TRIGONOTARBIDA, PETRUNKEVITCH, 1949

Jones, Fiona M., Dunlop, Jason A., Friedman, Matt & Garwood, Russell J., 2014, Trigonotarbus johnsoni Pocock, 1911, revealed by X-ray computed tomography, with a cladistic analysis of the extinct trigonotarbid arachnids, Zoological Journal of the Linnean Society 172 (1), pp. 49-70 : 60-61

RELATIONSHIPS WITHIN TRIGONOTARBIDA

Trigonotarbid monophyly

We found no evidence that Trigonotarbida is paraphyletic (contra Petrunkevitch, 1949), although we did not include the extinct order Haptopoda in our analysis. Petrunkevitch grouped this order with the trigonotarbid family Anthracomartidae as a novel subclass: Stethostomata. However, numerous arguments for why Stethostomata is an artificial assemblage have been presented previously (e.g. Dunlop, 1996b) and the analysis of Shultz (2007) also recovered Haptopoda close to ( Amblypygi (Thelyphonida + Schizomida)) but not with Trigonotarbida . Again, a fuller analysis of living and fossil arachnids should resolve this issue more clearly, but we see nothing in our data set that would fundamentally challenge trigonotarbid monophyly. Characters like the apparent absence of a first sternite of the opisthosoma emerge as potentially diagnostic for Trigonotarbida .

Status of the families

Within Trigonotarbida , Archaeomartidae and Lissomartidae only contain a single genus, and in both cases only one species was included in the analysis. The present data support the monophyly of three further families: Anthracomartidae , Anthracosironidae , and Eophrynidae . The anthracomartids are principally united by their division of the opisthosomal tergites into five plates. The anthracosironids share a rounded prosoma and a long and fairly narrow opisthosoma. The eophrynids are united by a reduced locking ridge, absence of a diplotergite 2 + 3 (both characters are also seen in Tr. johnsoni ), the presence of four tubercles with a cross-shaped arrangement immediately behind the median eyes, and limb podomeres possessing longitudinal ornamentation.

This leaves Palaeocharinidae , Trigonotarbidae , Aphantomartidae , and Kreischeriidae without strong cladistic support based on the taxa and characters considered here. To approach these in order of decreasing support, the heavily tuberculate Aphantomartidae and Kreischeriidae are shown in the agreement subtrees ( Fig. 5 ) to be paraphyletic grades within the eophrynid assemblage (see below). The only reliably placed palaeocharinid is the Rhynie Chert genus Palaeocharinus . In analyses in which a position is recovered for Gilboarachne (see Supporting Information File S4), it is sister to a clade including Palaeocharinus , the anthracomartids, and the archaeomartid, suggesting that these members of the family Palaeocharinidae could be a grade recognized on the basis of plesiomorphic characters. The putative palaeocharinid Gigantocharinus is one of the least stable taxa because of a combination of characters that suggest parallels to both the eophrynid assemblage and Anthracomartidae (see discussion in Shear, 2000). Its current placement in Palaeocharinidae thus looks unlikely.

Despite the increased morphological information provided in the current study, Trigonotarbidae remain very difficult to place definitively. In some IW analyses (k = 0.25, 1; see Supporting Information File S4 for trees), the two included Trigonotarbus species are recovered as sister taxa, whereas in other analyses they are variously a grade at the base of the nonanthracomartid/ Palaeocharinus clade (k = 3) or widely separated (k = 10). This could be partly because of misinterpretation of the Devonian species, Trigonotarbus stoermeri , for which Schultka (1991) confused dorsal and ventral features. A redescription of this species with a test of its generic affinities would be welcome. It is interesting to note that in several trees the Trigonotarbus species group with Palaeotarbus , the earliest known trigonotarbid, which appears superficially similar in appearance to the Trigonotarbidae .

A framework phylogeny and evolutionary trends

Our data set was unable to produce a well-resolved tree for Trigonotarbida as a whole; thus, we are reluctant at this stage to propose formal changes to higher systematics and nomenclature. Additionally, the incertae sedis genera did not clearly resolve with any of the established families and remain difficult to place. Yet we hope to have made a starting point towards understanding trigonotarbid evolution and we can offer a series of characters into which future discoveries (or revisions of existing taxa) can be scored in the hope of refining the overall picture. The lack of resolution is probably not because of missing data, rather we suspect that uncertainties and errors in interpretation of the fossils may create conflicting signals that may be contributing to unreliable placements.

Despite several problematic, highly mobile, taxa removing the signal in the strict consensus trees, a consistent result of our analysis is a split at the base of the trigonotarbids. One reliable ‘anthracomartid’ clade ( Fig. 5 ) comprises ( Palaeocharinus ( Archaeomartidae + Anthracomartidae )), which is principally defined by the presence of a somewhat box-shaped carapace. Our data thus offer cladistic support to the suggestion by Poschmann & Dunlop (2010) that the Palaeocharinus species from the Rhynie and Windyfield Cherts are closely related to Anthracomartidae , with Archeaomartidae as a putative intermediate taxon.

The other major clade recovered contains all remaining trigonotarbids. Amongst this assemblage, Lissomartus invariably resolves close to the eophrynid assemblage, as implied by the beginnings of a lateral division of the carapace into lobes ( Fig. 1 ). The most consistently recovered clade in all trees is thus the aforementioned eophrynid assemblage sensu Dunlop & Brauckmann (2006), supported here by a lobed carapace and heavily ornamented dorsal cuticle. In our agreement subtrees ( Fig. 5 ) this clade takes the form of ( Aphantomartus ( Alkenia ( Pseudokreischeria ( Kreischeria ( Eophrynus + Pleophrynus ))))). This implies that Eophrynidae is monophyletic, but that the inclusion of Alkenia in Aphantomartidae and the resurrection of Kreischeriidae as a family may have been premature.

The resolved phylogeny also demonstrates that the split between these two major clades must have occurred by the Early Devonian at the latest. The earliest (Silurian) trigonotarbid Palaeotarbus is consistently recovered within the non-anthracomartid clade (see Supporting Information File S4), whereas Palaeocharinus from the Devonian Rhynie Chert is placed in the ‘anthracomartid’ clade. This implies either an early split as part of rapid diversification after terrestrialization within the trigonotarbids, or it could reflect early terrestrialization and then a long period from which we have no trigonotarbids prior to the earliest fossils, a possibility given the patchy fossil record of pre- Carboniferous terrestrial arthropods. Either scenario lends credence to the suggestion that the peak in richness seen in Coal Measures deposits is more likely to be taphonomic, or sampling bias, rather than representing a genuine interval of diversification within the trigonotarbids.

The evolution over time within these clades is also interesting to note: it could reflect increasing ecological stress for Carboniferous trigonotarbids. From the (presumably cursorial) predators of the Rhynie and Windyfield Cherts (Garwood & Dunlop, 2014), the anthracomartid clade developed morphologies such as laterigrade legs consistent with a group adopting ambush predation by the Carboniferous: presumably a less exposed mode of life (Caraco & Gillespie, 1986). Convergent evolution of this trait is apparent within the Anthracosironidae ( Dunlop, 1994 b) , which also have semiraptorial forelimbs ( Fig. 1 ). A strong trend towards the development of heavy, probably defensive, tuberculation and later spination in the eophrynid assemblage is apparent during the Carboniferous ( Dunlop & Garwood, 2014). The changes in both lineages could have resulted from an increasing risk of predation for trigonotarbids by tetrapods, which diversified prolifically during the Carboniferous (Coates, Ruta & Friedman, 2008). Arachnids had, in the Early Devonian, been subject to no such risk.

Figures

Figure 1. Reconstructions of representatives of nine trigonotarbid families, shown to scale. A, Palaeocharinus rhyniensis (Palaeocharinidae). B, Archaeomartus levis (Archaeomartidae). C, Anthracomartus hindi (Anthracomartidae). D, Anthracosiro woodwardi (Anthracosironidae). E, Trigonotarbus johnsoni (Trigonotarbidae). F, Lissomartus schucherti (Lissomartidae). G, Aphantomartus pustulatus (Aphantomartidae). H, Eophrynus prestvicii (Eophrynidae). I, Kreischeria wiedei (Kreischeriidae). Scale bar = 10 mm.

Figure 5. The results of the cladistic analysis. The four trees show, as labelled, the agreement subtree and strict consensus of both the equally weighted analysis and implied weights analyses with concavity constants (k) of 0.25, 1, 3, and 10.