1. INTRODUCTION TO PALAEOENTOMOLOGY
1.1. SCOPE AND APPROACH
by A.P. Rasnitsyn
This book tries in general to cover the whole extent of the history of insects in time and space in detail (Fig. 1). The exception is the Quaternary Period which is only occasionally considered here in depth because Quaternary palaeoentomology is a specialised field with its own, rather specific methods and approaches. Quaternary palaeoentomology has recently been dealt with in considerable detail by Elias (1994), and this might excuse our omission here. We understand that the selection of which subjects to present and discuss here reflects our personal views and tastes, and that many important subjects may have been missed, either by mistake or ignorance. We hope that our readers will help us to make any future editions better than the present one.
The geochronological scale employed here is shown in Fig. 2, and the majority of insect fossil sites are displayed on the maps (Figs. 3-5). The sites/site groups numbered and labelled there are the same as those catalogued in Appendix (Chapter 4).
The class Insecta is taken here in its narrow sense, that is, not including the entognathous orders (Acerentomida = Protura, Campodeida = Diplura, Podurida = Collembola). Although the relationships of these three orders is not fully resolved, there is growing evidence that they may form a monophyletic group together with the myriapods (Remington 1955, Handshin 1958, Melnikov 1974a, b, Rasnitsyn 1976, Melnikov & Rasnitsyn 1984, Shcherbakov, submitted) as explained in some detail below.
1.1.2. PHYLOGENETIC APPROACH
It may come as a surprise to many members of the western scientific community that cladistics has not been universally accepted as the one and only method to be applied to classifying organisms by entomologists and other systematists of the former Soviet Union. That is not to say that cladistics and parsimony analysis are not regarded as being very useful tools, or that groups should not be clustered on the basis of shared derived characters (synapomorphies) wherever possible. Rather, many believe that phylogenetic inferences are exactly the same as all other scientific statements in that they can never be proved nor rejected, but are for ever destined to persist as more or less likely hypotheses. Indeed, the inaccessibility of the final verification of a scientific statement has become almost a truism in post-Popperian times, and yet the final falsification of one statement is nothing more than the verification of an alternative statement, that is, the falsifying result is neither by chance nor due to unconsidered influences or circumstances. As a result there is no single best method for inferring phylogenies, such as the outgroup approach comparison praised by some authors. Contrarily, many methods are good, each in its own place, and their application can be efficiently regulated in the form of presumption. Thus much of the work described in this book relies on making presumptions. This is not a trivial matter because of the often incomplete nature of the evidence that scientists in general, and palaeontologists in particular, have to deal with. A presumption is a statement based on observations that a particular kind of result occurs more commonly in particular circumstances, and thus it is to be considered as most likely irrespective of the existence of confirmatory evidence, but not, of course, in the presence of reasonably sound contrary evidence (Rasnitsyn 1988a, 1992a, 1996, Rasnitsyn & Dlussky 1988). For instance, we should consider any similarity between organisms as inherited from a common ancestor and not gained independently (as homoplasy), unless and until strong contrary evidence is presented. This has been termed the presumption of cognisability of evolution, and it is equivalent to the auxiliary principle of Hennig (1966a). Another type of presumption that is particularly relevant to palaeontology (the palaeontological presumption) is that of two apparently closely related groups, the one entering the fossil record earlier should be considered as ancestral unless and until sound contrary evidence is presented. Likewise, we can recognise a biogenetic presumption, that is, a transformation series should be polarised in agreement with the ontogenetic succession of the respective character states; the outgroup presumption, that a character state found only within a group should be considered apomorphic in respect to that found both within and outside the group, and of course, the presumption of parsimony, i. e. that the most likely cladogram is that one necessitating the least number of homoplasies (i.e. the most parsimonious one). All of these presumptions are accepted below unless there is convincing evidence to the contrary. For a more thorough discourse on the use of presumptions in phylogenetics see Rasnitsyn & Dlussky (1988) and Rasnitsyn (1996).
For the most part, the authors in this volume try to follow the cladistic approach to phylogeny reconstruction (Hennig 1966a, etc.). In short, we appreciate that a clade should be considered definable by a synapomorphy, that is, an advanced character state hypothesised to have been acquired by the first member of the clade. The clade itself as the main notion in phylogenetics can be defined as a part of the tree-like figure representing relationships between groups of organisms, which is formed by cutting off a single ancestral line. Symplesiomorphy (similarity in a character state inherited from a more distant common ancestor) has no value in phylogenetics, as well as any kind of the homoplasy (independently gained similarities, including convergence and reversals, that is the convergence with an ancestral state of a character).
Another minor deviation here is that phylogenies based extensively on fossil information, in comparison with modern phylogenetic reconstructions prepared by students of living insects, pay relatively less attention to synapomorphies of detailed internal anatomy. Such characters are commonly simply unknown for the phylogenetically most interesting extinct groups, and even the living representatives have sometimes been explored very unevenly for these character systems.
A more important deviation results from understanding that the cladistic concept is a methodological claim rather than an ontological one. By this it is meant that cladistics makes available a very efficient method for the identification and, in more advanced its versions, for the calculation of relatedness, that results in enhanced objectivity and reproducibility of the results. At the same time, the basic hypotheses that either implicitly or explicitly form the basis of cladistics are not necessarily correct ontologically, that is, they do not necessarily agree with what we can observe in nature. Indeed, among the most important such hypotheses, there is the claim that each taxonomic group appears as a result of a divergence event accompanied by the gain of one or more apomorphy(ies), and that these groups are each destined to either die out, or to disappear in the next divergence event resulted in appearance of two (neither more nor less) new lines, each marked with their own apomorphies. Biological theory is aware of no mechanisms possible to secure these rules and to prevent, say the ancestral line not dying out when it bears a daughter line, but survives and to give birth to two or more further daughter lines. In other words, a polychotomy (the case of a line giving birth to more than two daughter lines) is not necessarily a case of incomplete knowledge: indeed it well might be effectively correct. Equally, no mechanisms are known to prevent a line from give birth to only one, and not to two daughter lines. Of course, it is a rare occasion when we have evidence for one or another of the possibilities discussed above. Nevertheless, this makes it evident that a cladistic concept is an hypothesis and not the final truth, and that a cladistic result does not mark the end of a study but rather is the material for further testing by different methods (Rasnitsyn 1996). The most powerful of these tests are various ways of comparison of the cladistâ€(tm)s cladogram with the palaeontological succession of fossils and their characters, e.g. tracing of the morphological transitions formed by fossils (e.g. the ghost range method; Rasnitsyn 2000a). Of course, these methods do not give us any ultimate truth either, but mutual hypothesis testing and correction is always beneficial.
It is clear from the above discussion that the principles of taxonomy followed here differ somewhat from those employed by many, if not all, current cladists (see Wiley 1981, Quicke 1993). This particularly concerns the interrelationship between the taxonomic system employed and the hypothesised system of relatedness between the organisms concerned (as derived from cladograms). The cladistic demand is to construct the first system so as to make it isomorphic to the second, as if the ultimate goal were to prepare the genealogical system of organisms. Thus, for most cladists only monophyletic taxa are appreciated as legitimate, with monophyly being taken in its narrowest sense, referring to a taxon which covers all descendants of the single first (oldest) its member. This kind of taxon has the single ancestral line and no descendant lines beyond its limits.
Because of the existence of another, older and broader understanding of the term monophyly which is no less use or popularity, a new term was coined for the cladistic version of monophyly, viz. holophyly (Ashlock 1971). This latter term is subordinate to monophyly sensu lato, which implies a taxon equally with the single ancestral line beyond its limits, but with no limitation imposed on its descendant lines. The second subordinate term of monophyly s.l. which is complementary to holophyly is paraphyly. A taxon is paraphyletic if it has a single ancestral line but has one or more descendant lines beyond its limits. So monophyly s.l. is definable through possession of only a single ancestral line beyond its limits. A taxon with more than one ancestral line is called polyphyletic irrespective of the presence of descendant lines. It is only monophyly sensu lato that is used throughout the present book, with the alternative notion, whenever employed, being termed holophyly.
Thus the purist cladistic approach to taxonomy has been largely abandoned here for several reasons (for details see Rasnitsyn 1996). First of all, reflection of genealogy is not the ultimate goal of taxonomy, even for many cladists, for the cladogram coupled with necessary explanations (technically termed scenarios) reflects genealogy much better than any nomenclatural system. Implicit here is the belief that the system must reflect the full balance of similarities and dissimilarities out of the totality of all possible characters including not only morphological, physiological, behavioural, aesthetic and so on, but also those that are already explored and those that are as yet completely unknown. Because of the obvious predestination, which history (as it is materialised in the realised type of organisation) imposes on the structure, functions and further evolution of an organism, history (in form of a genealogical scheme, a cladogram) has been hypothesised to be best correlated with the above mentioned total balance of similarities and dissimilarities. This inference, which has much truth, albeit not all the truth, was the starting point of cladistics.
The first and most important problem is that even in theory, good correlation between a cladogram and the total balance of similarities and dissimilarities is possible only under the condition of essentially equal evolutionary rates, which is not generally held. For example, the birds are usually considered to form a separate class, in spite of the fact that they are monophyletic with crocodiles (in terms of living reptiles) because they have been deviating (at least morphologically and physiologically) from the common ancestor of both at a much greater rate in comparison with the crocodiles.
Another problem arises when we appreciate that the cladistic system (that is one that is isomorphic with the cladogram) can be constructed only from taxa whose origin can be justified by the possession of at least one synapomorphy. In fact we could consider synapomorphy as ultimate evidence that a taxon has appeared, and a divergence event as a heuristic method permitting us to locate the proper place of the synapomorphy on the cladogram. Alternatively, we can use synapomorphy as a heuristic method to determine the sequence of past divergence events so forming a cladogram. Irrespective of this dichotomy, considerable uncertainty results from the absence of a precise correlation between the divergence event and the gain of a defining synapomorphy. As discussed above, one can take place without the other: a population of an ancestor can evolve into another species due to the gain of different synapomorphies, while the residual population remains essentially unaffected. And conversely, a synapomorphy can be gained in the course of so called phyletic evolution, that is without any divergence events. What is worse, there is no evolutionary mechanism known that would make any of the above processes rare. We should consider such cases as fairly common, and yet each of them makes impossible precise reconstruction of the cladogram and, as a result, the delimitation of the respective taxa.
One more problem with the cladistic taxon deserves mention. It is equally rooted in that the taxon can be legitimised by either a gained synapomorphy, or by the occurrence of divergence: both sorts of event are important but by only their absence or presence, not by any of their other features. That is why any synapomorphy and any divergence event is of equal value in the ranking of taxa. Therefore subordinated ranks must either be as numerous as the number of succeeding divergence events (conservatively counting a divergence event each million years gives hundreds of subordinate ranks), or such ranking should be abandoned and replaced by simple numbering (cf. Hennig 1981). Alternatively, taxonomic decisions can be arbitrary. Consequently, the claimed consistency of cladistic taxonomy is delusory, and its other advantages are equally problematic, thus permitting a chance for other taxonomic principles.
One alternative to the cladistic approach is the phenetic one which claims superiority because the classification relies on weighting of the raw similarity (for details see Sokal & Sneath 1973). This approach has its own advantages and deficiencies. At present it has been practically abandoned and can be recommended only as an auxiliary, albeit useful, tool of taxonomic study.
Another alternative was termed phylistic. It is similar to the traditional one but can be explicated as follows (Rasnitsyn 1996). Both similarity and relatedness are involved in a taxonomic system's organising principles, but their application is not arbitrary. In fact, the total balance of similarities and differences in all, even unavailable, characters is considered as the ultimate but directly unattainable goal to be reflected in the system. Raw similarity is used there as a heuristic method to construct (delimit) taxa while relatedness (monophyly) works as a method to test the resulting taxon. It is believed that the history (monophyly) is at least as suggestive concerning the ultimate goal of taxonomy as is raw similarity, and therefore, in case of contradiction (for example, when a taxon seems to be polyphyletic), it is recommended to retest the whole case on a wider basis (using more material and deeper analysis). Technically, the taxon is termed a monophyletic continuum, with monophyly defined as above (i. e. holophyly + paraphyly) while the continuum can be understood as a chain of taxa, either simple or branching, with every neighbouring pair of links being more similar to each other than to members of other continua. Thus taxa can be delimited by tracing the area of the least similarity, that is along hiati. Metaphorically, a taxon is a cloud of characters in a space, with the cloudâ€(tm)s integrity being of primary importance, and the particular characters of secondarily so ("Scias Characterem non constituere Genus, sed Genus Characterem", that is "Know, the character does not constitute the genus, but the genus the character"; Linné 1751, # 169).
Since Rohdendorf's proposal (Rohdendorf 1977, Rasnitsyn 1982), the thorough typification of all insect taxa including those of the highest rank is possibly one of the most striking features of the Moscow palaeoentomological school, and this practice is generally retained here. There are several reasons for this (Rasnitsyn 1996). The first is the old, but still valid, observation that a consistent order in the formation of taxonomic names makes the work of taxonomists easier. This applies equally to names of any rank, and the urgency for progress in the unification of higher taxon names can be still be appreciated even by entomologists (e. g. Boudreaux 1979), not to mention other zoologists (e. g. Starobogatov 1991), and by particularly botanists who have even changed their Code respectively (ICBN 1980). The second reason is based on another observation, namely that typification works well when applied to taxa of lower rank, and therefore should also be of help in respect to higher taxa.
Both of these reasons are clear and have been well known for centuries, indeed consistent typification of names of insect orders was first proposed 220 years ago; Laicharting (1781). Nevertheless, resistance to the proposed changes is strong, because a considerable cost would have to be paid in exchange for the advantages promised by comprehensive typification. The cost is the necessity to abandon many customary descriptive names and to learn new, typified names, and this will concern many more people than do routine nomenclatural changes that typically affect only tens of, and rarely hundreds of, taxonomists. Nevertheless, the cost is worth the price, and it will be paid sooner or later, for the cost of refusal is also high, and additionally is accumulating, with a consequent detrimental effect on taxonomic practice. This is because, contrary to the ICZN (p. xiii), nomenclature is not neutral in respect to taxonomy.
The present book is not the proper place to discuss details of the problem, for which see Rasnitsyn (1996). Only those aspects that seem more important concerning the issue of the typification of higher rank names will be briefly considered here.
"The name-bearing type provides the objective standard of reference by which the application of the name it bears is determined, no matter how the boundaries of the taxon may change" (ICZN #61a). The type principle is thus a method of introducing a taxon into the system. Indeed, although in practice we introduce it by means of a taxon diagnosis rather than by direct comparison with the type, the diagnosis must always agree with the characters of the type, and in cases of uncertainty only such comparison can finally resolve the issue. At the same time, application of the type principle is a rather sophisticated job, though the principle is nevertheless universally applied to lower ranked taxa, thus indicating that other possible ways to introduce a taxon into the system are not efficient. This is because the taxon-continuum (see above), being a cloud of subordinate taxa with highly unstable boundaries and content, cannot be effectively worked with unless its name is pinned by a type as a tag.
When we refuse to typify higher rank taxa, we should imply that it is because they either differ cardinally from that of lower rank and thus can be introduced in the system in another way, or they are so unimportant that any rigorous system of formation and application of their names is superfluous. The first alternative does not seem real, for there is no essential differences between lower and higher rank taxa that has been reported yet. Higher rank taxa are not particularly well characterised to be introduced just by referring to their characters. Nor do they exhibit especially high integrity which could permit us to consider them, unlike lower ranked taxa, as true individuals: this would make it possible to introduce the taxon into the system referring to its untypified (attributed to the individual as a whole) name.
As a result the inference appears inescapable that the common resistance to the thorough typification of taxonomic names in zoology rests on the belief, probably unconscious, that the names of the higher ranked taxa are of only secondary importance, and therefore that the problems born by arbitrariness in their application may be neglected comparing the labour which should be otherwise spent on learning new names. This myopic arrogance in respect to higher taxa is rather natural as a result of ongoing specialisation of taxonomists for lesser and lesser ranked taxa, coupled with the lowering prestige of taxonomy itself whose needs and demands meet the decreasing respect of other scientists and laymen. And yet such an approach is treating taxonomy itself, because the longer the disregard of higher level taxonomy lasts, the deeper will be its disorganising effect on lower level taxonomy. It is our appreciation that this danger is a real one which has forced us (i.e. the Moscow school) to continue using typified names of higher taxa, in spite of understanding that this will hardly increase the rating of this book, at least in short term.
In spite of a full understanding of the ultimate advantages of thorough typification, the painful proximate disadvantages of our hard adherence to its ultimate goal have forced us to soften our attitude in a way approved by botanists. They recently made exception for a short list of traditional, non-typified names that can be legitimately used, along with the typified ones, for a while at least (e.g. Umbelliferae together with Apiaceae) (icbn 1980). Similarly it was agreed among the contributors of the present volume, that several non-typified names should be permitted to be used herein, viz. Insecta, Odonata, Hemiptera, Coleoptera, Neuroptera, Mecoptera, Trichoptera, Lepidoptera, Diptera, Hymenoptera and Orthoptera. All other names are left typified (see Fig. 1 above).
Another topic worth discussing in respect to nomenclature is the use of parataxa. This problem is not unique to palaeontology but is particularly important therein. Parataxa are those groups created in direct violation of particular, taxonomy-dependent principles of nomenclature, as discussed in detail by Rasnitsyn (1996). The violation is inevitable when we have to classify insufficiently known organisms - whether it is due to their incomplete preservation, as is normal in palaeontology, or because the stage of development or of the life cycle or the sex at hand, bears no taxonomically important characters (as in some fungi and parasitic worms, as well as in some insects with pronounced sexual dimorphism).
The following kinds of parataxa have been proposed: taxon incertae sedis, formal taxon, and collective taxon. A taxon incertae sedis is a group that is properly housed at some taxonomic levels but not at some others. For instance, a genus incertae sedis is that for which only the encompassing order and not the family is identified as yet. The formal taxon is created for material with a particular type of taxonomic deficiency: it can be treated as a seemingly normal taxon when compared to other formal taxa of the same kind, e.g. genera of 'Fungi Imperfecti' (a group of fungi with the sexual stage of the life cycle unknown) or formal taxa of the detached beetle elytra [184.108.40.206.2.1]. However, they cannot be properly compared with normal taxa (orthotaxa). The weakest form of the parataxon is the collective group which is a collection of species differing from each other and possible to be assigned to an orthotaxon of any rank, but which are impossible to be either distributed among its subtaxa or to be organised in a system of parataxa of their own. For example, Carabilarva Ponomarenko is the assemblage of fossil beetle larvae that can be reasonably attributed to the family Carabidae but not to any of its particular subtaxa (Rasnitsyn 1996).