible range of characters. This property enables the system to fulfil two further most important functions. First, it compresses information about living organisms into an accessible form. Recall that each of us knows, say, something about a cockroach, including information about the characters of higher taxa. Second, it allows the prediction of properties of insufficiently studied taxa. Only a negligible fraction of insect species has been studied in any detail, yet we have no doubt that they consist of eukaryotic cells of the corresponding structure and function, and we describe with sufficient confidence many details of the biology of even extinct species. For example, the discovery in Lower Cretaceous deposits of Transbaikalia of an incomplete forewing identified as that of a wheat stem sawfly (Cephidae) served as a basis for predicting that angiosperms, on which modern wheat stem sawflies exclusively develop, had already existed at the beginning of the Cretaceous (Rasnitsyn, 1969). And angiosperms were indeed found in the same Baisa locality in the very year when the prediction was published (Vakhrameev and Kotova, 1977). In the systematics of recent decades, three principal schools can be distinguished (Table 3): phenetics, cladistics, and yet another, often unfortunately termed "evolutionary systematics" — in contrast to "phylogenetic systematics," the original and equally unfortunate designation of cladistics (indeed, how do the terms "evolutionary" and "phylogenetic" differ in this context?). For this school, which attempts to identify and formulate the principles of traditional systematics, the term "phyletics" was proposed (Ponomarenko and Rasnitsyn, 1971). The differences between the three schools are far more profound than one might suppose: they correspond to entirely different conceptions of the structure and origin of biological diversity. One of the listed schools is compatible with the synthetic, or more precisely, the population-genetic theory of evolution. The latter, as we have seen, represents the evolutionary process as the dynamics of alleles in populations, controlled by selection and stochastic processes. Selection shapes the composition of populations and the properties of organisms, doing so with a precision limited only by stochastic factors and gene exchange, which unifies population structure at the level corresponding not to local but to averaged characteristics of selection. The organisation of living beings appears here as a set of characters freely shuffled by selection, and the evolutionary process as uniform to the degree that environmentally determined variations in selection are uniform. This uniformity is disrupted only by divergences and extinctions, which alone can be employed for constructing a natural system of organisms. That is, the natural system answering to the population-genetic theory of evolution can only be strictly genealogical, with the boundaries of taxa drawn at the points of divergence (nodes of the cladogram). Taxa are delimited in accordance with reconstructed facts of their history, not their present properties, which are important only as historical evidence. This is a precise description of the cladistic system: strictly hierarchical and excluding paraphyletic, i.e. ancestral, taxa (Table 1). Cladistics reduces the system of a group to its genealogical history, phenetics to the characters of similarity accessible to analysis, while traditional systematics (viewed through the lens of phyletics) attempts to use both, not arbitrarily, but in an ordered manner (Table 3). Table 3. The Three Taxonomies
Type
Evolutionary Basis
Setting process
Working criterion
Similarity
Blood relationship
Cladism
synthetic
act of divergence
synapomorphy
ignored
single basis of the system
Phenetics
epigenetic
discreteness of biodiversity
giatus
single basis of the system
ignored
Philetica
epigenetic
discreteness of biodiversity
giatus
assigned criterion
controlling criterion
Unlike the synthetic theory, the epigenetic theory of evolution, as we also know, views the evolutionary process primarily as a process of evolutionary transformation of ontogenesis and, in particular, of established (selected over generations) epigenetic processes and their stabilized ensembles – creodes. Conversely, mutations and alleles act here as switches between alternative developmental trajectories, rather than as creators of structures, properties, and traits. The deep interdependence between ontogenetic processes and, consequently, between properties and traits makes the optimization of the whole organism an extremely complex task, a product of strenuous compromise. A successfully achieved compromise is found to be firmly stabilized (buffered), as a successful organismal structure is very difficult to change beyond normal intraspecific variability (which is itself already stabilized by selection of previous generations). Therefore, evolution (successful exit beyond the stabilized epigenotype) is complicated, unpredictable, and more or less jumpy, and the resulting biodiversity is more or less discrete. Discreteness is exploited by both phenetics and traditional systematics, which build their systems by identifying and tracing hiatuses – breaks in the continuity of biodiversity. But let's take things in order. 1.2.2.1. Cladistics The deep understanding of cladistics by its various followers differs so much (see Hennig, 1966a and Platnick, 1979) that it has been proposed, for example, to use the old name 'phylogenetic systematics' for Hennig's version, and leave cladistics for the so-called transformed cladistics (de Queiroz, Donoghue, 1990a). This proposal did not gain wide support, and I retain a broad understanding of cladistics here. However, the central assertion of cladistics, that 'phenetic classification is built by grouping by overall similarity, whereas phylogenetic classification is built by grouping by inferred synapomorphies' (Farris, 1979: 478), remains generally accepted. Justifying it, W. Hennig (Hennig, 1966a) wrote that taxa designated by synapomorphies form a system isomorphic to phylogeny and therefore the most effective. Hence the name of the system – phylogenetic. But there is another point of view, that, firstly, the very structure (pattern) of nature is such that the system of organisms can be represented by a branching diagram (hierarchical classification). Secondly, this pattern can be reflected as a result of studying traits and searching for a hierarchical (nested) set of synapomorphies. Thirdly, our knowledge of evolution, as well as of the system, is derived from the same hierarchically organized set (pattern) of synapomorphies (Platnick, 1979: 538). Thus, the pattern is recognized as primary, and phylogeny as secondary. This approach raises the question: who brings the pattern into nature? what is its generating process, if not evolution? As long as no answer is proposed, I see no possibility of a separate analysis of pattern cladistics (transformed cladistics) and will return to its classical version. The first question that arises concerns what exactly is meant by the central position of cladistics (synapomorphy as the sole basis for a taxon). Synapomorphy is usually defined as a character state acquired by the ancestral species of a group and therefore characteristic of all its members and only them. This definition is imprecise, as any acquired character state can later be lost, which is considered impossible for apomorphy (snakes lost legs, but not quadrupedality as a synapomorphy of tetrapods). Apomorphy is not a character state, but the fact that such a state was once acquired (a fact of the group's history). The question of groups lacking apomorphies deserves special attention. Such groups are called paraphyletic or polyphyletic, depending on whether they are distinguished by symplesiomorphy or homoplasy, respectively. The cladistic solution is to break down such groups into smaller ones so that they all become monophyletic (more precisely, holophyletic, see below), i.e., characterized by apomorphies. However, this creates the problem of paraphyletic taxa, which cannot be divided into holophyletic components (metataxa according to de Queiroz, Donoghue, 1990b). These are many extinct and some extant groups, not all of whose necessary traits can be studied. In most cases, this problem could be considered temporary and attributed to incomplete knowledge, which will be overcome. However, this refuge of ignorance does not save, if only because parallel evolution, as we have seen, leaves us no hope of deciphering the cladogram in all its details. Even more significant is that the mandatory division of a paraphyletic taxon into holophyletic subgroups will inevitably lead us to the ancestral species, which by definition is a paraphyletic taxon and which cannot be divided into holophyletic components, as it is a single species. Some researchers resort to the refuge of ignorance here too, arguing that the ancestral character of a species can never be verified (Nelson, 1973a). However, as Popper showed, scientific hypotheses in principle cannot and should not be verified. On the other hand, if each branch of the phylogenetic tree has an ancestral species at its base, then a simple calculation shows that such species must constitute half of all species that have ever existed. Indeed, in the case of strict dichotomy of branching and the absence of phyletic speciation, the number of ancestral species equals the number of terminal species minus one. Multiple branching (bushy, polytomy) somewhat reduces the proportion of ancestral species, phyletic evolution increases their number, but as a first approximation, it must be admitted that half of all described fossil species (and probably a significant part of extant ones) are direct ancestors of other species. A species ancestral to more than one taxon presents an insurmountable theoretical problem for cladistics, unless its central tenet (a taxon is defined exclusively by apomorphy) is softened. This necessary step was already taken by Hennig, who believed that apomorphy legitimizes a taxon not by definition, but only as a consequence of another, stricter (basic) assertion about the isomorphism between the system and phylogeny. 'The duration of a species is determined by two speciation events: the one to which the species owes its origin as an independent reproductive community, and the one that splits it into two or more such communities' (Hennig, 1966: 66). This line of reasoning makes divergence the sole process that creates taxa, and apomorphy turns out to be simply a marker of divergence and, consequently, of a taxon. Another consequence will be the extinction of the ancestral species as a result of divergence – even if one of the products of divergence shows no differences from the ancestral species (lacks apomorphies). This statement seems somewhat strange, but at least it is consistent. Thus, in the cladistic paradigm, some (perhaps many) species do not differ from their ancestors. Then we do not even have a theoretical possibility to accurately reconstruct the topology of the cladogram: only some approximation to it is possible. This circumstance reduces the advantages of the approach and opens up other possibilities, for example, to invert the central cladistic assertion and consider the cladistic system ultimately definable by a sequence of apomorphies. Then divergence events can be considered as a tool for defining this sequence. Such an approach was used by D. Brothers (Brothers, 1975) in his classic study of the phylogeny and systematics of predatory migratory spiders. G.D. Nelson likely holds a similar position, arguing that 'origin without change... is not a sufficient explanation for a taxon' (Nelson, 1989: 280-281). Just as divergence is not always accompanied by apomorphic change in all its products, so the acquisition of apomorphy in the case of phyletic evolution is not necessarily associated with divergence. Thus, establishing the exact sequence of apomorphies is unattainable even in theory. Therefore, the two approaches are at least comparable in their significance for phylogenetic research – not in the sense that they are necessarily equivalent, but in that their advantages and disadvantages equally deserve special study. For example, in Brothers' version, apomorphies are hardly accessible not only for precise sequencing (ordering in a strict sequence), but even for simple counting. In most cases, apomorphies are too complex to be represented as occurring in one step (during one speciation event). But the sequence of elementary steps in acquiring a complex apomorphy must correspond to the hierarchical system of taxa, which in principle are not distinguished. Other, more serious problems of cladistics will be discussed in comparison with the advantages and disadvantages of phenetics and phyletics after their more detailed consideration. Thus, cladistics asserts that the system of organisms must be strictly genealogical, i.e., reflect the blood relationship of organisms as accurately as possible. For this, taxa are divided at divergence nodes using synapomorphies, i.e., changes that arose in the ancestor and are found in its descendants. Similarity is interesting only as evidence of genealogy, and is not significant in itself. A cladistic taxon cannot give rise to another taxon: in the cladistic world, there are no ancestor taxa and descendant taxa. Of course, except for species, but since identifying an ancestral species is impossible according to cladists, the ancestor is not a real being or a real species, but simply a collection of primitive traits (plesiomorphies) found in its descendants. There is indeed reason in this, as the widespread occurrence of parallelisms makes discussion of an ancestral species in most cases irrelevant, but this also makes illusory the main assertion of cladistics about a single divergence event (or, in its other version, about the acquisition of a single apomorphy) as the sole event that gives life to a taxon. 1.2.2.2. Phenetics Unlike cladistics, phenetics and traditional systematics (or, in a refined form, phyletics) start from the integrity of living systems and, accordingly, from the discreteness of biodiversity and are engaged in searching for hiatuses. Therefore, the fundamental concept must include a process that generates such discreteness. Initially, discreteness was easily explained by creationism; now, only the epigenetic theory of evolution (see above) seems to claim this role. Of the three competing approaches, phenetics is the most straightforward in its goals and means. Since the discreteness of biodiversity is described in terms of similarity, not kinship, the task of systematics, according to pheneticists, should be formulated in the same terms, namely: the system should be explicitly built in terms of similarity. Subjectivity can be avoided by calculating similarity. The calculation methods are collectively called taxometrics or, more often, though less felicitously, numerical taxonomy. Unfortunately, these methods have not been developed to the stated level – in my opinion, not because this level is unattainable, but because phenetics was supplanted by cladistics before it completed its useful work. Indeed, taxometric methods are numerous and diverse, and choosing a method adequate to the task is not simple and at the current level is barely possible without a significant degree of subjectivity. However, this is a problem for any approach, and cladistics is no exception (see, for example, Mickevich, 1978): the optimal area of application of a specific method, its strengths and weaknesses can only be identified through costly efforts. Before yielding the battlefield to cladistics, phenetics made significant, but insufficient, progress in this direction. However, the fundamental problem of phenetics, in my opinion, is different. The similarity it works with is not quite the similarity required by the system of organisms. Pheneticists work with studied traits, the number of which should be reasonably large, but not more. The system, to optimally perform its functions, must rely on similarities and differences in all traits, studied and unstudied, including those that may never be studied. Of course, such similarity is not available for direct study, but it is important enough to try to estimate it indirectly. This is precisely the goal pursued by phyletics.