Phylogeny

Analysis of the paleontological data resulted in a remarkably resolved topology for Isoptera (fig. 3). Structure of the consensus tree generally agrees with that from prior morphological and molecular studies (Kambhampati et al., 1996; Donovan et al., 2000; Thompson et al., 2000; Bitsch and Noirot, 2002; Inward et al., 2007a, 2007b), but depart in some respects from the recent molecular study of Legendre et al. (2008) (e.g., the relatively basal placement of Kalotermitidae). Legendre et al. (2008) had a dramatically reduced taxon sampling relative to other, more comprehensive treatments (e.g., Inward et al., 2007a).

In our analysis unequivocal relationships include Cryptocercus as the living sister group to the termites and Mastotermes Froggatt as the living sister group to all other termites—the lineage we are designating here as Euisoptera. Mastotermes darwiniensis Froggatt, from northern Australia and southern New Guinea, is the sole survivor of the formerly global Mastotermitidae (Thorne et al., 2000; Wappler and Engel, 2006), and it retains striking plesiomorphic features with roaches, such as laying its eggs in a vestigial pod or ootheca. Interestingly, the Mastotermitidae as it has historically been classified was recovered as monophyletic, despite opinion to the contrary (e.g., Jarzembowski, 1981). Cratomastotermes Bechly, from the Early Cretaceous Crato Formation of Brazil, was formerly placed in the Mastotermitidae (Bechly, 2007); in our analysis it is the most basal species of termite since it retains even more plesiomorphic features than Mastotermitidae (Grimaldi et al., 2008). This genus is accordingly placed in a new, extinct family, Cratomastotermitidae, new family (see Classification). Cretaceous fossils traditionally classified as Hodotermitidae, such as the extinct “genus” Meiatermes Lacasa-Ruiz and Martínez-Delclòs, actually comprise a grade of stem-group species that fall basal to the divergence of Termopsidae, true Hodotermitidae, and all other termites. Termopsidae stat. n. (see Classification) should be restricted to the Baltic amber species of Termposis Heer (Engel et al., 2007b). What we are designating as the true Hodotermitidae (see Classification) is a group of three genera of the “harvesters”—Anacanthotermes Jacobson, Hodotermes Hagen, Microhodotermes Sjöstedt—that feed on grasses in savanna and steppe biomes of Africa and Eurasia; monophyly of this group is confirmed by other analyses (Inward et al., 2007a, 2007b; Legendre et al., 2008). Three genera of highly disjunct wood feeders that are traditionally placed in the Termopsidae s.l.—Archotermopsis Desneux, Zootermopsis Emerson, and Hodotermopsis Holmgren—were not found to be monophyletic, contrary to other studies. Inclusion or exclusion of the fossils made no difference to the definition of “Termopsidae s.l.” except that Archotermopsis was sometimes found to be monophyletic. As such, the traditional concept of Termopsidae s.l. could not be supported and these genera are accordingly classified in the new family Archotermopsidae, new family (see Classification).

The austral disjuncts, Stolotermes Hagen and Porotermes Hagen, are sometimes placed within Termopsidae s.l. (e.g., Donovan et al., 2000; Thorne et al., 2000; Legendre et al., 2008) but should clearly be the separate family Stolotermitidae, stat. n. ( =  Stolotermitinae + Porotermitinae: see Classification), since they are the living sister group to Kalotermitidae plus higher termites in this study and in some molecular analyses (Inward et al., 2007a, 2007b). The divergence between Stolotermitidae and Kalotermitidae + higher termites was certainly in the Early Cretaceous, since there are three intermediate genera from the Early and mid-Cretaceous. The family Kalotermitidae, or dry-wood termites, is a cosmopolitan group of 457 living species, which has two known Cretaceous and three Tertiary stem-group taxa (only two of the latter were included in our study), with living species comprising a monophyletic group of probably Tertiary origin. Interestingly, a kalotermitid-like nest exists from the Late Cretaceous Javelina Formation of Texas (Rohr et al., 1986). Kalotermitidae is the sister group to an unequivocally monophyletic group we are calling the Neoisoptera, which is defined in part by the distinctive opening of the frontal gland called the fontanelle. The Neoisoptera is comprised of the Rhinotermitidae (13 living genera, 380 living species, all of which are wood feeders), the monotypic family Serritermitidae from Brazil, and the largest family, Termitidae. It also includes Archeorhinotermes rossi Krishna and Grimaldi in 100 Ma Burmese amber, the most derived termite from the Cretaceous, formerly placed in the Rhinotermitidae (Krishna and Grimaldi, 2003), but actually an extinct stem group to the rest of the Neoisoptera and here considered as the sole member of Archeorhinotermitidae, stat. n. (see Classification). Parastylotermes Snyder and Emerson and Stylotermes Holmgren and Holmgren (neither genus studied in prior analyses [Donovan et al., 2000; Inward et al., 2007a, 2007b; Legendre et al., 2008]) should be separated from Rhinotermitidae s. str., the latter doubtfully monophyletic (Donovan et al., 2000; Inward et al., 2007a, 2007b; Legendre et al., 2008) but clearly closely related to the Termitidae. These genera have been at times considered a separate family, as Stylotermitidae, stat. rev. (see Classification), and this status should be reinstated. The earliest rhinotermitids are Reticulitermes antiquus (Germar) and Heterotermes eocenicus Engel in Baltic amber (Engel et al., 2007b; Engel, 2008); the divergence of both families probably occurred in the Early Tertiary.

Classification

Several taxonomic changes are required in order to have the classification of termite families reflect our cladistic results. The revised, higher-level classification of Isoptera is outlined in table 3 (modified from Engel and Krishna, 2004a, 2004b, 2007c). The classification is that which is employed for (and will be further elaborated in) the forthcoming world catalog of Isoptera (Krishna et al., in press).

Table 3

Synonymical Hierarchical Classification of Isoptera (modified and updated from Engel and Krishna, 2004a, 2004b, 2007c)

Cratomastotermitidae, new family

Ecology and Evolution

The circumtropical family Termitidae, or “higher termites,” comprises approximately 70% of all termite species and appears to be one of the most recent radiations of all insect groups that are ecologically significant. Monophyly of the Termitidae is well established; the family includes such familiar groups as the Macrotermitinae and Nasutitermitinae, some of which build huge mounds in grassland and scrub biomes; other nasute taxa build large arboreal nests of cartonlike, fecal material in tropical forests. The diets of Termitidae are extremely diverse, primitively being sound and rotting wood but also including humus, leaf litter, soil, grass, herbivore dung, and even the mycelia of a symbiotic fungus, Termitomyces R. Heim, that they cultivate in the nest like attine ants (e.g., Sands, 1969). The huge colonies and diverse diets of the Termitidae account for the overwhelming biomass of termites in tropical and subtropical environments. The earliest apparent termitid is an incomplete compression of an imago from the Oligocene of Brazil, ca. 30 Ma (Martins-Neto and Pesenti, 2006;, a putative termitid from the Bembridge Marls [Jarzembowski, 1980] is probably a rhinotermitid). The first diverse paleofaunas of Termitidae—more than 30 species—are in Miocene amber from the Dominican Republic (Krishna and Grimaldi, 2009) and Mexico, which are very similar to modern Neotropical faunas. This dramatic appearance is probably due to the fact that these ambers are the only major fossil insect deposits from the Neogene that were formed in the tropics. Were Termitidae abundant during the Eocene they should have been preserved in Baltic amber (Lutetian: ca. 45 Ma), since this deposit has yielded thus far most other living termite families (Engel et al., 2007b; Engel, 2008), as well as species belonging to a diversity of other tropical insect groups (Grimaldi and Engel, 2005). We estimate that Termitidae diverged from Rhinotermitidae sometime in the Early Paleogene (perhaps Late Paleocene or Early Eocene) and subsequently began its diversification in the latest Eocene (Priabonian-Bartonian: ca. 40 Ma) to Early Oligocene, continuing to radiate throughout the remainder of the Neogene and Quaternary.

Though Jurassic remains of Isoptera have not been found, Isoptera appear to have diverged from cryptocercid roaches in the Late Jurassic. This would make termites the oldest group of eusocial animals, predating the origins of ants by some 35 million years. Major geological and biotic events in the Cretaceous probably had little effect on termites, since basal divergences appear to have preceded the drift of Gondwanan continents and the angiosperm radiations. Unfortunately, the stratigraphic sampling is too poor in the Late Cretaceous and Paleocene to determine any effects of the end-Cretaceous extinctions. The Tertiary thermal maximum of the late Paleocene and Eocene, however, probably had a profound effect on termites, specifically on the global spread of Mastotermes and the radiations of living Kalotermitidae and Neoisoptera. The rapid spread of C4 grasslands in the Miocene (Jacobs et al., 1999) doubtless promoted a minor diversification of the harvesters and the explosive diversification of many Termitidae, such as the Macrotermitinae. A macrotermitine nest, in fact, is known from the Miocene of Chad (Duringer et al., 2006).

Throughout the Cretaceous and early Tertiary, including the mid-Eocene, termites represented less than 1% of all insect specimens in all fossil deposits (fig. 3). Their abundance rises in the late Eocene to approximately 2%, and then spikes from 5%–10% during the Miocene as both amber and compression fossils, to the present day (in copals, or subfossil resins) (table 2, fig. 3). This spike in abundance is due to the diversification of the Termitidae. The abundance of ants rises dramatically in the Eocene (Grimaldi and Agosti, 2001; Dlussky and Rasnitsyn, 2002), and ants are generally much more abundant in Tertiary insect deposits than are termites, probably because termites feed within the wood where they nest or they travel through tunnels from nest to food sources, so foragers are rarely exposed and imagoes are exposed only during brief nuptial flights. Interestingly, many termitids will forage in the open and these are concomitantly the most abundant termites in Dominican amber and copal.

Our analysis indicates the importance of including fossils in cladistic analyses rather than mapping their putative ages onto molecular-based trees for purposes of dating. The identity of stem groups is obscured in the latter method, resulting in overestimates for divergence times. In our analyses this overestimation is highlighted by the traditional taxonomic placement of Termopsis and most Cretaceous fossils as Termopsidae s.l. and Hodotermitidae, respectively (fig. 3). These taxa in fact represent either a grade to more nested termite lineages or stem groups to some modern families (fig. 3). Using such fossils to calibrate the basal nodes for Hodotermitidae or Termopsidae would result in significant overestimates of the ages of these groups. A failure to distinguish stem groups in an analysis probably explains prior overestimates of the age of other insect lineages (e.g., Moreau et al., 2006; Hunt et al., 2007).

Patterns in termite diversification are very similar to those of the ants (Grimaldi and Agosti, 2001): throughout the Cretaceous both groups were rare and consisted of basal lineages. The diversity and abundance of termites and ants spiked in the Tertiary when speciose groups that form large colonies with highly specialized castes (for ants, the subfamilies Dolichoderinae, Formicinae, and Myrmecinae) eclipsed the smaller colonies of more basal taxa—the concept of “dynastic succession” (Wilson and Hölldobler, 2005). For ants, approximately 70 million years passed from origin to ecological dominance; in termites, this period was 100 million years. Thus, eusociality per se does not result in ecological success, but living in very large colonies with extreme division of labor does. Why, then, did it take so long for large colonies to evolve? We suggest that social evolution is like any other highly adaptive feature, such as the evolution of flight in feathered theropods, and thus may take tens of millions of years to refine.

Similarly, the symbiosis of termites with intestinal protozoa or bacteria which aid their break down of lignocellulose does not alone explain their ecological success as basal termite lineages exhibit the same mutualistic relationship. While their critical role as carbon recyclers is made possible by this symbiosis, this association existed for tens of millions of years before their rise in abundance and diversity (fig. 3).

One question remains: how was wood decomposed in Mesozoic forests with few or no termites? Patterns of coal and oil deposition suggest that lignocellulose did not rapidly decompose prior to termites and the actions of fungi or other organisms at the time must have been either slow or negligible. Coal is formed from ombrotrophic, or waterlogged, peat (Scott, 1987). While there are some tropical peatlands, such as the coastal “moor” forests of western Borneo and southern Sumatra, these are dwarfed in area compared to the boreal peatlands of sphagnum and heaths that comprise some 3% of earth's land surface. Tropical ecosystems produce more biomass, but much greater plant detritus accumulates in boreal forests and peatlands, which is traditionally explained by lower boreal evapotranspiration and because seasonality limits decomposition (Scott, 1987). Indeed, boreal peat lands lay hundreds of miles north of the most northern termites in the genus Reticulitermes Holmgren (Rhinotermitidae). Maximum termite diversity is equatorial, and half of that diversity falls between 18° N and 30° S latitudes; by 48° N and S it is 1%–4% that of the diversity at the Equator (Eggleton, 2000). Tropical ecosystems where termites are most abundant and diverse have notoriously thin humus layers (Richards, 1996). This may explain why coals that were formed prior to the appearance of termites in the Paleozoic and Early Mesozoic, and in largely the same regions and habitats, decomposed less (i.e., contained significantly more vitrain) than Tertiary and modern peats (Shearer et al., 1995; Raymond et al., 2000), as well as the formation of some vast reservoirs of petroleum, like those in the Early Cretaceous Nubian sandstones of present-day Africa and the Middle East. While some Miocene coal formations are astonishingly thick (Shearer et al., 1995), these were formed in palaeoclimates that today would have very few or no termites.