Unpublished for the purposes of zoological nomenclature (Art. 8.2, ICZN)
Traditional contributions of the insect fossil record are listed. Fossil material indicates the earliest occurrence of a group, which in turn is useful for inferring clade divergence dates and net diversification rates. Fossil material provides complementary information on the dynamics of taxonomic diversity. Geographical occurrences outside the extant range of a taxon can be used to infer climatic macro-fluctuations. In short, the fossil record of insects is essential for pointing out the major factors responsible for the mega-diversity of the group, and of some of its internal lineages. Reliable taxonomic assignments and phylogenetic hypotheses underpin broader generalizations. In that respect, a problem is the inadequate integration of data from fossil and extant insect taxa in phylogenetic investigations. Stumbling blocks lie at various systematic levels. Unreliability of specimen-based data, of species delimitation, and of homology assumptions, might have been responsible for a disdain by some entomologists for palaeoentomological literature. Idiosyncratic (and in cases flawed) methods aimed at investigating phylogenetic relationships used by a fraction of the palaeoentomological community might also have contributed to this situation. Concurrently, the traditional nomenclatural procedure might prevent effective communication between neo- and palaeoentomologists. Augmenting the available information on the wing venation of extant taxa would significantly advance palaeoentomology, and provide a relevant broad-scale character system. Furthermore, the entomological community should contribute to experimentations of various nomenclatural procedures, with the aim of developing an optimal approach in terms of communication and information retrieval.
The fossil record of insects is a valuable source of information for investigating the origin of the tremendous extant biodiversity of the group. It supplements data based on extant taxa in various ways. This contribution explores existing and potential interactions between these two areas of investigation.
The resulting paleontologically calibrated data can be coupled with extant species richness in order to estimate net diversification rates (Farrell, 1998; Mayhew, 2003), with the aim of identifying factors responsible for the mega-diversity of some lineages. It is then of prime interest for evolutionary biology. However, the consideration of actual fluctuations as observed in the fossil record (Labandeira & Sepkoski, 1993; Jarzembowski & Ross, 1996; Benton, 1997; Dmitriev & Ponomarenko, 2002) should be used to supplement results obtained from extant species richness figures, because the evolutionary history of insects is far from linear. For example, holometaboly has been regarded as a decisive factor in the mega-diversity of Holometabola. However, the fossil record indicates that holometabolous insects occurred during the Carboniferous (ca. 310 Mya; Nel et al., 2007), whereas actual taxonomic radiations (if so) of the main holometabolous lineages occurred at least 50 My later, during the Early–Middle Triassic (ca. 240 Mya; Grimaldi & Engel, 2005; Blagoderov et al., 2007). This example indicates that reliance on extant species richness alone can be misleading, because radiations could have been delayed, and diversifications interrupted by extinctions.
However, it must be acknowledged that available data on fluctuations of taxonomic diversity based on fossil material rely on debatable taxonomic frameworks [e.g. use of the paraphyletic and polyphetic assemblage ‘Protorthoptera’ in Labandeira & Sepkoski (1993), Jarzembowski & Ross (1996); abundant paraphyletic families in Dmitriev & Ponomarenko (2002)], and the debatable assumption that families (and any other taxonomic unit above the species level) are comparable units.
Addressing this evolutionary question is laudable, but the basic data need examination. The aforementioned investigations primarily rely on taxonomic assignments of fossil species. As a consequence of the fossil record incompleteness, such identifications are frequently based on a few selected character states, which have not always been subjected to a congruence test. Incorporation of fossil species in phylogenetic analyses is a preferable option. However, although fossil species can exhibit combinations of character states unknown among extant relatives, and therefore potentially affect phylogenetic hypotheses, the role devoted to this dataset is limited. Fossil species are seldom incorporated as outgroups (but see Rehn, 2003; Vilhelmsen, 2007; Bybee et al., 2008) or as part of the ingroup (but see Andersen, 1989; Andersen & Poinar, 1998; Pohl & Beutel, 2005; Pohl et al., 2005; Hippa & Vilkamaa, 2006). Exceptions are produced mostly by palaeoentomologists. In other words, data integration in phylogenetic investigations, clamoured for by Wheeler (2008), is still lacking. Issues that might be responsible for this lack of data integration, and their possible remedies, are considered below.
The assumption that fossil material should be discarded because of its incompleteness is briefly addressed. The reliability of basic data, alternative methods of phylogenetic investigations, and nomenclature, are considered as more substantial issues. Collaborative projects that would allow a better integration of information are pointed out. Some preliminarily conventions are proposed.
Conventions used in the following regarding crown- and stem-groups, and nomenclature, need definition. A crown-group is defined as all extant species of a given group, their common ancestor, and all its descendants. A crown-group can then include fossil species (e.g. ‘crown-A’ on Fig. 1B). The stem-group comprises fossil species that are more closely related to this group than to any other extant group, and that are not part of the crown-group (Hennig, 1981: 31). It is, by definition, exclusively composed of fossil species, and is paraphyletic if it includes at least two distinct monophyletic units (species or taxa; Fig. 1). The total-group includes both the stem- and the crown-group. These conventions are widely accepted in the palaeontological literature.
Whether a stem-group fossil species should be included within the group, or excluded from it, may cause confusion. Should a group be extended so that to include its stem-species, or not? How taxa are conceptualized is connected to this point. Let us assume that a non-holometabolous fossil insect species is more closely related to Holometabola than to any other extant lineage. It is then a stem-Holometabola, but should it be included within Holometabola? This issue is tricky under the traditional nomenclatural procedure, because taxa are not unambiguously conceptualized [especially under the International Code of Zoological Nomenclature (ICZN) above the family group]. Coping with this issue is the particular advantage of using the qualifiers ‘stem-’, as it only makes reference to a phylogenetic branching pattern, and a number of species ‘extant’ at some (unspecified) time and (somehow) conceptualized altogether as a taxon. Under the traditional nomenclatural procedure: (i) crown-Holometabola is composed of extant Holometabola; therefore, all are holometabolous, so is their common ancestor; (ii) stem-Holometabola is composed of non-holometabolous and/or holometabolous fossil species. The question ‘should non-holometabolous species related to Holometabola be included in Holometabola?’ is nonsensical because how ‘Holometabola’ is conceptualized (or indeed the meaning of the word ‘Holometabola’) is not clear in the first place. The qualifier ‘stem-’ is a stopgap solution as it allows the ambiguity taxon conceptualization inherent in the traditional nomenclatural procedure to be circumvented when communicating about fossil species (see also application of taxon names ‘A’ and ‘B’ on Fig. 1B).
Regarding conventions in nomenclature, most taxon names used in this contribution are the result of the application of the traditional Linnaean procedure. This choice was prompted by a lack of a sufficiently developed nomenclature based on an alternative procedure, and does not imply the author's support for the traditional procedure. Names erected under the cladotypic procedure (see references below) are italicized, and preceded by the word ‘taxon’, in order to differentiate them from generic names (themselves preceded by the word ‘genus’). For the sake of discussion on nomenclature, species are referred to by uninominal species names coupled with authorship information, according to Dayrat et al. (2004). Figure captions refer to names under the Linnaean approach.
An early step that could frustrate a neontologist willing to incorporate fossil species in a phylogenetic analysis concerns primary data on fossil insect material. Frequently conflicting interpretations of the same fossil material occur in the literature. Beyond inaccuracy of observation, some palaeoentomologists might have a distinctive propensity to extrapolate from the actual material. However, apart from one particular case evidenced by Béthoux & Briggs (2008), most observations subject to controversy date back to late 19th-century and early 20th-century literature. A number of species have been left in need of re-description for about a century, especially those assigned by Carpenter (1992) to the problematic ‘Protorthoptera’ wastebasket (Grimaldi, 2001). This issue is related to the comparatively limited size of the palaeoentomological community. Nevertheless, revisions of inappropriately described material are steadily produced. Contributions by neontologists at re-interpreting (Dalgleish et al., 2006) and re-re-interpreting (Smith et al., 2007) fossil material are also valuable. In addition, in order to improve the quality of specimen-based contributions, experiments and applications of innovative illustration techniques, such as surface laser scanning (Béthoux et al., 2004) and X-ray micro-tomography (Grimaldi, 2003; Lak et al., 2008) are carried out (see also Quetscher & Ilger, 2007). The unreliability of specimen-level data is less and less diagnostic of the palaeoentomological literature.
Reliable delimitation of fossil species matters for (cladistic) phylogenetic investigations, because incorporating distinct terminal taxa that actually belong to a single species, i.e. have reticulate relationships, might alter the result of phylogenetic analyses. It must be acknowledged that the typological species concept plagues palaeoentomological literature. Cases of species erected without attention to intra-specific variability abound, especially for compression material (as opposed to amber material), for which wing venation provides most of the available information. In such cases, conspecific individuals can be identified when fossil specimens exhibiting a pair of wings in connection are available (Schneider, 1983; Béthoux & Nel, 2003; Béthoux, 2008b). If so, the range of intra-specific variability can be inferred from that of intra-individual variability. Abundant material can also help to solve such cases (Zessin, 1987). An additional option is to consider wing venation variability of related extant taxa (Schneider, 1977), but available data are limited in the neoentomological literature. For example, among articles published in the journal Illiesia dedicated to stoneflies, wing venation is illustrated for two of the 77 newly described extant species for which adults are provided with wings (at the time of writing). The few works on venation variability of extant taxa at the species level have been carried out mostly by palaeoentomologists (e.g. Yablokov et al., 1970; Schneider, 1977; Makarkin, 1995, 1996; but see, for example, Akahira & Sakagami,1959; Devetak, 1991; Walia, 2004). An insect wing collection, composed of wings of extant insects mounted on slides, obviously provides a solution (Béthoux, 2008a).
As for specimen-level data, alternative homology hypotheses of structures observed in fossil taxa can be found in the literature. This is especially true of the wing venation [e.g. for Dictyoptera, see Deitz et al. (2003: table 1); for Orthoptera, see Béthoux & Nel (2002: tables 1, 2)]. This issue is relevant at the super-ordinal level, as contributions consistent at the ordinal level are available [see Sharov (1968), mostly involving fossil taxa; see Needham & Claassen (1925), Ragge (1955a, b), Rehn (1951) mostly involving extant taxa]. However, a consensus has not been reached on wing venation homologies across Neoptera or Pterygota (compare Gorokhov, 2005; Rasnitsyn, 2007; Béthoux, 2008c), preventing the use of fossil species in phylogenetic investigations at this level.
As for intra-individual variation, the insect wing collection appears a relevant solution (Béthoux, 2008a). Recent investigations based on such material demonstrated that, in adult wings, the pattern of tubular sclerotizations usually referred to as ‘venation’ can be distinct from the tracheal pattern. It entailed a re-evaluation of homology hypotheses based on the former only (Béthoux, 2005b; Béthoux & Wieland, 2009). In addition, a more detailed knowledge on actual wing venation transformations, based on extant taxa, would be valuable to palaeoentomologists. For instance, it is now assessed that the course of a main vein can be altered in such a way that it ‘follows’, or ‘captures’, a pre-existing cross-vein (Béthoux, 2005b). In addition, investigations on the wing venation of Mantodea (Béthoux & Wieland, 2009) revealed that ‘vein translocation’, defined as the fusion of a vein with another from the base of the latter, is a plausible transformation. Extant material allowed a complete transformation series to be documented (Fig. 2), with the occurrence of a polymorphic state involving vein and trachea reticulations (Fig. 2D). This discovery entailed a concurrent re-investigation of the forewing venation of the fossil taxon Titanopterida, interpreted as the result of successive translocation events (Béthoux, 2007c; Fig. 3). This option was not considered by previous authors because the ‘transformation series’ is recorded too incompletely in the fossil record, making translocation events hardly traceable (indeed not even considered). In turn, the new interpretation allowed the debate on the phylogenetic position of Titanopterida to be closed: as stated by Sharov (1968), these insects derived from a lineage of Permian Orthoptera (Béthoux, 2007c), contraGorokhov (2001) and Béthoux (2005a).
The amber record is comparatively immune to the issues highlighted above, because its information content is higher: more characters of external morphology are preserved, sometimes exquisitely [e.g. Aspöck & Aspöck, 2004; Amorim & Grimaldi, 2006; Azar et al., 2008; see also Grimaldi & Engel (2005) among many other contributions] in addition to behavioural features (Poinar, 1993; Arillo, 2007). Preparation techniques aimed at extracting fossil material from amber allow scanning electron microscopy observations (Grimaldi et al., 1994; Azar, 1997). In addition, the relatively young age of the amber material makes direct comparison with extant material more relevant than for the pre-amber record. Unlike for the compression record, there is no significant information gap preventing the development of reliable homology hypotheses among extant and fossil species, because external morphology other than wing venation is routinely described in the neoentomological literature.
Methods of phylogenetic investigation and nomenclature
Assuming optimal specimen descriptions, accurate species identifications, grounded and testable homology hypotheses, a sufficient number of characters and a good sample of character state combinations, the way most neoentomologists investigate phylogenetic relationships differs from methods used by some palaeoentomologists. For instance, the ‘groundplan approach’ advocated by Kukalová-Peck (2008) favours a mental computation of a groundplan prior to phylogenetic analysis. This option is considered as circular (Béthoux et al., 2008)—hypotheses are built on hypotheses rather than observations—and is not widely followed.
However, the methodological discrepancy has limited implications. Although several groups identified under the method defended by Rasnitsyn (2002a) might be difficult to use for cladist neoentomologists because of nomenclatural issues (see below), synapomorphies open to discussion and congruence tests are provided for many groups in Belayeva et al. (2002), the largest contribution making use of phylistics. This important contribution by proponents of phylistics cannot be discarded just because of methodological discrepancies. In addition, taxonomic discussions and phylogenetic analyses carried out by palaeoentomologists using the cladistic paradigm, and relevant for modern taxa, are available (e.g. Willmann, 1987; Engel, 2001; Grimaldi & Engel, 2006; Blagoderov et al., 2007; Engel & Grimaldi, 2007; Azar et al., 2008—and see Grimaldi & Engel, 2005).
Nonetheless, phylistics is problematic in that paraphyletic stem-groups are given formal taxon names, even if they possess no genuine apomorphy. For example, the taxon Eoblattida possesses no apomorphies and was established instead ‘because of supposed stem position of the order in respect to other Gryllones (= Polyneoptera)’ (Rasnitsyn, 2002b: 256). In other words, ‘Eoblattida’ is a synonym of stem-Gryllones, and therefore the concept ‘Eoblattida’ is arguably uninformative with respect to ‘stem-Gryllones’. The proliferation of such names as proposed in Belayeva et al. (2002) surely results in a blurring of relevant information.
The issue can become troublesome to a cladist once ranks are considered. For example, in Belayeva et al. (2002), a noteworthy case is provided by Rasnitsyn (2002b) and Vršanský et al. (2002) (see a ‘cladistic-inspired’ representation on Fig. 4). According to Rasnitsyn (2002b; personal communication, 2009) (contraBelayeva et al., 2002: fig. 1), the super-order Blattidea (= Dictyoptera) includes cockroaches (order Blattida), termites (order Termitida) and praying mantids (order Mantida). According to Vršanský et al. (2002: 265), Blattida is paraphyletic ‘in respect to mantises and termites’. In other words, stem-Blattida are stem-Blattidea, and some extinct Blattida are stem-Termitida and stem-Mantida; in other words, an order includes two other orders. In addition, the super-family Phyloblattoidea is part of the stem-Blattida, and is paraphyletic, as some Blattida, Termitida and Mantida derived from the lineage from which Phyloblattoidea derived. Within the latter, the family Phyloblattidae is paraphyletic: it includes a part of Blattida, and Termitida, and Mantida. Finally, a family includes two orders and a part of a third one. This is problematic to one used to seeing families included hierarchically within orders, and not the opposite.
If the attribution of ranks appears as lawless, the application of the approach is performed in such a way that supernumerary ranked names are avoided. Following the rule that sister taxa should be of the same rank, and positing that Termitida is an order, would require an inflation of names for naming stem-Blattidea (with at least six orders according to the phylogeny presented in Fig. 4). As a matter of fact, Rasnitsyn's (1996, 2002b) approach exemplifies the difficulty in applying the rule that sister taxa should of the same rank.
Indeed, under a cladistic/phylogenetic application of the traditional Linnaean approach, a palaeontologist describing a new species belonging to the stem-group of an extant taxon could erect a series of new monotypic taxa (at least a genus) including the new species. This is grounded especially if the new fossil species exhibits derived features (e.g. Aspöck & Aspöck, 2004; Nel et al., 2005; Wappler & Petrulevičius, 2007; Kirejtshuk et al., 2009; among many other contributions). Further discoveries will then follow this pathway, with each new stem-group species being assigned to its own monotypic taxon of high rank, until phylogenetic relationships are investigated (resulting either in a broad synonymy, or in the erection of taxa of higher ranks). Arguably, this approach can complicate the work of neontologists: identifying relevant species among a plethora of monotypic high-rank fossil taxa can deter incorporation of fossil species in phylogenies.
Another nomenclature-related issue is the lack of rules governing taxon names above the family group in the ICZN. Again, termites constitute a relevant example. A consensus is now reached about the subordination of Isoptera, considered previously as an order, within Blattaria (or Blattodea, or Blattidea, or Blattida) (Deitz et al., 2003; Lo et al., 2003; Inward et al., 2007), itself an order. How to cope with this issue under a formally ranked nomenclatural procedure is unclear. The case can be fairly considered as undecided at the moment (compare Lo et al., 2007; Eggleton et al., 2007).
Arguably, nomenclature is a source of substantial dissonance, but some options might provide solutions. Regarding the phylogenetic position of fossil species, the combination of uninominal species names with taxonomic addresses (Dayrat et al., 2004) might facilitate information retrieval. For example, the combination ‘Lestinoidea germanicaWappler & Petrulevičius, 2007’ provides exhaustive information on the phylogenetic affinities of the species germanicaWappler & Petrulevičius, 2007: it belongs to the taxon Lestinoidea, but cannot be assigned to any less inclusive taxon at the moment. The erection of a number of redundant taxon names (in our example the monotypic genus Priscalestes and family Priscalestidae) becomes unnecessary.
A number of authors have proposed alterations of the ICZN in order to solve the issue of animal taxon names above the family group (Rohdendorf, 1977; Alonso-Zarazaga, 2005; Dubois, 2006). Among other initiatives, Nixon et al. (2003) proposed that taxa could be named without being associated to a rank and intercalated between ranked taxa, in order to avoid the proliferation of taxa of high rank. The opinion that the traditional Linnaean nomenclatural procedure, as governed by the ICZN, is not the most optimal one, prompted the development of approaches that are more fundamental alternatives, such as that governed by the PhyloCode (Cantino & de Queiroz, 2007), and the cladotypic procedure (Béthoux, 2007a, b). Under these approaches, taxa are defined or taxon names are associated with a definition, respectively. Although the usage of ranks is not prohibited, it is not encouraged under these procedures.
Connecting fossil and extant insects
In summary, from my limited perspective on the topic, I identify two points that would need a decisive input and could involve collaboration among palaeo- and neoentomologists. First, there is a lack of interest (if not disdain) for wing venation pattern by a part of the latter community. However, it can hardly be maintained that wing venation carries limited and/or inconsistent phylogenetic information. Although this can be true at low taxonomic level, and for groups with a uniform wing venation (Pilgrim & von Dohlen, 2008), there are examples of highly complex wing venation transformations allowing robust phylogenetic inferences (e.g. Béthoux, 2007c). Furthermore, it is not totally surprising that wing venation characters could be homoplasic if fossil species are not considered, as in Ware et al. (2007). Just as in other character systems, wing venation does not provide relevant information at all taxonomic levels, for all groups, and if taxon sampling is incomplete.
The resulting lack of information has been detrimental to palaeoentomology. To some extent the persistence of wastebasket groups such as ‘Protorthoptera’ or ‘Eoblattida’ is the result of this lack of data. A digital insect wing collection widely accessible would allow palaeoentomologists to better circumscribe species, develop wing venation homology hypotheses testable on a broad scale, and properly identify stem-groups (Béthoux, 2008a). The prominent vertebrate palaeontologist Georges Cuvier demonstrated the importance of such collections for fossil vertebrates a couple of centuries ago. Large collections of cleared leaves of extant angiosperms provided a wealth of characters useful for the identification of fossil species, resulting in significant outbreaks in the study of the evolutionary history of the group (Dilcher, 2000). It is my opinion that entomology needs to be endowed urgently with such a facility, the development of which I envision as a freely accessible internet database, contributed by individual entomologists.
Second, it is clear that the ICZN, as currently designed, prevents efficient communication and information retrieval among entomologists. A ranked nomenclatural procedure is of uneasy use to palaeontologists, because they routinely unearth species that are sister groups of extant taxa of high rank. A traditional application of the procedure results in an inflation of ranks and of taxon names, and/or a blurring of information required by neontologists. In addition, binominal species names are chronically unstable [see van der Linde et al. (2007) and associated comments on the Drosophila melanogaster vs Sophophora melanogaster case], and do not warrant independence between taxonomy and nomenclature, clamoured for by ‘linnaeists’ (e.g. Dubois, 2008). Dogmatic positions on the prevalence of the traditional procedure, as currently designed in the ICZN, can no longer be afforded.
The community of entomologists as a whole could contribute to the current debate on nomenclature. The field is particularly well suited, with experiments on virtually all options apart from the PhyloCode. Benefits are expected from what I identify as a scientific crisis (sensu Kuhn, 1996): either the traditional procedure will be successfully implemented, or an alternative and more optimal approach will be preferred. It is my belief that attempts at databasing biodiversity, such as Zoobank (Polaszek et al., 2005; Pyle & Michel, 2008), should be prepared to comply with alternative approaches before massive efforts are undertaken.
I gratefully thank the editors of Systematic Entomology for inviting me to provide this contribution to the ‘Fossil insect’ virtual issue, and for their careful review and pertinent comments. I thank Professor Dr J. W. Schneider (Freiberg University, Freiberg) for valuable comments on an earlier version of this contribution. I am grateful to Dr G. Edgecombe (Natural History Museum, London) and Professor P. Štys (Charles University, Prague) whose comments and opinions contributed to the improvement of this contribution. The author is a research fellow of the Alexander von Humboldt Foundation.