Preservation of details
The relatively common preservation of muscles in specimens of Clausocaris lithographica (Figure 10D) is remarkable. The long muscles (Figures 7A, 8A-B, 9A, D), and their fan-shaped attachment (Figure 7D), are unusual. Even more so are the tightly arranged muscles in the trunk, which also preserve a distinct pattern (Figure 8C). These structures were observed by Polz [33], who interpreted them as parts of the trunk appendages. The distinction between trunk and appendages is evidenced by the setae on the latter (Figure 9C, F, G). The appendages are represented mainly by muscle and setae indicating that their cuticle was rather weakly sclerotised and lost through decay. Muscles are also preserved in each trunk segment of Thylacares brandonensis (Figures 2E, 4C) and help to define their boundaries and identify the segment count.
Assignment of Thylacares brandonensisto Thylacocephala
Thylacares brandonensis shows many characters typical of thylacocephalans, but also some features that are unusual for the group (Figure 10A). The large bivalved shield lacks the characteristic well developed optical notch (Figure 3A, B), a feature that was used to argue that some Cambrian taxa should be assigned to Thylacocephala [4]. Yet, other thylacocephalans appear to lack a pronounced optical notch [6]. This observation emphasizes that the shield morphology of bivalved arthropods is not a reliable guide to their affinity (e.g. [35]).
The eyes of T. brandonensis are also unusual in their small size and apparent stalked nature (Figure 6A-D). The large compound eyes of other species of Thylacocephala are generally regarded as sessile [11]. Yet the eyes of many other thylacocephalans are unknown and may also have been small and stalked.
T. brandonensis bears three pairs of sub-chelate appendages (Figure 2A) that resemble those of other thylacocephalans in general structure and size, i.e. the first one is the smallest, the third the largest. Yet they are significantly shorter than those described in most other species (compare Figure 10B and Figure 10E).
The trunk of T. brandonensis is very similar to that of other representatives of Thylacocephala, although it is comprised of more segments.
The differences between this Silurian form and other thylacocephalans can be interpreted as plesiomorphic. A marked optical notch in some later thylacocephalans may be a derived feature, yet shield structures are highly likely to be convergent, and the plesiomorphic condition is difficult to infer. From a functional point of view the evolution of a notch is likely coupled to the reduction of the eye stalks and enlargement of the eyes in some forms. The long raptorial appendages of some later thylacocephalans, which contrast with the short limbs in the Silurian form, also appear to be a derived character, as other younger forms retain relatively short raptorial appendages [6].
The absence of gills
The presence of eight gills has been advocated as an important character of Thylacocephala (e.g. [12],[17]). While it is difficult to judge whether a structure in a fossil is a gill [36], the exquisite preservation in some of the fossils from La Voulte allows virtually no other interpretation [12],[17].
We have observed no features in T. brandonensis that resemble gills. T. brandonensis may represent a sister species to all other thylacocephalans, in which case the absence of gills might be plesiomorphic; alternatively gills may have been lost through decay.
The features interpreted as gills in M. bucculata[34] may simply represent the upper preserved part of the anterior trunk segments; they are in a similar position to these segments in T. brandonensis. Eight poorly preserved gills have also been reported in C. lithographica[33], but they are not evident in the specimens we investigated. Thus, while some thylacocephalans appear to have eight sets of (supposed) gills, it is not clear whether this is a diagnostic character of the group.
Assignment to Eucrustacea
Most authors have considered Thylacocephala as an ingroup of Eucrustacea, yet unequivocal evidence for this assignment has been lacking. Lange et al. [37] noted that the presence of two pairs of antennae in Thylacocephalus cymolopos from the Upper Cretaceous of Lebanon supports the assignment of thylacocephalans to crustaceans. There is also evidence of two pairs of antennae in Thylocaris brandonensis. Although the second appendage is called an ‘antenna’ in Eucrustacea, it is antenniform only in Eumalacostraca. In other eucrustaceans it is used mainly for locomotion, and sometimes resembles the mandible but never the antennula (see discussion in [38]). Thus the morphology of the second antenna varies in different eucrustaceans and this character must be used with caution in determining affinity.
The morphology of the appendages of Thylacocephala appears to be highly derived, which makes comparison with other arthropods difficult. One new observation reported here supports a eucrustacean affinity. The proximal region of the raptorial appendages of Clausocaris lithographica bears up to five enditic projections with rows of setae. Median enditic armature is widespread among Euarthropoda. Slight elevations in early chelicerates bear just a single strong spine accompanied by two smaller spines. Similar arrangements are found in early crustaceans. In labrophoran crustaceans the endites are more strongly pronounced and bear more complex armature (see e.g. [38]). Several strong endites with setose armature are developed in entomostracan eucrustaceans and in at least some appendages of malacostracan eucrustaceans. Hence the presence of up to five pronounced endites with numerous setae on the raptorial appendages of Thylacocephala supports a eucrustacean affinity.
Systematic position: earlier ideas
Since their recognition as a separate group [16],[30] thylacocephalans have been assigned to a range of eucrustacean groups including stomatopods, decapods and cirripedes, and a superficial resemblance to Remipedia has also been noted [3],[39].
The comparison with stomatopods (see discussion in [3]and [40] was based on the shape of the shield of some Cretaceous species, which in lateral aspect resembles that of certain stomatopod larvae. The appendage morphology, however, is incompatible with stomatopods. Although the raptorial appendages of stomatopods are described as sub-chelate they differ strongly from those of thylacocephalans in overall structure. Most important of these differences is the double flexure that results in a Z-shape in stomatopods, whereas the raptorial appendages of thylacocephalans are only “folded” once, and do not close fully. The distal movable finger in stomatopods is formed by a single element, while it comprises four or five in thylacocephalans. There are five pairs of sub-chelate appendages in stomatopods, the second of the series being the largest (at least in extant forms), while in Thylacocephala the last of three pairs is the largest.
The arrangement of tagmata in thylacocephalans, and especially in Thylacares brandonensis, argues against a malacostracan affinity, including stomatopods. The trunk of up to 22 undifferentiated segments strongly differs from that of Malacostraca, which is consistently differentiated into a thorax of eight segments and a pleon of six (sometimes five in Eumalacostraca) or seven (Phyllocarida). The arrangement of tagmata in Thylacocephala also rules out a decapod affinity, a suggestion prompted by the similarity of the gills in certain thylacocephalan species to those in decapods [17]; the nature of the gills in fossil arthropods is difficult to infer without evidence of the ultrastructure of the surface epithelium [36].
The long trunk does not support a close affinity between Thylacocephala and Cirripedia (as pointed out by [41],[42]); the trunk of cirripedes includes only six segments. Like cirripedes, thylacocephalans possess pits on their shield [42], which may represent a dorsal organ. While the special arrangement of these pits in the so-called lattice organ has been argued to be an autapomorphy of Euthecostraca (which also includes cirripedes: [43]) such pits are widespread among crustaceans and even other euarthropods [44].
Systematic affinities: sistergroup to Remipedia?
The new details reported here prompt a reconsideration of a possible affinity of thylacocephalans to remipedes supporting suggestions by Schram [39]). Remipedes possess three pairs of sub-chelate appendages (Figure 10C): the posterior two head appendages (maxillula and maxilla) and the first trunk appendage (maxilliped).
Three pairs of limbs are present anterior to the raptorial appendages in Clausocaris lithographica, namely antennula, antenna and mandibles [19]. Our observations reveal at least two pairs of anterior appendages in Mayrocaris bucculata (most likely representing antennula and antenna, similar to strucures observed in Thylacocephalus cymolopos[37]) and possibly three in Thylacares brandonensis (antennula, antenna and mandibles). Thylacocephala in general appear to bear at least three pairs of appendages anterior to the first raptorial one.
The segmental affiliation of the three raptorial appendages in Thylacocephala is uncertain. They have been interpreted as belonging to the anterior trunk segments (see summary in [19]), but their position in the specimens described here makes this unlikely. The raptorial appendages are broad proximally and apparently too robust to attach to the short trunk segments, except perhaps the most anterior ones. Both their position and size indicates that at least some of the raptorial appendages belong to the posterior divisions of the head.
Thus the three pairs of raptorial appendages in Thylacocephala could represent maxillula, maxilla and trunk limb one (maxilliped), or maxilla and trunk limbs one and two. Additional material is required to resolve this question but a positional homology (homotopy) between the raptorial appendages of Thylacocephala and Remipedia is at least plausible.
Morphological similarities between the raptorial appendages in the two groups strengthen this assumption. The proximal part of the appendages (probably a basipod) bears setose endites in both. More importantly, three or more distal elements form the functional finger of the subchela in thylacocephalans as well as in remipedes (Figure 10B, C, E). This is an unusual character state, as the functional finger of other subchelae in crustaceans comprises only the most distal element (e.g. in mantis shrimps and gammarids), or the distalmost two (e.g. in slipper lobsters).
The multisegmented and relatively undifferentiated trunk in Thylacares brandonensis, which bears more than twenty appendages, is unusual among eucrustaceans. Apart from the modern branchiopod ingroups Polyartemia and Phyllopoda, which, unlike thylacocephalans, possess phyllopodous appendages (i.e. limbs differentiated into a basipod with median endites, a reduced endopod, a paddle-shaped exopod and lateral epipods; e.g. [45]), only Remipedia show such a high number of trunk segments. The eucrustacean trunk is usually differentiated into at least two tagmata: thorax and pleon in malacostracans and thorax and abdomen in entomostracans [46], in contrast to the trunk in thylacocephalans and Remipedia, where there is only one tagma. The specialisation of the posterior head appendages and anteriormost trunk appendage as sub-chelate raptorial appendages with setose endites and a finger made up of the three or more distal elements represents a potential synapomorphy of Thylacocephala and Remipedia. This, together with the multisegmented trunk, suggests a sister-group relationship.
Additional material is necessary to determine whether the three raptorial appendages in Thylacocephala are homologous in position with maxillula, maxilla and maxilliped and to test the possibility of a sister group relationship with Remipedia. Such evidence is a prerequisite for a rigorous phylogenetic analysis.
Functional morphology and 3D modelling of Thylacocephala
Previous authors (e.g., [4],[20]) have considered the thylacocephalans to be nectic or necto-benthic predators. The new evidence presented here allows us to refine these interpretations.
Thylacocephalans are usually reconstructed with their raptorial appendages pointing forward, i.e., more or less in the axial plane of the body. However, our new reconstruction (Figure 10A, D) shows that such an arrangement is unlikely. Given the narrow ventral gape, which has been reconstructed in different species [4], and the relatively large size of the appendages, only certain positions are possible. The proximal podomeres with their endites and armature must have interacted with the opposing appendage of the pair and therefore cannot face directly forward, but at most antero-medially. When the valves are closed, the largest (third) appendages would occupy almost the entire width of the ventral gape; there is no space for the two other pairs of raptorial appendages to lie inside or outside the valves. This problem is overcome by rotating the proximal podomeres about 45° abaxially to accommodate all three appendages within the narrow ventral gape in an anterior-posterior sequence.
In remipedes the appendages are also held in an abaxial orientation, at an even higher angle than in thylacocephalans. Mantis shrimps, although not closely related, provide a further functional comparison. They accommodate four pairs of sub-chelate appendages which are also rotated away from the axis. The arrangement of the raptorial appendages in mantis shrimps, near parallel to the axial plane, is achieved by the presence of an additional joint, resulting in a z-shape.
The attitude of thylacocephalan raptorial appendages reconstructed here is supported by several specimens that preserve them directed posteriorly (Figures 7E, F, 8A). Preservation of appendages projecting forwards in some specimens and backwards in others is less likely if the appendages are held in an orientation parallel to the axial plane of the body.
The appendages are reconstructed here forming a kind of basket. Such an arrangement would have facilitated predation. In the absence of a second, strongly flexed joint, such as that in mantis shrimps and some great-appendage arthropods [47], the raptorial appendages of thylacocephalans could not extend forward. More likely, the crustacean swam forward using the trunk appendages and trapped prey or at least grabbed food items with the raptorial appendages.
The paddle-shaped trunk appendages identified here in Clausocaris lithographica are compatible with a swimming function. They are equipped with setae enlarging the surface, and powered by well-developed muscles. The last pair of trunk appendages, which are more elongate than those lying anterior to them, may have functioned as a steering device, in a manner similar to the uropods of eumalacostracans and the furca of phyllocarid and entomostracan eucrustaceans.
