Introduction

With over 113,894 described species, the chelicerates (sea spiders, horseshoe crabs, and arachnids) represent one of the most species-rich clades on Earth today (Zhang 2011), second only to their primary prey, the hexapods (insects). Understanding the origin and early evolution of chelicerates has proven to be a difficult and contentious pursuit. Although molecular clock analyses indicate an early Cambrian origin for this clade (Lee et al. 2013; Rota-Stabelli et al. 2013), body fossils of crown-group euchelicerates are not recovered until the Early Ordovician (Van Roy et al. 2010), although unequivocal chelicerate trace fossils (Dunlop et al. 2004), and a putative larval pycnogonid are known from the late Cambrian (Waloszek and Dunlop 2002). A diversity of putative chelicerate ancestors have been identified from Cambrian Konservat-Lagerstätten. Older works tended to align chelicerates with artiopodans, a group of arthropods including trilobites and other trilobite-like arthropods, collectively known as the trilobitomorphs, and vicissicaudates, usually possessing a broad tergum, antennae, and bilobate exopods (Stein and Selden 2012). Within Artiopoda, chelicerates were typically allied to the vicissicaudates (Wills et al. 1995, 1998; Hou and Bergström 1997; Dunlop and Selden 1997; Bergström and Hou 2003; Edgecombe et al. 2011), a group consisting of cheloniellids, aglaspidids, and xenopods (Fig. 1(a,c,d,f,h); Ortega-Hernández et al. 2013); more recent studies have regarded megacheirans (‘short-great-appendage’ arthropods) as the putative sister-taxon of Chelicerata (Fig. 1(b,e,g)), based primarily on the morphology of their anterior-most appendage pair (Chen et al. 2004; Cotton and Braddy 2004; Scholtz and Edgecombe 2005; Haug et al. 2012).

Fig. 1
figure 1

Previous hypotheses regarding the relationships of fossil taxa, particularly artiopodans and megacheirans, to Chelicerata. Twenty common taxa were selected for ease of comparison. Megacheirans are represented by Jianfengia, Fortiforceps, Yohoia and Leanchoilia; trilobitomorph artiopodans by Retifacies, Kumaia, Cindarella, Naraoia and the trilobite Olenoides; and vicissicaudate artiopodans by Sidneyia, Emeraldella, Aglaspis and Cheloniellon. a Wills et al. (1995, fig. 1A; 1998, fig. 2.1). b Bousfield (1995, figs. 7, 10). Rather than proposing sister-taxon relationships, Bousfield (1995) postulated direct lineages. c Hou and Bergström (1997, figs. 87, 88). d Dunlop and Selden (1997, fig. 17.3). This analysis did not include megacheirans. e Chen et al. (2004, fig. 6B) and Haug et al. (2012, fig. 11). This analysis did not include artiopodans. f Bergström and Hou (2003, fig. 4). g Cotton and Braddy (2004, fig. 8). h Edgecombe et al. (2011, fig. 8)

Full size image

Sanctacaris uncata from the middle Cambrian Burgess Shale was originally identified as a stem-chelicerate with putative chelicerate features including a cephalon bearing six appendage pairs, a cardiac lobe, and a similar pattern of tagmosis (Briggs and Collins 1988). As with other putative chelicerate allies, the affinities of Sanctacaris were rejected due to a lack of chelate frontal appendages (Bousfield 1995), the cephalic appendages were instead interpreted as a ‘limb-basket’ (Dewel and Dewel 1997) and this taxon resolved amongst a paraphyletic assemblage of megacheirans in a subsequent phylogenetic analysis (Budd 2002). A restudy of Sanctacaris was therefore undertaken. The significance of purported chelicerate features was explored, and new observations on Sanctacaris and other purported stem-group chelicerates were coded into an extensive phylogenetic analysis of panarthropods to ascertain their significance and explore the early evolution of Chelicerata.

Material and methods

Specimens and geological setting

All previously described material of S. uncata, reposited at the Royal Ontario Museum (ROM), Toronto, was examined. This material, which was originally described by Briggs and Collins (1988), is currently the only known material of S. uncata. These specimens were recovered from Unit 3 of the Collins Quarry exposure of the Kicking Horse Shale Member (Burgess Shale Formation) (Collins et al. 1983; Briggs and Collins 1988), situated on the western slope of Mount Stephen in Yoho National Park, British Columbia, Canada. Associated trilobites indicate this horizon belongs to the Polypleuraspis insignis Subzone of the Glossopleura Biozone (middle Cambrian, Series 3, Stage 5) (Fletcher and Collins 1998, 2003). This fauna is dominated by Alalcomenaeus, which represents 58 % of recovered specimens (Collins et al. 1983; Briggs and Collins 1999), and the bivalved arthropods Nereocaris briggsi (190 specimens) and Loricicaris spinocaudatus (28 specimens), which combined account for a further 8 % of specimens from this site (Legg and Caron 2014).

Specimens of the xenopods Emeraldella brocki (Bruton and Whittington 1983; Stein and Selden 2012) and Sidneyia inexpectans (Bruton 1981; Stein 2013), reposited at the ROM and the National Museum of Natural History (USNM), Smithsonian Institution, Washington D.C., were also examined for comparative purposes. Stratigraphical information regarding examined material is provided in Bruton and Whittington (1983) and Stein (2013).

Phylogenetic analysis

New observations made herein were coded into a modified version of a phylogenetic data set of panarthropods (Legg et al. 2013; Siveter et al. 2014). To the original data set of 314 taxa and 753 characters, 13 taxa were added, 14 characters were added and 16 were modified, either recoded or combined or divided into discrete characters, and three were removed completely resulting in a total data set of 327 taxa and 763 characters (see online supplementary material for details). Cladistic analysis was performed using TNT v. 1.1. (Goloboff et al. 2008b). All characters were treated as non-additive (unordered) and weighted using both equal and implied weighing with a variety of concavity constants (k = 2, 3, 10), a justification of which is given elsewhere (Goloboff et al. 2008a; Legg et al. 2013; Legg and Caron 2014). Most parsimonious trees (MPTs) were found using New Technology search options with 100 Random Addition Sequences using Ratchet (Nixon 1999), Sectorial Searches, Tree Drifting and Tree Fusing (Goloboff 1999).

Discussion

Morphological interpretation of S. uncata

This section is not intended as a complete redescription of S. uncata, but rather a discussion of key characteristics either in need of reinterpretation or deemed important for determining relationships of this taxon, particularly those pertaining to either artiopodan and/or chelicerate affinities.

The morphology of the anteriormost appendages has been a cynosure of arthropod systematics (Scholtz and Edgecombe 2005, 2006). The chelicerate affinities of Sanctacaris were rejected due to a lack of chelicerae (Bousfield 1995), a pivotal synapomorphy of Chelicerata (Lamsdell 2013). Based on comparisons with Offacolus kingi (Orr et al. 2000; Sutton et al. 2002), from the Silurian of Herefordshire, it was considered possible that the chelicerae may be present in Sanctacaris yet unobservable (Boxshall 2004). Still others have interpreted structures in the holotype (Figs. 2a,b and 3) originally identified as ‘antenna-like rami’ of the sixth cephalic appendage (Briggs and Collins 1988) as biramous deutocerebral antennae (Budd 2002). Frontal appendages, ‘great-appendages’, chelicerae, antennae or other uniramous anterior appendages could not be identified in any of the available material. Instead, the structures interpreted as part of a biramous antenna are herein interpreted as individual exites which belong to the posterior cephalic appendages (Figs. 2a,b and 3). Each ramus is distinguished from the other by a slight variation in relief with the posterior rami preserved above the anterior ones. This interpretation is further supported by their resemblance to exites preserved in other specimens (Fig. 2c). The distal tips of the posteriormost exites bear a cluster of setae (Figs. 2b and 3), reminiscent of those in the basal chelicerates Offacolus (Sutton et al. 2002) and Dibasterium durgae (Briggs et al. 2012).

Fig. 2
figure 2

Sanctacaris uncata and Sidneyia inexpectans from the Burgess Shale: a, b Royal Ontario Museum (ROM) 43502b (holotype of Sanctacaris uncata), cephalic region. a Cephalic region, b enlargement of right cephalic appendages. c Cephalic appendages of ROM 43504b (S. uncata). d Cephalic appendages of ROM 920274a (Sidneyia inexpectans). Entire specimen figured in supplementary figure 1. e Posterior trunk, abdomen and telson of ROM 43506 (S. uncata). f Medial trunk region of ROM 43504b (S. uncata); arrows indicate bifurcation of putative axial lobes. g Medial trunk region showing proximal lamellal setae of ROM 43503 (S. uncata). h Distal exites of ROM 43504a (S. uncata). ab abdomen, ca II–VI cephalic appendages II–VI, ce compound eye, dex distal exite flap, en endopod, ex exite, ls lamellal setae, sb setal brush, st setae and tl telson

Full size image
Fig. 3
figure 3

Camera lucida drawing of the appendages of the holotype of Sanctacaris uncata (ROM) 43502b. Accompanying photo in Fig. 1(b). a Camera lucida drawing. b Proposed homology of limb elements showing a gradual increase from five elements in the anterior appendages to seven in the posteriormost appendage. Endopods are in brown, basipodites in purple, exites in blue, and the compound eye in green. ce compound eye, cs cephalic shield, en clusters of endites, sb setal brush, and tp terminal endopod podomere

Full size image

The putative limb-basket (sensu Bousfield 1995) has been interpreted as a pair of multi-ramous appendages, comparable to the short-great-appendages of megacheirans (Dewel and Dewel 1997; Budd 2002). An anterior clustering of the cephalic appendages has also been observed in other Palaeozoic arthropods, particularly the xenopods Sidneyia (Fig. 2d) and Emeraldella (Stein and Selden 2012). The reason for this is unknown but may be taphonomic, e.g. the appendages are pushed forward during burial, or due to moulting. Specimens preserved in this manner also tend not to have their antennae preserved (Fig. 2d), indicating that the lack of anterior appendages in Sanctacaris may not be a genuine absence. Each cephalic appendage of Sanctacaris possesses a distinct basipodite (Figs. 2a,b and 3) and is therefore not fused into a single multi-ramous appendage. Each appendage is associated with a pediform exite, but the nature of its attachment is unclear. The separation of these exites from the endopod in some specimens (Figs. 2a,b and 3) may indicate the exites were attached to the underside of the cephalon, as seen in Dibasterium (Briggs et al. 2012). The posterior cephalic endopod pairs of Sanctacaris resemble those of Sidneyia in possessing seven podomeres, the most distal of which bears an elongate endite on the medial-distal margins, and a subchelate terminal podomere (Fig. 2a,b). The anteriormost endopod of Sanctacaris possesses just five podomeres (Fig. 2a,b). A reduced anterior (first post-antenna/chelicera) endopod is also present in Sidneyia (Stein 2013), Emeraldella (Stein and Selden 2012) and chelicerates.

All appendages preserved in Sanctacaris are biramous; however, deutocerebral appendages in arthropods are universally uniramous (Boxshall 2013), thus indicating that either Sanctacaris has lost its anteriormost (deutocerebral) appendages through evolution or they were not preserved. Assuming, at least a single pair of unobserved anterior appendages then the cephalon of Sanctacaris bears six pairs of appendages, a characteristic feature of Euchelicerata (horseshoe crabs and arachnids) (Lamsdell 2013). Although Sanctacaris was originally described as possessing a cardiac lobe (Briggs and Collins 1988), another characteristic feature of Euchelicerata, this could not be identified in the available material. The cardiac lobe of unequivocal chelicerates is a sub-triangular bulge at the posterior of the cephalon. The equivalent area in Sanctacaris is sub-circular and much broader, more comparable to the interophthalmic area of extant horseshoe crabs, indicating the cephalon of Sanctacaris was also vaulted, but the eyes are located underneath the anterolateral margin of the cephalic shield.

The trunk of Sanctacaris consists of 11 segments, the posteriormost of which is limbless and has reduced lateral pleurae (Fig. 2e). An 11-segmented trunk is also present in aglaspidids (Hesselbo 1992) and the basal chelicerates Dibasterium (Briggs et al. 2012), Legrandella (Eldredge 1974) and possibly Weinbergina (Lamsdell 2013). Unlike Sanctacaris, however, the abdomens of these taxa all consist of three segments. The abdomens of cheloniellids and Emeraldella consist of a single segment (Stein and Selden 2012; Ortega-Hernández et al. 2013).

Although the axial region of the holotype of Sanctacaris demonstrates notable relief, it lacks distinct trilobation. Longitudinal ridges on the axial region may represent subaxial nodes, a feature common amongst basal chelicerates (Lamsdell 2013); however, in Sanctacaris, they are unevenly distributed and occasionally intersect transverse ridges. They could therefore be caused by post-mortem deformation rather than being a genuine morphological feature (Fig. 2f).

Trunk appendages occur on all but the posteriormost trunk segment. They are biramous, composed of a short endopod (the exact podomere count could not be determined) and an exite consisting of at least two parts, a proximal shaft bearing lamellar setae with accessory setules (Fig. 2g) and a distal lobe fringed with fine setae (Fig. 2h). The basipodite was not observed but is assumed to be present because the appendages are typically biramous and because basipodites are observed in similar appendages in the cephalon. A bi- or tripartite exite with a proximal shaft and distal lobe is characteristic of artiopods (Stein and Selden 2012; Stein 2013; Ortega-Hernández et al. 2013). A similar arrangement is also present in the opisthosomal appendages of extant horseshoe crabs, although the various elements have undergone fusion into a ridged operculum (Fig. 4). A subrhombic exite lobe, as seen in Sanctacaris (Fig. 2h), is also found in vicissicaudates, particularly Sidneyia (Stein 2013), Emeraldella (Stein and Selden 2012) and to a lesser extent Kwanyinaspis (Zhang and Shu 2005).

Fig. 4
figure 4

Putative homology of trunk exites in Cambrian arthropods and the book-gill operculae of extant chelicerates: a, b Second book-gill opercula of Limulus Polyphemus. a Museum für Naturkunde (ZMB) 32113; b interpretive sketch. c Interpretive sketch of a trunk appendage of Sidneyia inexpectans (redrawn from Stein (2013). Colours indicate putatively homologous elements: basipodite in purple, proximal exite lobe bearing lamellal setae in blue, distal exite lobe in green, telopodite (endopod) in orange, possible sternite in brown, and pre-basinal endite in red

Full size image

New observations and interpretations made herein, particularly the identification of pediform exites tipped with a setal cluster and the presence of a short differentiated posterior tergite, are indicative of artiopodan affinities and seemingly support affinities with chelicerate arthropods (Briggs and Collins 1988). Furthermore, the high number of features shared with Sidneyia and Emeraldella, including previous observations which were confirmed herein, like the presence of a bipartite exopod shaft suggests a close relationship between these taxa, as recovered in a previous phylogenetic analysis (Legg et al. 2013).

Chelicerate-type features in other artiopodans

Given the number of similarities between Sanctacaris, a putative artiopodan, and chelicerates, we might expect other artiopodans to possess features indicative of chelicerate affinities. There is a wealth of literature documenting similarities between trilobites and trilobite-like taxa and chelicerates (Størmer 1944; Lauterbach 1980; Weygoldt 1986; Hou and Bergström 1997). Features previously deemed important include trilobation, genal spines, dorsal positioning of the lateral eyes and the presence of lamellar setae. The importance of these features has been dismissed by others however (Scholtz and Edgecombe 2005) as they are lacking in the majority of chelicerates, particularly pycnogonids and arachnids. The absence of these features from these taxa may actually be an apomorphic reversal, caused, in the case of the arachnids, by an adaptation to a terrestrial habitat, especially as many of these features are prevalent amongst basal members of Chelicerata (Lamsdell 2013).

As well as the aforementioned features, a number of other features are shared by artiopodans, particularly vicissicaudates, and chelicerates but are lacking in megacheirans and other putative stem-lineage euarthropods and stem-mandibulates. To the previous list we can add the presence of a marginal rim on the anterior of the cephalon, differentiation of the posteriormost somites into a short, moveable abdomen, tipped with an elongate, styliform, and potentially keeled telson, and the presence of multi-partite trunk exopods (discussed above).

Additionally, a prebasal endite has been identified in Sidneyia (Bruton 1981) and compared to the prebasal endites of xiphosurans and eurypterids (Boxshall 2004), although these structures were not considered homologous by others (Stein 2013) based on their assumed view of arthropod relationships. Given the structural similarities of the prebasal endites, which are all lacrimiform with a bulbous and spinose medial edge, and their common positioning, on the dorsolateral edge of a gnathobasic basis with an elongate dorsal flange, they are all coded as homologous in this study.

Segmental affinities of the anterior appendages of megacheirans

Recent works on the origins of chelicerates have been dominated by hypotheses regarding the chelate frontal appendages of megacheirans as homologous to the chelicerate chelicerae (Bousfield 1995; Chen et al. 2004; Cotton and Braddy 2004; Dunlop 2005; Scholtz and Edgecombe 2006; Haug et al. 2012; Tanaka et al. 2013), inferring that chelicerates could not have evolved from an antenna-bearing taxon such as the artiopodans. Similarities between short-great-appendages may be overstated however, and there are a number of lines of evidence which place reasonable doubt on their supposed homology with chelicerae.

Perhaps the most compelling line of evidence regarding the segmental affinities of the short-great-appendages has come from recent finds of fossilized neural tissues in the megacheiran Alalcomenaeus sp. (Tanaka et al. 2013). Based on alignment with Limulus and scorpions, the short-great-appendages were considered to innervate from the deutocerebral neuromere of the brain; however, the deutocerebral neuropils of Limulus are expressed anterior of the oesophageal foramen, whereas the neuropils corresponding to the short-great-appendages are located laterally of the oesophageal foramen, which may instead indicate a tritocerebral origin for these appendages. If this is the case, then deutocerebral neuropils are absent. A similar phenomenon occurs in other arthropods which have lost either their deuto- or tritocerebral appendages, such as hexapods (Strausfeld 2012). Pre-great-appendage antennae are present in other, presumably more primitive megacheirans (Legg et al. 2012, 2013; Legg 2013), such as Fortiforceps foliosa (figures 31C–D and 32A in Hou and Bergström 1997), and great-appendage-like appendages are also present in other antenna-bearing taxa including the bivalved arthropod Loricicaris spinocaudatus (figure 3C in Legg and Caron 2014), suggesting that more derived megacheirans, such as Alalcomenaeus, lost an antenna-bearing, presumably deutocerebral, somite.

Fuxianhuia bears two pairs of specialized cephalic appendages, deutocerebral antennae and tritocerebral ‘specialized post-antennal appendages’ (SPAs) (Ma et al. 2012; Yang et al. 2013). The SPAs of Fuxianhuia are geniculate like the short-great-appendages of megacheirans, possibly indicating they are homologous, although it should be noted that the SPAs of Fuxianhuia are not chelate. Geniculation was also used to unite megacheirans and chelicerates; however, many basal chelicerates, such as Dibsterium, Offacolus and possibly Weinbergina (Briggs et al. 2012; Lamsdell 2013), appear to lack this joint (figure 1C in Sutton et al. 2002 and figure 1C in Briggs et al. 2012). In fact, the anterior appendages of these taxa are elongated, composed of numerous podomeres, reminiscent of antennae. If these taxa indeed represent the most basal chelicerates, then it is reasonable to presume that the chelicerae actually originated from an antenniform appendage, akin to that of artiopodans. A similar hypothesis was also proposed by Sharma et al. (2013).

Character acquisition in the chelicerate stem-lineage

Phylogenetic analysis with equal character weighting resulted in 16 most parsimonious trees (MPTs) of 1924 steps (consistency index (CI) = 0.505; retention index (RI) = 0.871), and implied weighted analyses with a concavity constant of 2, 3 and 10 produced 28 MPTs of 179.96797 steps (CI = 0.499; RI = 0.867), 26 MPTs of 146.21588 steps (CI = 0.499; 0.867) and 17 MPTs of 66.65010 steps (CI = 0.503; RI = 0.869), respectively.

In all analyses, Sanctacaris resolved within crown-group Chelicerata (the least inclusive group including pycnogonids and euchelicerates) as the basal-most member of total-group Euchelicerata (Fig. 5). During analyses utilizing equal weighting and implied weighting with a concavity constant of 10, Sidneyia resolved as the sister-taxon of Sanctacaris. This relationship was supported by two synapomorphies: the presence of a paddle-shaped telson (ch. 198:1) and the differentiation of endopod endites into primary and secondary subsets (ch. 298:1). When analysed using implied weighting with a concavity constant of two or three Sidneyia instead resolved as a vicissicaudate, the entire group of which in turn is resolved as the sister-taxon of Chelicerata (Fig. 5). Both topologies indicate elongate, antenniform appendages were present in the chelicerate stem-lineage and were retained in selected crown-members such as Dibasterium and Offacolus. If this is the case, then we need to account for the position of Pycnogonida, whose position in the current analysis indicates a convergent origin of short, chelate appendages. The position of pycnogonids may, however, be a long-branch artefact caused by the extensive modification of the pycnogonid body plan.

Fig. 5
figure 5

The phylogenetic position of Sanctacaris uncata: (a) topology produced using implied character weighting with concavity constants of two and three and (b) topology produced using equal character weighting and implied character weighting with a concavity constant of ten. The megacheirans resolved as part of a paraphyletic assemblage outside of the arthropod crown-group. Trilobitomorpha includes trilobites and a diversity of trilobite-like, pygidium-bearing taxa including Cindarella (Petalopleura), Kumaia (Conciliterga) and Naraoia (Nektaspida). See Fig. 1 for alternative placements of these taxa

Full size image

A sister-taxon relationship between chelicerates and vicissicaudates (regardless of the inclusion of Sidneyia) was supported by a number of synapomorphies mostly pertaining to tagmosis. In particular, this clade, termed Cheliceratomorpha (sensu Cotton and Braddy 2004) primitively possess a trunk composed of 12 segments (ch. 75:5), with a differentiated abdomen (ch. 93:1, 94:1), composed of three segments (ch. 95:2), although numerous variations on this body plan occur within derived members of this clade.

A bi- or tripartite exopod shaft (ch. 152:1) with lamellar setae restricted to the proximal shaft has previously been considered a diagnostic characteristic of artiopodans (Stein and Selden 2012); however, basal members including retifaciids, Kiisortoqia, Siriocaris, and Squamacula lack this feature. The retifaciids possess a flap-like exopod with extensive lamellar setae and appear to represent an intermediate stage between the simple flap-like exopod of non-artiopodans and the lamellate exopods of later artiopodans (Fig. 5). The fusion of the trunk limbs into operculae, as seen in the euchelicerate crown-group, may have been a crucial step in the diversification and terrestrialization of chelicerates as the operculae act as a protective cover for the respiratory lamellae, preventing them from desiccation during terrestrial excursions (Reisinger et al. 1991).

A single synapomorphy—the presence of an extensive cephalic doublure (ch. 24:1)—supports the monophyly of Artiopoda. This feature was excluded from prior diagnoses of this group (Stein and Selden 2012) but is retained in most members of this clade. This feature is lost within crown-group chelicerates, specifically within pycnogonids and arachnids. Other features previously used to link non-chelicerate artiopodans and chelicerates, such as short genal spines (ch. 38:1), an inflated posterior cephalon (ch. 31:1), a raised marginal cephalic rim (ch. 16:1) and a raised axial trunk region (ch. 71:1), show varying levels of homoplasy across tree topologies; however, these features tend to be found exclusively amongst artiopodans (including chelicerates) and may still serve as a good indicator of relationships between non-chelicerate artiopodans and chelicerates.

The topology obtained in this study indicates that a number of features prevalent amongst chelicerates appeared sequentially in the chelicerate stem-lineage (Fig. 5). Invoking a megacheiran ancestry for chelicerates involves an abrupt morphological change at the base of Chelicerata, whereby many of the features linking them to artiopodans are convergently acquired, thereby resulting in an unparsimonious view of their evolution.