Après ce constat, il s'avère que l'approche la plus sensée est de se concentrer sur des groupes dont la monophylie est établie, et dont la taxonomie est remise à jour. C'est le cas des clades super-ordinaux des Archaeorthoptera (sauterelles, grillons) et des Odonatoptera (demoiselles, libellules), dont les études systématiques sont en cours.

Malgré un registre fossile incomplet, il est possible de mettre en évidence des tendances communes chez les Archaeorthoptera et les Odonatoptera entre le Permien et le Trias. Après l'apparition et la diversification de certains groupes au passage Carbonifère–Permien, certains d'entre eux s'éteignent au passage Permien–Trias et sont relayés, au Trias, par d'autres groupes, sous-représentés auparavant. Le cas des Coleoptera et des Diptera est comparable. Comme la même dynamique évolutive se retrouve chez des clades phylogénétiquement et écologiquement distants, il est vraisemblable qu'un ou plusieurs facteur(s) extrinsèque(s) ai(en)t affecté leur évolution d'une manière similaire. La crise biologique permo-triasique représente peut-être un de ces facteurs.

Ces modifications ont eu lieu dans l'intervalle de 15 Ma qui sépare les enregistrements les plus proches du passage P/T. Leur contemporanéité et/ou leur soudaineté ne peuvent être démontrées, en raison de lacunes dans le registre fossile.

Une approche basée sur les concepts de diversité écologique pourrait permettre d'aborder la question de la restauration des écosystèmes après le passage P/T. L'entomofaune du Grès à Voltzia serait particulièrement intéressante dans cette perspective, car c'est la plus proche du passage P/T qui soit suffisamment riche et diversifiée. Cependant, l'analyse taphonomique proposée par Gall [26], qui permettrait de corriger le biais inhérent à l'échantillonnage fossile, n'a pas concerné les insectes de ces gisements. Aucun gisement permien qui pourrait servir de référence n'a été étudié dans cette perspective. De sérieuses avancées sont encore nécessaires dans ce domaine.

Bien que l'enregistrement fossile des insectes au Permien et au Trias soit incomplet, certains groupes permettent néanmoins de mettre en évidence des dynamiques évolutives comparables, au passage Permien–Trias. Chez tous les groupes étudiés, des clades inconnus ou sous-représentés dans les gisements connus du Permien supérieur se sont certainement rapidement diversifiés au Trias inférieur ou à la fin du Permien, au vu de leur importante diversité au Trias moyen. De toute évidence, l'importance relative de ce renouvellement reste à tester. Il est indéniable que la lacune paléoentomologique des quinze millions d'années autour du passage Permien–Trias constitue une période majeure dans l'histoire des insectes. Mais il serait important de s'assurer que le passage Carbonifère–Permien, par exemple, n'a pas été aussi important dans l'histoire du groupe.

Our current knowledge of patterns of insect diversity through time is based on the enumeration of taxonomic units within the class as a whole, or within super-ordinal or ordinal groups. The most widely used unit in such compilations has been the family [22], [33], [34] and [41], but a number of factors make the use of family level data unsatisfactory, particularly in the investigation of patterns of insect biodiversity during the Palaeozoic.

The main source of family data has been the volumes of the Treatise on Invertebrate Paleontology [18]. This treatment pre-dates the widespread use of cladistics and, even though it involved the lifetime work of a single author, does not apply the concept of a family uniformly throughout. Some Treatise families date to 1885 [16], when taxonomic practices differed from those today. Consequently, a number of these taxa are synonyms or paraphyletic, possibly causing problems in the analysis of palaeobiodiversity [61]. The proportion of data that is unreliable is unknown, but it was significant enough to support the abandonment of the previous familial frame (see [13] and [56] on families of the order Caloneurodea; [9] for ‘Protorthoptera’). In other words, the fundamentals of our current understanding may be biased by an analytically incorrect database.

A good example of the familial definition issue is provided by Dmitriev and Ponomarenko [22]. They synonymised 212 of the 1261 insect families recognized by previous authors, a reduction of nearly 20%. But they did not discuss the basis for their taxonomic decisions, and they continued to use paraphyletic taxa (at familial, super-familial, ordinal, and super-ordinal levels). Thus, their new taxonomic scheme may not be a significantly better basis for investigating trends in biodiversity.

A more fundamental issue concerns the assumption of the comparability of the families. Taxa of familial and higher taxonomic ranks are usually established based on morphological ‘difference’. A recent example is provided by the order Mantophasmatodea [37]. It is based on a unique combination of morphological traits, but contains just 12 genera [24] and [37]. Thus, the rank attributed to this taxon does not reflect its taxonomic diversity but morphological distance between it and previously erected orders. Hence, the family rank unit is not strictly related to the taxonomic diversity, but is rather a subjective index of morphological disparity. Since the two components of biodiversity, the taxonomic diversity and the morphological disparity, can provide very different signals [25] and [49], new approaches should be developed and applied in order to better discriminate between them.

From this point, one can suggest that instead of attempting to cover the evolution of the whole class across the Permian/Triassic boundary, it would be more meaningful to extensively review significant monophyletic groups.

There is no consensual phylogeny of the insect class including the fossil orders. The proposals of Kukalová-Peck [40], Grimaldi [30], and of the Russian team [2] are strongly conflicting (see also [33]). Consequently, the exact relationships of several fossil orders in respect of modern orders are yet a matter of debate. Even the validity of several orders is argued. For example, there is no consensus on the exact relationships of the non-archaeorthopterid ‘Protorthoptera’, i.e. Protoperlaria sensu Tillyard [65], i.e. Grylloblattida sensu Storozhenko [64]. The group might include – or be included in – the orders Plecoptera (stoneflies) [30] and [65], Grylloblattaria [64], and/or Embioptera (webspinners) [66] (see also a discussion in [14]). Each of these hypotheses implies a different evolutionary history during the Late Palaeozoic and the Early Mesozoic. Since the group represents 20% of the species recognized in the Permian Wellington Formation (Kansas, USA; see http://www.windsofkansas.com/elmodiv2.html), the lack of consensual data on its relationships is an important issue.

Although measurements of the quality of the fossil record of insects are still missing (but see [70]), there is some evidence that it could be improved. The faunas newly investigated are suitable empirical tests of the quality of our knowledge.

The Permian fauna of the Lodève Basin (Hérault, France; Permian) has been systematically studied for some years [50]. Three discoveries show that our current record of Permian insects is patchy. Among the Odonatoptera, the family Lapeyriidae represents the closest sister group of the lineage of Nodialata and the more inclusive lineage of Odonatoptera having developed a true nodus [50]. The family Saxonagrionidae is the oldest representative of the Panodonata, viz. the stem group of the Recent Odonata, e.g., damselflies and dragonflies [51]. Among the clade Archaeorthoptera (e.g., related to grasshoppers and crickets), the family Permostridulidae is the only known stridulating member of the order Caloneurodea and, incidentally, the oldest evidence of an insect sound production [12]. Béthoux et al. [13] demonstrated that this family, among the Caloneurodea, has very primitive traits in wing venation, apart from the stridulatory apparatus. Since more derived members are recorded in Late Carboniferous, it implies that the most basal representatives are, up to now, unknown. This gap is related to the lack of Early Carboniferous entomofaunas, when insect basal cladogenesis events likely occurred [5], or had already occurred [23].

Phylogenies of major fossil insect clades are generally unknown. Hence, it is impossible to provide a global measurement of the quality of the fossil record of insects using appropriate indices [4]. However, some examples are significant. Wills [70] reported several ordinal ghost ranges exceeding 100 Myr. Recent hypotheses on the phylogenetic relationships within the clade Archaeorthoptera suggest that the family Geraridae and the order Titanoptera are sister-groups [6] and [28]. This implies a gap covering the Permian period (but new material remains to be described, see below). In Odonatoptera, the discovery of Bechlya ericrobinsoni Jarzembowski and Nel, 2002, the oldest representative of the Protozygoptera, in the Late Carboniferous (Westphalian) of England, implies a significant gap in the record of Nodialata covering all the Stephanian and Early Permian [35].

Historically, the Late Permian deposits that have been the most intensively studied are located in the region of Perm (Russia), Central North America, Moravia. Those from the Triassic are from France, Central Asia, Australia, Southern Africa, and North America. Very wide areas are still poorly known (China, India, South America, main part of Africa, etc.). Due to this sampling bias, the fossil insect record cannot adequately represent the world insect diversity for these periods.

The clade Archaeorthoptera belongs to the Polyneoptera, i.e. is more or less related with the Dictyoptera (cockroaches, mantises, termites), the Plecoptera (stoneflies), the Dermaptera (earwigs), the Phasmida (stick insects), and some more cryptic orders [68] and [69]. As currently known, it includes mainly the extant order Orthoptera (grasshoppers, crickets) and the extinct orders Caloneurodea and Titanoptera [8]. The assignment of the order Caloneurodea to the Archaeorthoptera has been recently ascertained [10] and [13]. Besides the recognized orders cited above, the clade encompasses a part of the ‘Protorthoptera’ group (see [9] for a recent update). Membership of Archaeorthoptera is established based on complex characters of the wing venation [7] and [8].

Although the phylogeny of the Archaeorthoptera will require additional taxonomic investigations, several Permian and Triassic groups are commonly considered as monophyletic, i.e. the extinct orders Caloneurodea and Titanoptera, and the order Orthoptera. Within Orthoptera, the sub-order Ensifera (crickets) is well circumscribed and rooted [11]. The composition of the clade Panelcanida [8], present essentially in the Triassic, e.g. in the ‘Grès à Voltzia’ in northeastern France, requires further investigations and its trends will not be discussed herein.

Considering the Permian as a whole, the Archaeorthoptera clade is roughly composed of 70% of species belonging to the Orthoptera, 20% of species belonging to the Caloneurodea, 10% belonging to remnants of Carboniferous groups (including an inferred species of Titanoptera) ([6]; total of 66 species recognized). The Triassic profile of the Archaeorthoptera is clearly different (mainly compiled from the deposit of Madygen, Russia; total of 51 species recognized): the order Caloneurodea is missing, and the Titanoptera encompasses about 25% of the species. No major changes affect the order Orthoptera as a whole, which encompasses 75% of the species (70% in Permian) (but see below its inner composition – Fig. 1).

The extinction of the order Caloneurodea, though apparently not critical in terms of global taxonomic diversity because of its low diversity, is very significant in terms of morphological disparity: its flight mode was unique within Archaeorthoptera. Its representatives possessed two pairs of identical slender wings, which are typical of a bimotorist flight mode. Moreover, at least one Permian family acquired a stridulatory apparatus [12]. All these features depict insects that are very different from other Archaeorthoptera.

Besides the extinction of the Caloneurodea, the most striking modification is the new proportion raised by the sub-order Ensifera, with 25% of the total Triassic archaeorthopterid species (5% in Permian). This ratio is conservative, since several of the Triassic ensiferan species described by Gorochov [27], from the deposit of Madygen, are herein considered as synonyms, because the effect of post-depositional deformation was overlooked by this author (O.B., pers. obs.).

The cladogenesis events that led to the diversity of the Ensifera can be dated with the best possible resolution around the P/T boundary. The first known true Ensifera is Raphogla rubra Béthoux et al., 2002 (Fig. 2a), from the Lodève basin, i.e. one of the Latest Permian deposits that yielded Archaeorthoptera [11], the precise age of which is ranging from Ufimian to Tatarian according to Schneider and Roscher, 2001, http://www.geo.tu-freiberg.de/palaeo/. The Earliest Triassic Ensifera is Galliagryllavus vogesiacus Marchal-Papier et al., 2000, from the ‘Grès à Voltzia’ Early Anisian (France) [46] (Fig. 2b), i.e. the earliest known Triassic deposit that yielded Archaeorthoptera. Unfortunately, the two deposits are separated by at least 15 Myr. No more precise resolution can be currently reached. Since the Triassic Galliagryllavus belongs to the derived clade Gryllida [8], it implies that the most basal cladogenesis events within the Ensifera occurred previously. Although 15 Myr separates Galliagryllavus from Raphogla, we can at least state that a group that constituted about 25% of the Triassic archaeorthopterid species evolved during this gap. Naming this event a ‘radiation’ is a matter of scale.

The order Titanoptera deserves a special mention. It is recorded in only two Triassic deposits (Madygen, Russia; New South Wales, Australia), but two fragmentary specimens, one from the Late Permian of South France and the other from the Triassic of the Vosges (France) remain to be described. The order is very rare in the fossil record and might have narrow palaeoenvironmental preferences. It is rooted in the Carboniferous and is practically unknown during the Permian. Finally, despite an apparent rather high diversity in the Triassic, any attempt to draw ‘titanopterid’ evolutive trends is clearly premature.

In conclusion, the Triassic Archaeorthoptera differ significantly from the Permian ones by the absence of the order Caloneurodea and the apparent sudden rise of the sub-order Ensifera.

Before the recent discovery of the oldest Nodialata in the Late Carboniferous [35], it was supposed that the Carboniferous and Early Permian Odonatoptera belonged to basal ‘pre-nodialatan’ groups (Meganisoptera), and that the diversification of the group occurred later during the Late Permian. The presence of Bechlyia in the Carboniferous implies that the most basal cladogenesis events within the Odonatoptera are yet unrecorded (probably occurring during the Early Carboniferous – Fig. 3), but also that the rise and diversification of the Nodialata cannot be reliably correlated with a Late Permian–Early Triassic event.

Several other recent discoveries demonstrated that the diversity of the clade Odonatoptera was very important during the Permian. The ‘old’ famous group of Meganeuridae (commonly referred to ‘giant dragonflies’) was flourishing, even more than in the Carboniferous, with no less than four species in the Late Permian of Lodève, which have a wing size ranging between 4 and 30 cm (A.N., pers. obs.; Fig. 4a). This morphological disparity corresponds to a wide range of habitats. The highly specialized group of Protanisoptera is still strictly known from the Permian, with numerous families and genera, and wide morphological disparity [32]. Interestingly, this group is still unknown in the Late Permian of Lodève, while it is present in the localities of nearly the same age and close palaeogeographic location (North America), perhaps due to palaeoenvironmental preferences of the group. The Protozygoptera were present and very diverse during the Permian and Triassic (Madygen, Kirghisistan, Grès des Vosges, Northern France), with no decrease in diversity or morphological disparity around the boundary between the two periods. Until recently, the very wide group of the Triadophlebiomorpha was only known from Triassic deposits (Madygen Formation and ‘Grès des Vosges’), but a recent discovery demonstrated their presence in the Late Permian [52] (Fig. 4b). The first representatives of the modern clade Odonata (zygopteran and anisopteran lineages) are known from the Late Triassic but, as they belong to already very advanced subgroups, the group is certainly older and probably appeared during the Permian, as the oldest Panodonata (a clade including the Odonata and its closest sister-groups) is Late Permian [51]. Nevertheless, the Panodonata and Odonata were certainly less diversified during the Late Permian and Early Triassic than in the Late Triassic.

In summary, the recent discoveries demonstrated that our knowledge of the Permo-Triassic fauna of Odonatoptera is fragmentary, but some patterns can be reliably outlined. It appears that no less than six main clades were present in the Late Permian, but three of them were not found in the Triassic deposits (Meganisoptera, Protanisoptera, Lapeyriidae). Thus, the change in the fauna of Odonatoptera is evident. However, the timing is poorly constrained as the fossil record is extremely scarce in the Late Permian and Early Triassic.

The Coleoptera is a dominant order in terms of species richness in the Early–Middle Triassic of the Vosges, the oldest known formation that provided diversified Coleoptera, and in later periods [45] and [55]. This group is present in all the Triassic and more recent deposits. The most ancient Coleoptera are recorded from the Permian, though it has never been recorded in some Late Permian localities (e.g., Lodève Basin), where it was either absent or of minor importance [55]. Similarly, we could hypothesize that they were not yet adapted to some special Late Permian environments. On the contrary, this group has been widespread in all continental environments since the Mesozoic and Cenozoic, as the most important component of the entomofaunas. Its morphological and taxonomic diversification probably took place between the Late Permian and the Middle Triassic.

The Diptera are still not recorded during the Permian, but are diverse in the Early–Middle Triassic. In the deposit of the Grès à Voltzia, the order represents a diverse fauna with numerous families [38] and [39], phylogenetically distant (several ‘Nematoceran’ groups and Brachycera), even if these fossils are the first and oldest known representatives of this order. These discoveries demonstrate the great antiquity of the order Diptera, much older than the Early–Middle Triassic. Based on our current knowledge, it diversified during the gap of 15 Myr, as the Coleoptera.