I.4. Reconstructing phylogeny

I.4.1 Analysis of phylogeny based on synapomorphies

I.4.1.1 Terms «apomorphy» and «plesiomorphy»

I.4.1.2 Logic of phylogenetic based analysis based on synapomorphies

I.4.1.3 Interdependency of phylogenetic theory and character definition

I.4.1.4 Field of application of the phylogenetic analysis

I.4.1.5 Phylogenetic analysis and evolutionary scenario

I.4.1.6 Search for apomorphies

I.4.2 Misconceptions on phylogeny analysis

I.4.2.1 Fallacy of embryological method

I.4.2.2 Restriction of paleontological method

I.4.2.3 Non-obligation of structure complication

I.4.2.4 Non-obligation of oligomerization

I.4.2.5 Non-obligation of ancestor non-specialization

I.4.2.6 Why do numerical and matrix methods fail

I.4.2.6.1 Unavailability of parsimony principle

I.4.2.6.2 Incorrectness of data presentation in a matrix form

I.4.2.6.3 Fundamental shortcoming of numerical taxonomy

I.4.2.7 Substitution of molecular method of phylogeny analysis

  I.5. Principles of classification of supraspecies taxa

  I.6. Principles of nomenclature of zoological taxa

I.4. Reconstructing phylogeny

The word phylogeny or phylogenesis (from Greek φυλον – tribe, and γενεσις – origin) was introduced by E. Haeckel (1866) and means the same that Ch. Darwin termed genalogy. 

The relationship of different taxa (or phyla) can be graphically depicted as a scheme called a phylogenetic tree, or a cladogram. The phylogenetic tree is a graph of branching lines, where each line means the existence of a biological species in time with a common gene pool and branches out at the points of divergence, where  reproductive isolation appears (i.e. isolation between gene pools); the length and direction of the lines are either arbitrary, or are tied to a particular coordinate system (one or another axis can reflect either geological time, or accumulation of differences, or other parameters). Often, more generalized phylogenetic schemes are drawn, where each line means not a separate species, but a phylogenetic branch consisting of many species; this implies that all the species branches are indistinguishable on a given scale and therefore can be depicted merged, that is, in the form of an integral strip or line. In any case, each branching shown on the phylogenetic tree means a divergence – that is, the moment when one biological species is divided into two species, reproductively isolated from each other. It should be cautioned against over-passion for the analogy between the phylogenetic scheme and the tree: the phylogenetic scheme is similar to a tree only in that it branches; in contrast to a real tree, a phylogenetic tree has neither a root, nor a central trunk, nor thickened bases of the branches.

The reconstruction of phylogeny consists in correctly identifying the sequence of branchings of the phylogenetic tree.

I.4.1. analysis OF PHYLOGENY BASED ON SynapomorphIES

The conclusions about these or that phylogenetic connections and a general conclusion about origin of diverse animals from a common ancestor, are made on the base of the fact that diverse animals have common characters whose occurrence can not be explained neither by occasional coincidence, nor by any universal laws of biology. Presence of common characters of such kind testifies that these characters were inherited by their owners from their common ancestor. The fact that various animals which have a common character, differ one from another by other characters, testifies that at least some of these animals differ from the mentioned ancestor as well; thus, we come to the conclusion about existence of evolutionary changes. 

In order to clarify which concretely evolutionary changes and in which succession took place, it is necessary to analyse distribution of common characters in these animals. Such analysis is termed phylogenetic, or cladistic analysis (from Greek κλαδος – branch), because it clarifies out concrete branches of phylogenetic tree. Here under the word combination «phylogenetic analysis» we understand an analisys of phylogeny based on scientific hypotheses; however, the same term is often applied to a numerical method of dendrogram building, in which scientific hypotheses are premeditatedly ignored (see I.4.2.6 below).

As a founder of the phylogenetic analysis, German entomologist Willi Hennig (1913–1976) is regarded, who in 1950 published a book «Grundzuge einer Theorie der phylogenetischen Systematik» (English translation of this book «Phylogenetic systematics», which became most known, appeared in U.S.A. in 1966). However, the same methods which were described by Hennig, were used by biologists not less than one and half century before him. For example, even in the second volume of «Philosophy of Zoology» by J.-B. Lamarck (1830), a phylogenetic scheme is given which could be built only with usage of the principles of phylogenetic analysis (while comments to this scheme are extremely short and do not explain which methods were used by the author). Especially many works on phylogeny appeared beginning from the second half of XIX century, when thanks to works by Ch. Darwin the evolutionary approach got a general recognition. In the majority of works evolutionary reconstructions are in fact based on the phylogenetic analysis, however they usually lack necessary comments about methods of phylogeny reconstruction. Many these authors clearly understood the logic of phylogenetic analysis; other authors using this logic intuitively, by words explained principles of phylogeny reconstruction in other manner. Because of absence of distinct formula of principles of phylogeny analysis, some people who read papers on phylogeny, got a wrong impression that the phylogenetic schemes are drawn arbitrarily. Thus the Hennig's merit is not a discovering of a new method, but is that he was the first who clearly and consistently described methods of phylogeny reconstruction; his publications made subsequent authors to formulate clearly the principles which they use for their phylogenetic reconstructions.

One of the Hennig's merits is also that he introduced several important terms.

I.4.1.1. Terms «apomorphy» and «plesiomorphy»

The term apomorphy (from Greek απο-, from), or apomorphic character means a progressive (in the original meaning of this word, from Latin pro- – forward and gressus – movement), i.e. a derived, or secondary character. The term plesiomorphy (from Greek πλησιος – near), or plesiomorphic character means a primitive (in the original meaning of this word, from Latin primus, primitivus – first, primary) character (termed also original, ancestral, or generalized character). Plesiomorphy and apomorphy are two alternative conditions of a character, correspondingly before and after an evolutionary change happened with this character. These two categories are relative ones: if evolution goes from condition A to B and than to C, the condition B is an apomorphy in relation to A and a plesiomorphy in relation to C. Let's imagine that in course of evolution, shape of some part of an animal body changed from square to round and than to triangular; in this case the square shape should be apomorphic in relation to the round shape, but plesiomorphic in relation to the triangular shape:

The terms «progressive» and «primitive» are often connected with categories «more perfect» and «less perfect», or «more composite» and «more simple», that can lead to confusion. The terms «apomorphy» and «plesiomorphy» have only one meaning – a direction of evolutionary change of a concrete character in a concrete case, that makes these terms more strict.

An apomorphy which differs a certain taxon from all others is termed its autapomorphy, and an apomorphy common for several taxa is termed their synapomorphy. A plesiomorphy common for several taxa is termed their symplesiomorphy.

I.4.1.2. Logic of phylogenetic analysis based on synapomorphies

The phylogenetic analysis is based on an assumptions that evolutionary changes are unique and irreversible [the last is known as a Dollo's law (Dollo 1893)]. This assumptions, in turn, are based on consideration of probability: A probability that the same change in genotype can appear several times independently is so little, that it can be regarded to be equal to zero. Thus it is assumed that each apomorphy has originated only once and was inherited by all descendants of the species in which it had appeared. Consequently, if different species have a same apomorphic character (i.e. a synapomorphy), this testifies that all these species have inherited this character from their common ancestor, and therefore these species form an integral phylogenetic branch; at the same time, presence of common plesiomorphic characters in different species (i.e. symplesiomorphies) testifies about nothing. Thus, in the phylogenetic analysis we analyse synapomorphic similarity ignoring symplesiomorphic similarity.

It should be noted that in order to prove the phylogenetic relationships of animals, as well as to prove the existence of biological evolution in general, we use precisely the similarities between animals, but not the differences. Although differences between different species of animals arose during evolution, they are not sufficient to prove the fact of evolution, because differences between two objects can be provided not only by changes in the transformation of one object into another, but also by the independent origin of these objects.

In order to prove that several taxa form a common phylogenetic branch, it is necessary to find out at least one their synapomorphy, which at the same time would be an autapomorphy for all this branch. 

For this purpose it is necessary to determine, which of the alternative characters is the apomorphic one.

In order to clarify polarity of a character (i.e. in order to know out, which condition of the character is a plesiomorphy and which is an apomorphy), an out-group criterion is used, which means the following: 

If representatives of a certain phylogenetic branch have alternative characters, one of which is apomorphy and another is plesiomorphy, the same plesiomorphy can be found also outside of this phylogenetic branch (i.e. in out group), while the same apomorphy cannot be found outside of this phylogenetic branch. 

This statement proceeds from the assumption that the given apomorphy evolved as a result of a unique and unrepeateble evolutionary process occurred within the analysed phylogenetic branch, and because of this can not be found outside this group. In contrast to apomorphy, plesiomorphy was inherited from some earlier ancestral species, which may be the ancestor not only for the phylogenetic branch we are analysing, but also for other phylogenetic branches, which, in relation to the branch under consideration, are part of the external group. Some authors believe that for phylogenetic analysis it is enough to use the characteristics of one species as an «out-group»; this is a full misinterpretation of the «out-group criterion», because it is impossible to judge by any one or several randomly selected species whether a given characteristic occurs in the entire out-group. In reality, the out-group is not one species, and not any deliberately limited set of species, but the entire totality of living organisms outside the analysed phylogenetic branch. Since in practice the researcher cannot take into account all living organisms (if only because not all of them are discovered yet), one has to be limited by known data.

Task about three taxa. Let us take the simplest formal task – the identification of phylogenetic relations between the three taxa A, B, and C. If we assume that evolution consists of successive divergences, and the separation of one ancestor at once into three or more branches has too little probability, then three variants of phylogenetic relationships are theoretically possible between three taxa: 

In order to find out which of the three options is correct, it is necessary to analyse the characteristics of these three taxa. Let us designate in small letters a, a', b and b' characters where a and a' are two alternative states of one structural part, and characters b and b' are two alternative states of another structural part. Let us assume that the characters a and b are plesiomorphies, and the characters a' and b' are apomorphies, that is, their evolutionary transformations can be written as a→a' and b→b'. In our example, the taxon A has characters a and b, the taxon B has characters a and b', the taxon C has characters a' and b':

Thus, the taxa A and B have a common character a, while the taxa B and C have a common character b'. Since the character a is a symplesiomrphy and the character b' is a synapomorphy, we ignore the presence of symplesiomorphy a in the taxa A and B, and on the base of presence of synapomorphy b' in the taxa B and C conclude that the taxa B and C form an integral phylogenetic branch, to which the taxon A does not belong (integral lines show the branches whose existence is proven in this example):

We have got a grounded phylogenetic reconstruction, in  which each evolutionary transformation (a→a' and b→b') occurred only once. 

Here is important to note that for reconstruction of a tree whose branches contain more than one species each, it is necessary not only to find out synapomorphies between the branches, but also to find autapomorphies of each branch. If for a certain taxon composed of some number of reproductively isolated species, no one autapomorphy is find, we have no reason to figure this taxon in a form of phylogenetic branch. Thus, on the scheme given above, in addition to the apomorphies a' (for the branch C) and b' (for the branch B+C), autapomorphies for the branch A and the branch B should also be indicated; if such autapomorphies are unknown, it is not known whether the taxa A and B are separate phylogenetic branches, or the taxon A is identical to the common ancestor of A, B, and C, and the taxon B is identical to the common ancestor of B and C. Here, by the interrupted lines are conventionally shown those branches whose existence is not proved in this example.

In order to determine the polarity of the characters a and a', b and b' in the example under consideration, it is necessary to turn to an out-group, i.e. to phylogenetic branches outside the branch A+B+C. In the diagram below, for taxa D, E, F and G, characters other than the characters a, a', b, and b' are shown by three dots (...). Some of these taxa have the character a, some have the character b, but the characters a' and b' are not found outside of the branch A+B+C. From this fact we conclude that the characters a and b are plesiomorphies inherited from the ancestor common for all these branches, and the characters a' and b' are apomorphies evolved inside the branch A+B+C.

However, the facts presented here are not enough to prove just such a polarity of the characters a, a', b and b', and not enough to prove just such phylogenetic relationships between the taxa A, B, and C. Based on the same distribution of characters, we can admit a completely different polarity and a different branching of the phylogenetic tree:

Here we assumed that the apomorphy is not b', but b; in this case it turns out that the taxon A has synapomorphy with the branch F+G, while the relationship between taxa B and C appears to be not proven. Such assumption appears to be possible because we have not proved that the taxa D, E, F, and G belong to an out-group. In order to prove it, we must prove existence of the phylogenetic branch A+B+C. The only way to prove the existence of a phylogenetic branch is to find its autapomorphy, i.e. synapomorphy of its representatives. When we analysed the phylogenetic relationships between the taxa A, B, and C, we used only the characters a–a ′ and b–b ′ and ignored those characters which are identical for all three taxa A, B, and C. Now let us add a character which has a plesiomorphic state c and an apomorphic state c':

Now the existence of the branch A+B+C is proven by the presence of autapomorphy c'. Thereby, it is proved that D, E, F and G belong to an out-group; because of this, the presence of the character b in these taxa indicates that it is a plesiomorphy; therefore, b' is an apomorphy, and therefore the presence of this character in taxa B and C serves as a proof of their relationship. In this chain of reasoning, it is still necessary to prove that the character c' is an apomorphy, i.e. that its polarity looks as c→c'. To do this, it is necessary to turn to an out-group, i.e. to the phylogenetic branches outside the branch A+B+C+D+E+F+G and find an apomorphy proving the existence of this branch.

Thus, the process of phylogenetic analysis has neither beginning, nor end. Somebody believes that having objective facts about animal characters and correctly using rules of the phylogenetic analysis, we can reconstruct step by step a phylogenetic tree, not making mistakes; actually this is impossible. The reconstruction of phylogeny with help of phylogenetic analysis is a continuous and endless process of search of non-conflicting theory and correction of previously done mistakes. As demonstrated below, the characters on which reconstruction of phylogeny is grounded, are also not initial data, but themselves depend upon results of the phylogenetic reconstruction.

I.4.1.3. Interdependency of phylogenetic theory and character definition

In superficial view, it looks that animal characters are objectively existing and investigators only have to ascertain them. However actually in nature characters themselves are absent, and only structure of organisms exist. The characters are formulated by us on the base of the objectively existent organism structure, which is refracted by our perception. And our perception of the characters depend on our opinion about phylogeny of these animals.

For example, on such easy question for young pupils, how much legs have a beetle and a spider, no a single correct answer can be given (Table I.2).
           

Table I.2. Leg number counted by two different methods

  

Methods of counting legs	Selected groups of arthropods
	Hexapoda	larval Diplopoda	larval Monomero- somata	Arachnida	Xiphosura	Decapoda
Method 1: number of leg pairs	3	3	3	4	5	5
Method 2: presence of legs on postoral segments:
I
II					+
III			+	+	+
IV			+	+	+
V	+	+	+	+	+
VI	+	+		+	+
VII	+	+
VIII						+
IX						+
X						+
XI						+
XII						+

If we would simply count walking legs, we would get numbers shown in the line «Method 1»; this is only on of many possible methods of counting: here, for example, claws of cancers (Decapoda) are counted as legs, while the similar claws of scorpions (i.e. pedipalps of Arachnida) are not counted. Making such counting we can find similarity of Hexapoda, larvae of Diplopoda and larvae of Acari (each with 3 pairs of legs) and similarity between Xiphosura and Decapoda (each with 5 pairs of legs); in many classifications of XIX century, both Xiphosura and Decapoda were placed to the taxon Crustacea.

Recently this character is formulated in another way («Method 2»). When it is formulated in such manner, it appears that Xiphosura and Decapoda have nothing common, as well as larvae of Acari and Hexapoda; but a synapomorphy between Arachnida and Xiphosura is found, on which base they are united to the taxon Chelicerata (see Chapter IV). There is also a characters common for Hexapoda and Diplopoda, that can be interpreted as a synapomorphy. The interpretation of leg position which is shown here, is based on a number of assumptions accepted by us: (A) We regard that in arthropods legs of each determined pair are connected with a determined segment; with this statement some recent paleontologists do not agree; this statement is correct particularly for arthropods, but is not correct for other groups, for example for vertebrates. (B) We regard that the segment homologous in all arthropods is the segment situated directly behind mouth opening (i.e. the first postoral segment), and that for most certain this segment can be determined by its innervation [see Chapter IV: Metameria (2)]; not all zoologists agree with this statement: according to opinion by some authors, the innervation is not a reliable indication of the first postoral segment; according to opinion by others the mouth opening can change its position during the evolution, thus the first postoral segment is not homologous in various arthropods (see ibid.). (C) We count segments beginning from the anterior one and believe that in this case segments with equal numbers have to be homologous; at the same time it is evident that if count segments from the posterior end of the body, it would be not so. However, in some cases the opposite situation is found: for example, in all Chilopoda two last segments (the genital and the pregenital ones) are surely homologous; at the same time number of trunk segments in different chilopods is variable, thus if count the genital and the pregenital segments from the front, their numbers will differ in different species and even specimens. (D) We regard that the appendages of the first postoral segment of Hexapoda and Diplopoda have disappeared secondarily [see Chapter IV: Atelocerata (1)]; but according to opinion by some authors, this segment was never present, and in this case all scheme should be different. (E) We regard that the gnathochilarium of Diplopoda is formed by appendages of the 3rd postoral segment only, and the collum represents the 4th segment and lack appendages; this is also a subject of discussion [see Chapter V: Collifera (1)–(2)]. (F) We regard that the three pairs of legs are homologous in all Diplopoda; this is not evident at all, and had demanded special proofs [see Chapter V: Diplopoda (3)]. (G) We regard that in Diplopoda the segments marked here as V-VII are haplosegments; possibly this is wrong [see ibid.]. (H) We regard that the same segments of Hexapoda are haplosegments, but there are also doubts about this [see Chapter VI: Hexapoda (1)]. (I) We regard that the superlinguae of Hexapoda are not limbs, as some authors believed [see Chapter IV: Mandibulata (3)].

This list of assumptions an disagreements can be continued. Dependently which of the assumptions we accept and which do not, we find or do not find in these or that taxa common characters, and synapomorphies among them. Thus, the result of phylogenetic analysis depends on these assumptions. At the same tame each of these assumptions is made on the base of a certain opinion about phylogeny: for example, the first assumption (A) is based on the ideas about phylogenetic integrity of Articulata and of integrity of each of the arthropod group which has a certain set of limbs. These ideas about phylogeny themselves are based on the phylogenetic analysis of some other characters, which interpretation is also not evident and depends on ideas about phylogeny, and so on, endlessly.

In this simple example we see that an idea about phylogeny depends on formulation of characters, and the formulation of characters depend on idea about phylogeny. In connection with this, a danger exists to fall into a vicious circle of argumentation, when a wrong idea on phylogeny and a wrong formulation of characters would prove one another. A vicious circle appears when a set of arguments is limited, thus in order to broke the vicious circle, it is necessary to introduce new arguments. In phylogenetic reconstruction new arguments can be introduces by two ways: (1) to expand a set of species examined; this can be done if found out new species previously unknown to science (for many groups of insects this is a quite realistic way of investigation, as annually thousands of new species are discovered); (2) to expand a set of characters analysed; this can be reached if use mew methods and if examine such structures which previously were poorly examined or not examined.

I.4.1.4. Field of application of the phylogenetic analysis

For the phylogenetic analysis, can be and should be used all heritable characters of organisms – anatomical, histological, cytological, biochemical, physiological, ethological and others, belonging to any stages of ontogenesis.

The phylogenetic analysis in its classical form is intended for reconstruction of phylogeny based on divergences, when the phylogenetic scheme has a form of hierarchically branching tree, but not of a net. However, in some cases, factors such as hybridization or parallel gene transfer are admixed to the usual evolutionary mechanisms of divergence, in result of which genes are transferred not only from the common ancestor to its diverging descendants, but by other ways also, and the phylogenetic tree gets a net-like form.

In principle, the phylogenetic analysis can be used for taxa which have rank not less than species. Between infra-species taxa reproductive isolation is absent, thus their relations have a form of net but not of a phylogenetic tree, and cannot be clarified by the phylogenetic analysis.

Many authors use the phylogenetic analysis for reconstruction of phylogenetic relations of selected species. Theoretically this is rightful, but practically probably have little use, because the analysis at the species level does not allow to clarify and exclude from the analysis those characters c an exist in the genotype in hidden condition and to be displayed in phenotype as independently appearing homologous characters.

Besides synapomorphies (i.e. characters inherited from the common ancestor), there are characters which are also homologous but independently appeared. This is connected with the fact that animals can inherit from their common ancestor not a phenotypic character, but only genes which encode it; in further evolutionary changes such hidden character can appear in phenotype, and this can take place in different phylogenetic branches. Such independently appeared homologous character can be wrongly taken for a synapomorphy, and in this case the phylogenetic analysis can lead to wrong results. The more close is relation of the taxa, the more probable is that they contain hidden homologous characters. If each of the analysed taxa contains many various species, we can hope that the hidden characters common for each of these taxa would be exposed in phenotype of some of its species. In this case the independently appeared characters are founded out and we can exclude them from the phylogenetic analysis. Because of this, the higher is rank and the larger is volume (i.e. number of species) of each of the taxa included in the phylogenetic analysis, the more reliable is the result of this analysis. 

The less reliable is phylogenetic analysis for taxa of the species rank (the lowest rank for which the phylogenetic analysis is possible theoretically).

I.4.1.5. Phylogenetic analysis and evolutionary scenario

Before the out-group criterion was formulated, various other considerations were expressed on how to determine the direction of evolution. Actually, the only valid method for determining the polarity of a character is the out-group criterion, which consists of the fact that a plesiomorphic character, being inherited from an earlier ancestor, can occur outside the phylogenetic branch under consideration, but an apomorphic character that arose within the given branch is absent outside it (see I.4.1.2). 

For example, in limits of the taxon Hexapoda, in some representatives antennae have muscles in the first segment only, in others – in all segments except for the last one. If we would ground only on the antenna structure itself, we would have endless and useless discussion what is primary and what is secondary here – more simple antennae with limited mobility, or more complicate antennae able for various movements. None reasoning that evolution should be obligatorily directed somewhere – toward complication or, vice versa, toward simplification; toward increasing of adaptation to the medium or, vice versa, toward diminishing of adaptation; toward more perspective for further evolution or, vice versa, toward an evolutionary deadlock – such reasoning in this case do not give correct answer, because actually evolutionary changes can be infinitely variable and can have any direction. The principle of out-group comparison gives a single answer on a question about primary structure of antennae in Hexapoda: the plesiomorphy (i.e. the primary condition) is the presence of muscles in all antenna segments except for the last one, and the apomorphy – the presence of muscles in the first segment only: out of the taxon Hexapoda, multisegmented antennae with muscles in the first segment only are not found, while antennae with muscles in many segments are found in various groups of arthropods. From here the conclusion comes, that the structure of antennae with muscles in the first segment only, is an apomorphy which appeared in limits of the taxon Hexapoda, hence this character is considered to be an autapomorphy of a subordinate taxon Amyocerata inside Hexapoda (see Chapter VI).

Some authors believe that in order to reconstruct phylogeny, it is enough to build an evolutionary scenario, i.e. to describe an assumed way of evolution with description of possible reasons of each evolutionary change. In this case the evolutionary changes are discussed in connection with external conditions on which background these changes took place. For explanation why natural selection acts in certain direction, they use data about changes of climates and landscapes in various geological epochs. Actually each scenario of such kind is only one of many possible hypotheses.

For example some authors suggest scenarios describing a way from annelids – to onychophorans – to myriapods – to insects. In this case they assume that multisetose annelids (Polychaeta) which initially dwell in sea, came out to the land; as a result of adaptation for terrestrial king of life their parapodia protruded by sides were shifted to the ventral side of the body and transformed to non-segmented legs peculiar for Onychophora; further evolution of these animals represented a perfection of adaptation for terrestrial life, their legs got segmentation, anterior pairs of limbs were integrated to mouth apparatus, and other peculiarities characteristic of arthropods appeared, as a result of which Onychophora gave rise to Myriapoda and than to insects (Hexapoda). 

In this scenario no place is found for other arthropods – Chelicerata, Eucrustacea and Trilobitomorpha, because of this the apologists of this scenario regard that these animals derived from annelids independently from Myriapoda and Hexapoda. In connection with this they suggest scenarios of independent origin of chelicerates and eucrustaceans from annelids. In this case they assume that eucrustacean limbs, being derived directly from parapodia of Polychaeta, were initially non-segmented (swimming lamelliform appendages – fillopodia, peculiar for Branchiopoda), and were segmented during evolution of eucrustaceans. On the base of these evolutionary scenarios, a conclusion about polyphyly of arthropods was made (see Chapter IV: Gnathopoda:  «Classifications» II).

 However, a quite different theory about origin of arthropods exists, according to which arthropods form a holophyletic taxon (see Chapter IV: Euarthropoda). For this theory, another evolutionary scenario corresponds, which is not less convincing than the previous one: Polychaeta gave rise to Trilobitomorpha, and their limbs became segmented not in connection with terrestrial life, but in connection with inhabitancy on sea bottom; segmented limbs, inherited from trilobitomorps, were retained in eucrustaceans (and Branchiopoda secondarily lost segmentation of certain pairs of limbs in connection with transformations of them to swimming appendages); anterior pairs of eucrustacean limbs transformed, becoming a part of mouth apparatus; than follows turning to terrestrial kind of life and appearance of features characteristic of Myriapoda and Hexapoda. 

These two hypotheses are mutually exclusive. It is clear that the same insects could not arise by two different ways, and only one of these hypotheses is correct. The both hypotheses seem to be verisimilar when we are limited by building of evolutionary scenarios only. Besides these two ones, many other mutually exclusive theories about origin of insects were suggested, in each of them each selected step looks as verisimilar.

As seen here, the evolutionary scenario is not enough to prove a phylogenetic hypothesis.

An argued idea about phylogeny can be got only with help of phylogenetic analysis, i.e. a reconstruction of phylogeny on the base of apomorphies. To do this, we must first depict the hypotheses under consideration not in the form of evolutionary series, but in the form of cladograms: the evolutionary series consists mainly of hypothetical forms that are not amenable to direct study, and in the cladogram, each branch is terminated with a modern taxon, all of whose properties can be studied.

 Instead of the evolutionary row «Polychaeta → Onychophora → Myriapoda → Hexapoda» the following cladogram should be drawn:

 Instead of the evolutionary row «Polychaeta → Trilobitomorpha → Eucrustacea → Myriapoda → Hexapoda» the following cladogram should be drawn:

To proof the first of these theories and at the same time to disproof the second one, one have to find synapomorphies of the taxa Onychophora, Myriapoda and Hexapoda, which in this case would be autapomorphies of the taxon Ceratophora, because in the first theory the taxon Ceratophora is regarded as a separate phylogenetic branch. In the second theory, Onychophora are separated from the common ancestor of Myriapoda and Hexapoda earlier than it is separated from Eucrustacea, Trilobitomorpha and Chelicerata. Hence, according to the second theory, Onychophora, Myriapoda and Hexapoda can not have such synapomorphies which would be absent in Eucrustacea, Trilobitomorpha and Chelicerata. Actually synapomorphies of Onychophora, Myriapoda and Hexapoda (i.e. autapomorphies of Ceratophora) are not found yet.

Proofs of the second of these theories and at the same time a disproof of the first one, are firstly, autapomorphies of Gnathopoda; secondly, synapomorphies of Chelicerata, Trilobitomorpha, Eucrustacea, Myriapoda and Hexapoda – i.e. autapomorphies of Euarthropoda; thirdly, synapomorphies of Eucrustacea, Myriapoda and Hexapoda – i.e. autapomorphies of Mandibulata. These complexes of apomorphies really exist (see Chapter IV). The hypothesis about holophyly of arthropods (the second cladogram) explains all these cases of existence of common characters in diverse animals as a result of their inheritance from their common ancestor – i.e. as synapomorphies. In the hypothesis about polyphyly of arthropods (the first cladogram), existence of the phylogenetic branches corresponding to the taxa Gnathopoda, Euarthropoda, and Mandibulata is not recognized, so it is assumed that their common characters are not synapomorphies. Because of this, apologists of the theory about polyphyly of arthropods tried to explain the presence of these characters in diverse animals as a convergence, which appeared as a result of adaptation to similar conditions of life. These explanations seem to be rather strained, as it is difficult to decide which similar conditions can be present in deep-sea and on land surface where arthropods live, as well as it is difficult to decide which adaptive significance can have such features as the loss of coelom or the presence of an additional light-refracting lens in each ommatidium of faceted eye. Thus, when the two mutually exclusive theories are discussed, the phylogenetic analysis allows to make choice in favour of the theory about holophyly of arthropods: this theory is grounded by concrete apomorphies (while they are disputed by opponents), and the alternative theory about polyphyly of arthropods is not grounded at all.

I.4.1.6. Search for apomorphies

The main difficulty of phylogenetic analysis is the search for apomorphies proving the existence of certain phylogenetic branches (or, which is the same thing, holophyly of certain taxa – see I.5.1). Such apomorphies can be contained in the structure of any organs, but may be absent at all. If a researcher sets himself the task of finding the synapomorphy of taxa A and B (or, equivalently, the autapomorphy of a taxon uniting A and B), he can search for it anywhere – starting from the external shape or colour of the animal to the ultra thin structure of any part of the body or up to a nucleotide sequence anywhere in the genome. With this, the success of the search is not guaranteed. If an apomorphy is not found, this can mean either that the hypothesis about the sister relationship of taxa A and B is incorrect, or that it is true, but there are no synapomorphies of A and B in structure of the examined body part, or that the sought-for apomorphies exist, but were not found. In the search for apomorphies, it is not clear how long these attempts should be continued, and a lot of effort can be spent without result.

It is more rational to search for apomorphies from the other side, not from the phylogeny hypothesis to the search for apomorphies, but from the apomorphies to the search for their place in the hierarchical phylogenetic system of animals. This method is used in the cladoendesis. It consists in (1) writing down every known character as an autapomorphy of a particular holophyletic taxon and (2) giving a reference from each such autapomorphy to the initial autapomorphy of a higher taxon. The logic of this method is that each character, which could arose in evolution only once, is an autapomorphy of one and only one phylogenetic branch. If we already know about the existence of such a character, it remains only to find the phylogenetic branch, the autapomorphy of which this character is. In contrast to the searching for autapomorphies corresponding to a branch, the searching for a branch corresponding to an autapomorphy should always be successful: while not every phylogenetic branch has an autapomorphy, every apomorphy is an autapomorphy of some phylogenetic branch. To fixate the search result, it is necessary to give the phylogenetic branch status of a taxon (such a taxon is holophyletic – see I.5.1) and to give this taxon a name. Having the name of the holophyletic taxon, we can write under it all autapomorphies discovered for this taxon. If such a taxon was previously known, it is only necessary to supplement its characteristics with new autapomorphies; if such a taxon has not yet existed, it should be established and provided with a name.

For example, if the loss of the anus is indicated as an autapomorphy of a small specialized taxon of twisted-wing insects [see Strepsiptera (4)], then it is necessary to give a link to that superior taxon, the autapomorphy of which is the acquisition of the anus. This taxon appears to be a little-known taxon Haemataria, which unites most multicellular animals and is characterized by the presence of a through intestine [see Haemataria (1)]. Some modern classifications have no place for Haemataria; the authors of these classifications do not discuss the reason for abandoning this taxon, simply ignoring it. The need to make references to a taxon that has the through intestine makes us either to recognize the reality of the taxon Haemataria and abandon the hypotheses about phylogeny of multicellular animals that contradict this, or hypothesize the multiple occurrence of the anus, ground it and establish taxa characterized by the independent appearing of the anus. 

Until recently, the main difficulty was to supply with names all the taxa necessary for phylogenetic analysis. Since the traditional typified rank-based nomenclature allows to provide with names only a limited number of taxa (see I.6.3), and non-typified names were used chaotically until recently, many phylogenetic branches were left without names. To get out of the situation, they were designated by serial numbers or other arbitrary signs. Each such numbering of branches can be used only within the framework of one book or article, where, due to the limited volume of text, the author has to indicate not all autapomorphies. Other publications used different diagnoses for the same phylogenetic branches, and indicated other apomorphies, as a result of which summation and accumulation of knowledge about the apomorphies of these branches was difficult. The creation of rational universal rank-free nomenclatures (see I.6.4 and I.6.5) made it possible to solve this problem and to provide with unambiguous names all identifiable phylogenetic branches and all taxa (including plesiomorphons) which have to be created in the process of phylogeny reconstruction.

In order to record the autapomorphy of a taxon, it is necessary to report either a more general structural feature, or the ancestral state of a given structure. Each such mention should be provided with a reference to the corresponding autapomorphy of a concrete superior holophyletic taxon (which is now possible, due to the presence of each taxon's own name). Necessity to make such links makes the author critically review the distribution of known characters.

I.4.2. Misconceptions on phylogeny analysis

A widely spread opinion exists, that reconstruction of phylogeny is possible if combine three mutually supplementing methods – embryological (see I.4.2.1), paleontological (see I.4.2.2) and comparative (or comparative-morphological) ones. Combination of these three methods suggested in the second half of XIX century by Haeckel is called «Haeckel's triad». However it will be shown below that among these three methods, only the comparative method (more exactly, the phylogenetic analysis, or analysis of synapomorphies) has real significance (see above, I.4.1). Besides this, some authors try to use to justify phylogenetic reconstructions such, in their opinion, universal regularities of evolution as the principle of complication of organization, the principle of oligomerization, and others (see I.4.2.3, I.4.3.4, I.4.3.5). However in reality there are no any general laws of evolution (or at least they are not discovered yet) and according to modern view, evolutionary transformations can be quite diverse. In the literature of the late XX century, numeric methods of phylogeny reconstruction, which have no scientific justification, became widespread  (see I.4.2.6). Recently, large, but unfounded hopes have been placed on the so-called «molecular method» of phylogeny reconstruction (see I.4.2.7). 

I.4.2.1. Fallacy of embryological method

The embryological method is based on the «biogenetic law», according to which each organism during its ontogenesis repeats its phylogenesis in a shortened and modified form (here we use the word «phylogenesis» in its wide sense – see above). By other words, according to the «biogenetic law», the most ancient characters have to appear in individual development earlier than characters which are evolutionary younger.

In order to ground the «biogenetic law», specially selected examples are used, in which animals from different taxa at earlier stages of ontogenesis have more close similarity one with another than at later stages of ontogenesis.

For example, in many manuals a drawing is reproduced which shows selected ontogenetic stages of a fish, a bird and a mammalian, and at the earliest among the drawn stages these animals are very similar one to another (they have gill chinks, small limbs and large tail), and at the later stages they are sharply different. 

But numerous opposite examples are also well-known, when the greatest differences are found not at the later stage, but vice versa, at an earlier stage of ontogenesis. 

For example, differentiation of embryonic layers (ectoderm, entoderm and mesoderm) is a general character of Monostomata, while ways of formation of these layers at earlier stages of embryogenesis are characters of subordinate taxa [see Chapter III: Monostomata (1)]. Even if we turn to the mentioned above example with ontogenesis of the fish, the bird and the mammalian, the same animals at their earliest stages of orthogenesis (cleavage, gastrula et al.) have more sharp external differences than at the stage of gill chinks.       

Many researches repeatedly tried to use the «biogenetic law» for clarifying polarity of characters (i.e. to determine in which direction evolution of this or that character took place) and for reconstructing the ancestor. Often it leaded to evidently wrong hypotheses.

In entomology, examples of unsuccessful attempts to use the biogenetic law are the existent in XIX century theory about campodieform ancestor of insects, based on similarity of Campodea and some of insect larvae (see Chapter VI: Diplura: «Status»); the attempts to explain origin of the insect wings based on the wing developing in ontogenesis (see Chapter VII: Pterygota: «Classifications» VII); the opinion about homology of the ventral appendages of odonates with the cerci [see Chapter VII: Odonata (10)] and others. Some authors believe that the earlier difference between animals appear in ontogenesis, the earlier divergence of these animals took place in phylogenesis; on base of this they particularly assume a polyphyletic origin of arthropods, arguing this by the fact that in various groups of arthropods early stages of embryogenesis are different (Anderson 1973). However, if logically continue this idea, we would come to an absurdity conclusion that various groups of arthropods and even various representatives of such surely holophyletic taxon as Collembola (see Chapter VI) arose from various protozoans, because they have various types of cleavage (while the cleavage represents a changing from an unicellular to a multicellular condition). However generally the «biogenetic law» appeared to be not popular among entomologists, because a deep larval specialization in majority of insects visually demonstrates incorrectness of this law.

The «biogenetic law» was formulated in the XIX century, when nothing was known about molecular bearer of heredity and about mechanism of their transformation in evolution. Recently we know that during ontogenesis genotype is not changed, and the organism on all stages of its ontogenesis is built on the base of the same programs encoded in the same genotype; during phylogenesis the genotype is changed, but this takes place not by raising of new genes which would encode newly added stages of ontogenesis, but by changing (mutations) of formerly existed genes. Thus, the mechanisms of heredity does not provide the repeating of phylogenesis in ontogenesis. Characters inherited from an ancestor and indicating relations of a ceat in early stage, and in this case somebody say about expression of the «biological law»; sometimes, vice versa, such characters are found at a late stage.

An assumption can be made that the embryological method as a special method of phylogeny reconstruction is not grounded neither by theoretical ideas, nor by empirical data.

At the same time, embryological data are quite important for reconstructing of phylogeny, if they are used not for a special embryological method, but as a material for the phylogenetic analysis.

For example, adult crustaceans (Eucrustacea) have so diverse structure, that it is difficult to report any common characters of this taxon; at the same time in all Eucrustacea larvae (nauplii) or late embryos have evident common features [see Chapter IV: Eucrustacea (1)], that allows to make a conclusion about phylogenetic integrity of this taxon. At the same time it is difficult to imagine that the nauplius repeats a structure of the common crustacean ancestor, since it has non-segmented body and only two pairs of large postoral appendages (Fig. 4.7.1E); judging by everything, crustaceans being multisegmented animals arose from a multisegmented arthropodous ancestor with large number of homonomous limbs, and that one arose from multisegmented annelids.

In many arthropods during ontogenesis body segments originate as homonomous ones (the nauplius of Eucrustacea in this sense represents one of exceptions). Often in certain stages segmentation is well expressed, while in adult animals it partly disappears or masks – for example head of insects and many other arthropods appears to become practically non-segmented. Investigation of embryogenesis allows to clarify segmental composition of the head and some other divisions of the body, that allows to understand common characters of certain segments and basing on this to make conclusions about phylogenetic relations of various groups of arthropods.     

I.4.2.2. Restriction of paleontological method

The paleontological method means a supposition that we can directly see a result of evolutionary change if compare recent animals with fossil remnants of their ancestors. It is true that we can find in fossil condition the specimen which gave a posterity from which a recent species arose. In some cases preservation of fossils and exactness of estimating of their age are enough to make important trustworthy conclusions about structure and time of existence of these animals. However, now we have no any direct methods to recognize if this or that fossil animal is a direct ancestor of any of recent animals. Conclusions about relation ancestor-descendant can be made by the phylogenetic analysis only. In order to do this, it is necessary to reconstruct a common ancestor of several recent species on the basis of comparison of these species, and than to reveal that this reconstruction does not contradict structure of the certain fossil animal. In this case we have no any guarantee that the fossil animal is really a direct ancestor of the recent ones, but not a member of a side branch which retained that features of the ancestor which we are able to reconstruct.

The paleontological method can be accepted in a form of such rule: fossils which have plesiomorphies (i.e. ancestral features of structure) are found mainly in earlier deposits than fossils of animals which have apomorphies (i.e. derived features of structure). But there are many exceptions from this rule, that is explained by interrupted and occasional nature of paleontological data. It is clear that if in the moment of geological history A in a certain phylogenetic line a morphological character X₁ was transformed to X₂, the character X₂ can be found in some animals living later than the moment A, but can not be present in any animals living before the moment A. However there are no any limits for distribution of the character X₁, because when the character X₂ appeared, there was not any reason for disappearance of the initial character X₁, and it can be preserved for indeterminately long time in phylogenetic branches related to the branch where the character X₂ had appeared. Because of this in the rule formulated above, there is a word «mainly», and if the fossils interesting for us are known as a few specimens of a few deposits, this rule is non-acceptable.

Paleontological chronicle is in many respects occasional in relation to phylogeny, but it has its own regularities (see Chapter II.2). For example, soft-bodied animals have much less chances to be preserved than animals with hard skeleton; because of this, presence of animals with skeleton and absence of soft-bodied animals in ancient deposits in no manner testifies that recent soft-bodied animals arose as a result of loss of the skeleton. Since terrestrial animals and plants were preserved mainly in lake deposits, the largest number of their fossils belong to species living in lakes and banks; this does not mean that these species are more ancient than others. Reach collections of fossils are got from selected local deposits where conditions of sedimentation and all geological history helped to the good preservation of the fossils; but the fact that these or that species did not fall to these deposits does not mean that these species did nit exist at that time. Large number of exceptions make the rule formulated above to be of little use.

An example of unsuccessful interpretation of paleontological data is a history of investigation in Palaeodictyoptera and other Protorrhynchota (see Chapter VII). In the second half of XIX – beginning of XX century it was generally accepted that this group of insects should be the most primitive one, because it is one of the most ancient known groups of winged insects (they appeared in the middle of Carboniferous and died out in Permian). Palaeodictyopterans were called a «synthetic group», which contain inside itself ancestors of recent orders – mayflies, odonates, orthopterans, rhynchotans, hymenopterans and others. Particularly, in connection with this, the structure of Palaeodictyoptera s.str. was regarded to be an ancestral structure of recent odonates, and on this base dragonflies (Anisoptera) were regarded to be the most primitive among recent odonates, because they are more similar to the paleodictyopterans, than damselflies (Zygoptera) (see Chapter VII: Odonatoptera: «Classification»). However, when more paleontological material was accumulated, it was founded out that the paleodictyopterans and other protorrhynchotans form a holophyletic group which is characterized by a unique modification of mouth apparatus and is not ancestral to any of recent insects; correspondingly, that theories lost their meaning, which explained phylogeny of selected groups of winged insects basing on the idea about origination of these insects from the paleodictyopterans. In this case the mistake appeared probably because the large and actively flying paleodictyopterans were buried in lake deposits of Carboniferous, while really primitive insects had another kind of life and did not fall into that deposits.

The paleontological method as a special method of phylogenetic reconstruction does not give desired results. At the same time, paleontological data have great importance for reconstruction of phylogeny. In many cases fossil animals have such characters or combinations of characters, which are not found in recent species. Because of this, a comparative analysis of fossils give results which would be impossible to get studying recent forms only. For instance, examination of Palaeozoic mayflies with homonomous wings (Permoplectoptera) allows to understand phylogeny of mayflies, and without examination of Jantardachus and other Mesozoic groups allied to thrips, it would be difficult to clarify phylogenetic relations of Thysanoptera. But in these cases not a special paleontological method is used, but a phylogenetic analysis applied to paleontological objects.

I.4.2.3. Non-obligation of structure complication

An opinion occurs that evolutionary changes are directed toward complication of structure of the organisms. Such «law» of obligatory or preferable complication of organization in evolution could be explained if accept the position by J.-B. Lamarck, which was actual in the end of XVIII – beginning of XIX century. 

According to this antique theory, organism is influenced from outside by fluids, and it has some mechanisms because of which it reacts adequately in such a manner, that inside it are formed channels for guiding these fluids (i.e. veins, nerves and other integration systems); this complication of structure is inherited in some way; as a result of this, during adaptation to a new environment, besides structures which already existed, new ones are added; i.e. new adaptation leads to complication of structure, that leads to regular transformation of more simple organisms to more composite ones. The fluid theory was disproved even in the Lamarck's time, and his assumptions about adequate reaction of organisms to influence of environment and about inheritance of acquired characters have not got further confirmation, and recently are regarded to be wrong.

It is doubtless that earliest living organisms had simple structure; that more complicate eucaryotes arose from more simply organized procaryotes; that multicellular animals arose from unicellular ones and that ancestors of such complicatedly organized multicellular animals as arthropods or vertebrates were multicellular animals with more simple structure. Since initially organisms with complicate structure did not exist, they originated as a result of complication. But this in no way allows to conclude that there is a law of complication: the evolution equally can lead to complication and to simplification. When we discuss phylogenetic relations of certain animals, in some of which structure is more complicate, and in others more simple, we can with equal success assume either an origination of the more complicate from the more simple one, or vice versa, origination of the more simple from the more complicate one. Because of this comparison of organisms by their complexity gives nothing for phylogeny reconstruction.

The discussed below opinion about a «law» of oligomerization of homologous organs is an opposition to the «law» about complication of structure.

I.4.2.4. Non-obligation of oligomerization

Many modern phylogenetic theories, including that in entomology, are based on an assumption that evolution is supplied with the oligomerization (i.e. reduction in number) of these or that organs initially existed in greater number. For example, it is assumed that arthropods initially had large number of equal leg pairs and subsequently the number of leg pairs was reduced (see Chapter IV: Euarthropoda: «Classifications»); winged insects initially had two pairs of equal wings, and subsequently only one pair was retained (see Charter VII); and so on. Most of these theories are reliable as are argued on the base of phylogenetic analysis.

However, it would be necessary to warn against absolutization of the idea about oligomerization of homologous organs. Discussing phylogeny of a selected group of animals, somebody tries to chose the most simple explanation among all possible ones. If one animal has some organ, and another animal has no such organ, it is easier to assume that reduction of this organ took place in the second animal, than to suggest a hypothesis which would explain appearing of this organ in the first animal. In such manner, going by the way of easiest explanation, some authors try to restrict all evolutionary changes to reductions.

For example in the papers by Kukalova-Peck (1983, 1985, 1987, 1991) a theory of phylogeny of winged insects (Pterygota) is suggested, according to which the hypothetical ancestral form bears nearly all details of structure found in any really existent insects, and these details are in a polymerized condition. Each its segment besides a pair of limbs, bears homologues of eyes and wings; each limb has maximum number of segments and claws known for any limb of any insect; each segment bears a stylus, each stylus is segmented, and so on. Reasoning in such manner the author liberates herself from necessity to explain appearance of new organs in evolution: In this case somebody other have to work out the questions about emergence of details peculiar for Pterygota – somebody who reconstructs earlier stages of arthropod evolution.

However, it is clear that if animals have these or that organs, origin of these organs should be explained.

In some cases a phylogenetic hypothesis based on opinion about oligomerization of homologous organs appears to be wrong.

In the past many authors regarded representatives of Myriapoda with large undetermined number of segments as primitive forms. Particularly, it was assumed that Epimorpha are the most primitive among centipedes (Chilopoda) because they have large varying number of segments (see Chapter V: Chilopoda: «Classifications» I). However, phylogenetic analysis reveals that this version is be wrong and we have to accept a hypothesis according to which increasing of segment number in Chilopoda took place secondarily (see ibid., II). Similarly, according to recently existed theory about phylogeny of millipedes (Diplopoda), a large undetermined segment number which is peculiar for some Helminthomorpha is secondary, while initially millipedes had relatively small and determined segment number (see Chapter V).      

Thus, comparison of animals by their oligomerism or polymerism is not enough itself to make conclusion about direction of evolution.

I.4.2.5. Non-obligation of ancestor non-specialization

Often a species ancestral to a certain taxon is assumed as something non-specialized, combining in itself features of all members of the taxon together. With such point of view is connected a term «generalized» character or organism. Actually each animal species has its own peculiarities, its own specialization and evolves under natural selection in particular conditions of medium, but not under an aspiration to produce in future a large and diverse taxon. During its evolution not only specialization takes place, but also a despecialization and a changing of specialization.       

As one of numerous examples of disappearance of a primary specialization, evolution of mayfly larvae (Ephemeroptera) can be reported: It is convincingly shown that larvae of mayflies initially had a specialization for active swimming [see VII: Ephemeroptera (1e)] which in various groups of mayflies disappeared or changed to specialization for another kind of life; sometimes such larvae, having lost their swimming specialization, look as «generalized» ones, because resemble many insects together.

No rise to doubt is given by theories according to which the evolutionary way leading to bees with their social kind of life and feeding by pollen of plants passed through a specialization for parasitism on insects (which is peculiar for primitive representatives of Apocrita); the way leading to tetrapods and human passed through a specialization to swimming (peculiar for fishes); and many others.

Most probable that the unknown for us ancestral species, such as the common ancestor of winged insects, the common ancestor of insects with complete metamorphosis and others, were quite specialized insects, and such fundamental characters of large taxa as wings or complete metamorphosis initially arose as particular adaptations for concrete conditions but not as an inexplicable aspiration to give birth to biological diversity.

I.4.2.6. Why do numerical and matrix methods fail

The distinct formulating of the principles of phylogenetic analysis provoked in many investigators a temptation to formalize the process of phylogenetic reconstruction and to use for this purpose a computer. Recently there are compiled computer programs for building phylogenetic trees on the base of previously founded apomorphies. However the reconstruction of phylogeny by the formalized methods runs against insuperable difficulties and is scarcely expedient. Since a passion for these methods is widely distributed recently, here it is necessary to look into their shortcomings of principle.

For computer building of phylogenetic tree, at first a matrix taxon/character is compiled (which is incorrectly itself – see below, I.4.2.6.2), where for each elementary taxon (called operational unit) included to the analysis, a state of character is marked and it is indicated if this state is an apomorphy or a plesiomorphy. Than on the base of this matrix a phylogenetic tree is built; the program which builds a tree is based on the same principle as the phylogenetic analysis (see I.4.1 above) – i.e. on an assumption that the same apomorphic conditions of characters are synapomorphies, i.e. are inherited from the common ancestor. If everything would be limited by this, the usage of computer program would be evidently superfluous, because it is easier to make all these procedures by hands than by computer. But during building of tree on the base of matrix taxon/character, often non-solvable combinations of characters are found:

For such matrix no one phylogenetic tree can be built, because in each case it will appear that some of characters (either x₂, or y₂) had appeared independently twice. If change polarity of any character (for example, x) to an opposite one, on the base of this matrix a non-conflicting tree can be built:

Let us assume that x2 is a plesiomorphy, x1 is an apomorphy;  y1 is a plesiomorphy, y2 is an apomorphy WWWWWWWWWWWWWWWW

It is possible to use a program in which polarity is not given presumably, but it clarified during building of the phylogenetic tree: as the correct one, is regarded that polarity of characters, which allows to build a non-conflicting tree (for example, as in the previous example). However, the matrix can contain such combination of characters, which do not allow to build a non-conflicting tree independently of assumption about polarity of characters:

Theoretically, such non-solvable combinations of characters can not exist, but practically they are found very often. Reasons of existence of such combinations can be various: (1) As a result of convergence, during adaptations to the same conditions, organisms can get similar characters which looks as identical ones but have different genetic bases; such cases cause difficulties not only for computer methods, but for any method of phylogeny analysis. (2) Identical characters which have the same genetical basis, can appear independently in related groups (see I.4.1.4 above). (3) The seeming synapomorphy can be actually an artificial character – i.e. several completely different characters which are equally worded. While the first two reasons bring difficulties for any kind of phylogenetic analysis, the last of these reasons is characteristic specially for formalized computer methods: investigator can distinguish an evidently artificial character from a real one, but computer which deals not with characters but with their wordings only, is unable to do it. When a non-solvable combination of characters is found, it is necessary to reexamine these characters of the investigated animals, in order to find a mistake. But the aim of computer programs is to avoid reexamination of animals, and to dissolve a problem by purely computer methods.

In order to circumvent the appeared obstacle, an assumption is introduced that evolution can be reversible and (or) repeatable. If not to add here other supplementary conditions, the phylogenetic analysis loses any meaning and the computer builds many possible trees. In order to select among many trees a single one, a principle is used which is called a principle of economy, or parsimony. This principle is following: among all trees, such a single tree is selected, in which the violation of principle of irreversibility and (or) non-recapitulation takes place the less number of times. 

I.4.2.6.1. Unavailability of parsimony principle. The principle of parsimony has no relation to any biological theories. Usually, referring to the principle of parsimony in the reconstruction of phylogeny, they do not specify which parsimony is meant – either it is the parsimony of nature, or the parsimony in the actions of the researcher.

In the first case, the principle of economy is in direct contradiction with modern science. The most parsimonious phylogetic tree selected with help of this principle, hardly can be regarded to be the most correct one, as we do not know a natural mechanism which would make animals to chose the most parsimonious (or optimal in any respect) way of evolution. Phylogeny of every group of animals is unique. The evolutionary processes which took place, happened only once, and because of this no one mechanism of optimization of evolutionary process could exist. One could suggest some theory of economical evolution, alternative to Darwinian; in this case, the use of the principle of parsimony for reconstructing the phylogeny of a particular group of animals would be based on this theory and, thus, would be within the framework of scientific knowledge. However, those authors who refer to the principle of parsimony in the reconstruction of phylogeny, at the same time, refer to Darwin's theory as the only possible one. Such approach directly contradicts the principle of scientific knowledge.

In the second case, the idea that the researcher should act parsimoniously (that is, not to create extra entities without necessity) does not fit in with the actions that the researchers perform when analysing phylogeny using matrix-numerical methods. To calculate the «parsimony», a huge number of very complex programs are written, the use of which in itself requires certain knowledge and skills.

As the only justification for using the principle of parsimony in phylogenetic analysis, they refer to the works by the philosopher K. Popper. This author, not being a biologist, had no idea about the problems of the analysis of phylogeny, but his publications contain discussions about methods of cognition in physics. Explaining how the regularly repeating physical phenomena are investigated, Popper referred to the principle of refutability of scientific theories (for some reason he called it the principle of «falsifiability», although in the common vocabulary this word has the opposite meaning). According to this principle, a universal scientific theory (i.e. a theory explaining an infinitude of particular facts) cannot be proved, because in order to do this it would be necessary to examine an endless number of facts; but it can be disproved, if one would find a single fact which contradicts this theory. Because of this, a valuable scientific hypothesis should be disprovable – i.e., should be formulated in such a manner that it would be principally possible to assume existence of facts which would contradict this hypothesis. Until these facts are actually not found, the hypothesis is regarded to be correct (more exactly, not disproved). In order to explain the same fact, endlessly much mutually exclusive theories can be suggested, and in order to disprove each of them, a time should be spent. In order to regulate this process, as a working hypothesis is chosen the most parsimonious (i.e. the most simple) hypothesis among all possible ones; if this hypothesis is disproved, the most parsimonious among the rest ones is chosen. Here the principle of parsimony is served not as a criterion of truth, but only to chose a succession in which the hypotheses will be verified by help of other criterions. In the numerical cladistics, vice versa,  the principle of  parsimony is used as the only and the final criterion to chose a «true» phylogenetic hypothesis among many hypotheses which are already disproved (!), and no one of which will be verified in future. For detail discussion of this question, see Pesenko (1989).

I.4.2.6.2. Incorrectness of data presentation in a matrix form. To use numeric methods, a rectangular two-dimensional matrix of taxon/character is first created (see I.4.2.6). At first glance, such a representation of the data seems reasonable; sometimes it’s really useful for a researcher to compile for himself such a table of characters that is perceived easier than a set of verbal descriptions of taxa.

However, all the information about characters distribution should not be reduce to a matrix.

Firstly, when compiling the matrix, one has to come up with a new artificial set of characters. As shown by the experience of biological systematics since the time of Linnaeus, and as is clear from the modern theory of evolution, the diversity of living organisms fits not into a matrix, but in a hierarchical form. Linnaeus (1751) quite rightly wrote in the «Philosophy of Botany»: «Quae in uno genere ad Genus stabiliendum valent, minime idem in altero necessario praestant» (that which is important in one genus for establishing a genus, in another does not matter at all). The taxonomists formulate each character in such a way that, with this formulation, it is stable within a certain taxon and allows one to reliably distinguish this taxon from one or several taxa closest to it. However, this does not mean that the same formulation of the character will allow to characterise other taxa. 

For example, the taxon Raptoriae (praying mantis) is clearly characterized by a certain grasping specialization of the fore legs [see Chapter VIII: Raptoriae (1)]; however, in other insects the fore legs are very diverse – from non-specialized to the same grasping ones as in mantis, so this character cannot be entered into the matrix for all insects.

 In order to put all the characters into a strict two-dimensional matrix, one has to apply artificial methods – throw out inconvenient characters and change the formulations of characters, abandoning those that have already justified themselves in building a hierarchical system. As a result of this, when at the next stage of analysis a hierarchical system is obtained from the matrix (in the form of a branching phylogenetic tree), it turns out to be based on a smaller number of facts than was originally known. Often, to compile a matrix, a complex feature is replaced with a set of simple features, believing that such an action is an analysis. In fact, such a decomposition of a character into its components is not a natural-scientific analysis, but only a purely logical trick, as a result of which information is not added, but, on the contrary, disappears. For example, if some part of the animal’s body has a characteristic composite shape, we understand that such a shape cannot be repeated randomly, and therefore we successfully use this character in phylogenetic analysis. However, it is impossible to enter such information into the matrix: if we denote this complex shape by one arbitrary symbol and write it into the cells of the matrix related to species possessing this shape, then we will not be able to fill in the corresponding cells for other species. To fill all the cells of the matrix, it will be necessary to replace the designation of a unique shape with a set of epithets, each of which has its own antonym: for example, «long» (antonym – «short»), «widened distally» (antonyms – «not widened distally» and «narrowed distally»), «with a projection» (antonym – «without a protrusion»), etc. In this case, all cells of the matrix are filled, but at the same time, the characters with which they are filled have lost their uniqueness and therefore have become unsuitable for phylogenetic analysis.

Secondly, when compiling the matrix, one has to come up with a new artificial set of taxa. Taxa entered in the matrix are called «operational units». The operational units may be taxa of the equal ranks or have different ranks in the original formal classification (i.e., among them there may be species, genera, families, etc.); in both cases, they are all analysed as if having the same conditional rank. The choice of taxa that are equalized in rank is arbitrary (as in the traditional ranking classification – see I.5.4.2).

Since the formulation of features depends on the choice of operational units, the matrix is ​​largely arbitrary and artificial.

Often the matrix is ​​compiled in order to process it with mathematical methods based on the «principle of parsimony». Such an action is clearly absurd, because the only justified understanding of the «principle of parsimony» is parsimony in the actions of the researcher (see I.4.2.6.1), and compiling a matrix instead of a dendrogram in order to obtain a dendrogram, as well as the use of cumbersome mathematical calculations, are not parsimonious actions. 

The compilation of a taxon/character matrix is ​​quite justified if input there characters having a mosaic distribution among taxa. The mosaic distribution of a character is obtained if the gene program encoding the character is present in the entire set of these taxa, but only in a part of them is realized. The characters distributed in such a way cannot be used in phylogenetic analysis.

I.4.2.6.3. Fundamental shortcoming of numerical taxonomy. Usage of numerical (particularly computer) methods of phylogeny reconstruction lack meaning by the following reasons.

(A) In the methods discussed above, as well as in many others, all calculations are done on the base of arbitrarily compiled «convenient» mathematical formulas; this kind of work cannot be considered scientific. In science, the application of mathematics is expressed in the fact that a scientific theory is written in the form of a mathematical formula, and it is this formula which is used for calculation in the cases foreseen by this theory. In its turn, the theory which is reflected in the formula, is a subject of scientific discussions, because it is formulated in such a manner that allows possibility to its disprovement.

(B) Most of the currently used mathematical methods for phylogeny reconstruction use such a quantity as the number of characters – the number of common or different characters of several taxa, the number of apomorphies or plesiomorphies, etc. 

Phylogenetic tree constructed using these methods depends on the number of synapomorphies in these or that taxa. Here we should pay attention to the fact that in contrast to the numerical ones, in the real phylogenetic analysis (see I.4.1) the number of synapomorphies does not affect the result in any way: there a phylogenetic branch is regarded to exist if its members have at least one synapomorphy; a larger number of synapomorphies gives more confidence in correctness of the tree reconstruction, but does not affect the tree itself. The characters number used in numerical methods, is a category meaningless from biological point of view, because one and the same character can be formulated in various manners, being short or long.

For example, the character of the taxon Entognatha [see Chapter VI: Entognatha (1)] can be written as a single word: «enthognathy». The same character can be written much longer (see ibid.). A long text can be made in the form of one compound sentence, and in this case it will be regarded as a single character. We can make the same text as many sentences, and give separate number to each of them, in this case all of them will be regarded as different characters. In the same manner the character of the taxon Pleomerentoma, connected with abdominal limb structure (see Chapter VI: Hexapoda: «Classifications» III) can be written as one or several sentences and can be regarded as one or several characters. If number of sentences (i.e. counted «characters»)  in the characteristic of Entognatha will appear to be greater than in the characteristics of Pleomerentoma, the following phylogenetic scheme will appear: 

If vice versa, quite different conclusion about phylogeny will appear:  

Even if we will take absolutely simple objects, for example a square and a ring, we would not be able to answer the question how many differences are between them, i.e. how many peculiar characters each of them has.

It is possible to say that there is a single difference – the first object is a square, and the second one is a ring. It is possible to say that there are 4 differences – fore angles in the first case and no one in the second; with the same success we can count both angles and sides, curvation of sides and so on. Actually each geometrical object includes endless number of points and because of this is described by an equation which has endless number of values.

Structure of an organism is determined by genes; number of genes in a multicellular organism is very large, but not endless. However, we take into account not genes, but characters; in contrast to genes, number of character is not large, but endless.

In numerical methods, arithmetical actions are done with numbers of characters – division, subtraction, et al. Actually numbers of characters in all cases are equal to infinity, and only a person who does not understand what does he do, can make arithmetical actions upon them.

Everything said here about calculation of characters, is related not only to numerical methods in phylogeny reconstruction, but to any other numerical methods in systematics. In 60th years of the XX century there was such movement as phenetics, or numerical systematics, which disclaimed significance of phylogeny in systematics and suggested to base classification on an «objective» calculations. The modern methods of numerical cladistics are in great part adopted from phenetics (Mair 1969, Pesenko 1989–1991).

(C) Independently if inadequate mathematical methods are used (see A) or not, if an artificial character calculation is used (see B) or not, usage of computer for phylogenetic reconstruction can not bring a desired result.

The whole advantage of computer is that it is capable of quickly carrying out a long chain of logical operations; if it is required to perform only a few logical operations one after another, then this is easier to do in the mind. If it would be possible to divide the process of phylogeny reconstruction in to two successive steps – on the first step to make a description (i.e. formalization) of characters, and on the second step – reconstruction of phylogeny on the base of these formalized characters, this second step could be done with help of computer. However formulating of characters is possible only if we have an idea about phylogeny of the investigated group, because each description of animal already contains the author's opinion about nature of this animal or of this character (see I.4.1.3 above). Thus, all work on phylogeny reconstruction should be done simultaneously, without division in to steps. It means that after a short chain of logical operations (so short that it can be easily done mentally) it is necessary to return to reexamination of animals in order to specify or correct the formulation of characters.

For example, if we got two conflicting trees shown above, it is useless to continue logical operations with the existent formulation of characters of Entognatha and Pleomerentoma. Instead of this, it is necessary to dissect heads of various representatives of Entognatha, compare them with other insects, and taking into account the appeared doubt about common origin of Entognatha, anew describe structure of their mouth apparatuses. The same should be done with abdominal appendages of various insects, in order to revalue critically the characters of Pleomerentoma.

I.4.2.7. Substitution of molecular method of phylogeny analysis

The rapid increase in knowledge of the structure and function of DNA allows us to hope that this knowledge can be used to understand phylogeny. At the same time, modern scientific literature is full of publications on the so-called «molecular phylogeny», which have no scientific significance.

The often used collocation «molecular phylogeny» should not be taken literally, because phylogeny (that is, the formation of related groups) can exist only with whole animals which are related in the literal sense of the word, but not in their molecules or any other parts. Under the «molecular phylogeny» they understand molecular methods of phylogeny reconstruction, more precisely, phylogeny reconstruction methods based on DNA analysis. So much is being written about these methods that some people even think that it is possible to improve DNA analysis to such an extent that they automatically receive a phylogenetic tree. However, the phylogenetic tree is not recorded in DNA, so that with any perfection of methods, the phylogenetic tree will remain only a hypothesis that cannot be conclusively proved, but which can be infinitely approximated to reality, supplementing existent data with new ones.

An opinion exists, that molecular methods of phylogeny analysis can now replace and supplant phylogeny analysis based on phenotypic characters. Such an opinion may arise as a result of the following reasoning.

When we reconstruct phylogeny using the traditional comparative-morphological method, we use only inherited morphological characters, and only in order to reveal relationships of organisms by analysis of inheritance of these characters. At the same time, everyone clearly understands that morphological characters as such are not inherited: when an animal dies, its body is destroyed or eaten by representatives of other species, so that no morphological parts of it are transmitted to its descendants. Besides one ovicell (which does not have the structure of an adult animal), the descendant inherits only information recorded in DNA molecules. Thus, when reconstructing phylogeny, we analyse morphological characters only in order to make conclusions about the sequence of nucleotides in DNA: the knowledge about the presence of identical morphological characters in different animals (among which there are synapomorphies interesting for us) is needed only to make an indirect conclusion that these animals have the same inherited genes encoding these characters. In turn, the presence of the same genes interests us only because on its basis we can draw an indirect conclusion about the common origin of those animals that possess the same genes. Thus, when analysing phylogeny using morphological characters, we make an indirect conclusion twice: (1) assuming that in this case the same characters are encoded by the same genes and (2) assuming that in this case the same genes are inherited from a common ancestor. That is, with such an analysis, we have two sources of possible errors. If, instead of revealing morphological similarities in order to indirectly judge gene similarities, we immediately identify gene similarities by direct DNA sequencing, we will eliminate one of these two sources of error. The second source of errors remains, because the genealogy is not recorded either in morphology or in the genes, and in any case it can only be detected indirectly, i.e. by erecting hypotheses that may appear to be erroneous. Thus, a special molecular (genotypic) method of phylogeny reconstruction can be created, alternative to the morphological (phenotypic) method: in the molecular method, instead of analysing apomorphies at the morphological level, the same apomorphies are analysed at the gene level. Such a method would be fundamentally more accurate and reliable than the currently used morphological method (although the molecular method, like any other, will not allow error-free reconstruction of phylogeny with a given accuracy). If the molecular method would exist, one could argue about in which cases which of the two methods is better – the more expensive, but more reliable molecular, or more affordable, but less reliable morphological.

However, at present there is no such molecular method, as well as there is no scientific knowledge on which it should be based. Modern science only knows how genes encode proteins, and has some idea of ​​the mechanisms of gene interaction. We only know about the encoding of morphological characters by genes that they somehow encode them, but we have no idea how. Nothing is known about which genes encode those morphological characters that we use in phylogenetic analysis.

At present, it is widely practiced to substitute a real molecular method for its surrogate: instead of analyzing those genes that encode the apomorphies of interest to us, they analyze those genes that are available for this – such as the ribosomal RNA gene, genes of certain enzymes or mitochondrial genes. In these cases, the study of molecules is not a special molecular method for the analysis of phylogeny, which could be contrasted with the morphological method. Indeed, if we are interested in large structural details of the animal, we can simply examine them with our own eyes, and if we are interested in what is not visible to the naked eye, we use certain technical devices; but the use of these devices is not a special method of phylogenetic analysis. If a person does not see well, he puts on spectacles; if we are interested in the smaller details of the structure, we use a microscope. If a light microscope is not enough, we use an electron microscope; however, with an electron microscope, our eye cannot see anything, so we judge the structure of the object indirectly, based on the reconstruction of the image created by this device. In order to examine the characters by which ribosomes differ, the resolution of the electron microscope is not enough, and these characters can be judged indirectly, based on the sequence of nuclear DNA which encodes the ribosomal RNA. As in the case of spectacles, a microscope or an electron microscope, DNA sequencing in this case is only a technical method for studying the morphology of a given object. When, under the mask of a special «molecular method» of phylogeny analysis, instead of traditional morphological characters, 18S ribosomal DNA is examined, this is not a replacement of the morphological method with the molecular one, but only a replacement of traditional characters with structural features of the 18S ribosome subunit.

The study of ribosomal RNA, enzymes, mitochondrial DNA, or other structures available for study by molecular methods, may be useful. These data can be successfully used in phylogenetic analysis, if  correctly understand their significance. For example, if in addition to the previously known external structure of the animal, we get knowledge about the structural features of its ribosomes, we can try to find apomorphies useful for phylogenetic analysis, not only among the characters of the external structure, but also among the details of the ribosome structure. If such apomorphies can be found in the studied ribosomal gene, they can be used in the reconstruction of the phylogenetic tree, in which individual phylogenetic branches are proved by certain characters of external morphology and/or of the nucleotide sequence in the ribosomal DNA. Like any other character, a molecular character can be used in phylogeny reconstruction in addition to other characters used for this.

Instead of this, many publications contain dendrograms told to be «phylogenetic trees», which are built solely on the basis of data on the sequence of several known genes for few species, while completely and deliberately ignoring all other known facts.

By analogy with the concept of conservatism of certain characters, which is known in taxonomy (see I.2.4), some authors justify the choice of a gene for phylogenetic analysis by the assumption that this gene should be more conservative than others. Such statements are based on the erroneous idea that conservative should be those genes and characters which do not affect the interaction of the organism and the environment and therefore are not subject to natural selection. From the experience of constructing the classification and analysis of phylogeny by phenotypic characteristics, we know that this is not so.

It may seem that if the procedure of formal phylogenetic analysis is applied to nucleotide sequences, the results of such an analysis will be unambiguous and objective. While for the mathematical processing of phenotypic characters they must first be artificially formulated (as a result of which the characters lose their objectivity, and the use of mathematics becomes meaningless), the nucleotide sequence can be written objectively and unambiguously. However, in reality, mutations, due to which a difference in nucleotide sequences occurred, took place in the ancestors of the studied animals against the background of complex gene regulation, so that the replacement of certain nucleotides during these mutations was not equally probable. It is well-known from comparative morphology that a particular character in a particular taxon can be very stable and is not disturbed by any mutations, while a similar character in another taxon can be highly variable (see I.2.4). It is the certain persistent characters of concrete taxa (but not any characters) that are traditionally the basis for the reconstruction of phylogeny and the construction of a classification [«what is important in one genus for establishing a genus, in another genus has no meaning» (Linnaeus 1751)]. To discover the molecular mechanisms which ensure the stability of certain characters, knowledge of the sequence of nucleotides is not enough, but it is necessary to understand complex molecular interactions. So the attempts of some authors to calculate phylogeny by simple mathematical means, based only on the crude results of DNA sequencing, are premature.

About how implausible is the modern so-called «molecular phylogeny» can be judged by such a comparison. Eucaryotic 8S ribosomal DNA conatains 1900 nucleotids. Since there are only 4 types of nucleotides, each of them carries 2 bits of information, so that all 18S ribosomal DNA contains 3800 bits of information. Let us compare this with the information contained in a text. If the alphabet contains more than 32 letters, then each of them carries more than 5 bits of information; so a text containing 3800 bits of information should consist of no more than 760 letters. In a medium-sized book, such a text will take about 12 lines. Even if somebody would try to intentionally write down all the information about the phylogenetic position of the owner of this ribosome in the 18S ribosomal DNA, he would hardly be able to place there so much encrypted information (not to mention that the ribosome, besides performing of the dubious function of the historical archive, must also synthesize protein). Nevertheless, analyzing phylogeny by 18S ribosomal DNA only, some authors managed to find information about phylogenetic position of one and the same animal at all levels – from the initial divergences of living organisms to relation with the nearest species and populations, and to write about this much more than twelve lines of text.

Currently, despite the rapid development of molecular techniques and increased efforts to use molecular data for phylogeny reconstruction, the contribution of these studies to the phylogeny reconstruction is small. In particular, in the plentiful literature on the «molecular phylogeny» of insects, I could not find a single fact that could be used in this book.