AM F: Australian Museum, Sydney, NSW, Australia; BMSC: Buffalo Society of Natural Sciences, Buffalo, New York State, USA; CM: Carnegie Museum of Natural History, Pittsburgh, Pennsylvania, USA; YPM IP: Yale Peabody Museum, Diversion of Invertebrate Paleontology New Haven, Connecticut, USA; USNM PAL: United States National Museum, Washington, DC, USA.

“Bellinurus” carteri (note that, although Pickett (1993) emended the feminine suffix to “B”. carterae, the species spelling of carteri must be retained as “Article 32.5.1. (ICZN, 1999) states that incorrect latinization must not be corrected, and the original spellings be maintained” (Krell et al., 2017, p. 8)) was originally described by Eller (1940) from one specimen (BMSC E 9644) from the Hanley Quarry, an outcrop of the lower Cattaraugus Formation (Late Devonian, Fammenian), Pennsylvania, USA (Tesmer, 1975). BMSC E 9644 is preserved as a slightly domed internal mould without cuticle on a slab of dark-reddish sandstone and lacks a counterpart. BMSC E 9644 was photographed with a Canon EOS 5D Mark IV under normal and low angle light. Measurements were obtained from photographs using ImageJ.

Subphylum Chelicerata Heymons, 1901.

Class Xiphosura Latreille, 1802.

Pickettia gen. nov. (Fig. 1).

Pickettia gen. nov. (Fig. 1).

1940 Belinurus carteri Eller, p. 134, Fig. 1.

1982 Belinurus carteri Eller; Fisher, 1982, p. 177, Fig. 1.

1984 Belinurus carteri Eller; Fisher, p. 199, Fig. 2.

1987 Bellinurus carteri Eller; Siveter and Selden, p. 384, Fig. 1F.

1993 “Bellinurus carterae” Eller; Pickett, p. 282.

1994 Bellinurus carteri Eller; Schultka, 1994, p. 347.

1997 Bellinurus carterae Eller; Anderson and Selden, p. 22.

2013 “Bellinurus” carterae Eller; Lamsdell et al., p. 368.

2016 “Bellinurus” carterae Eller; Lerner et al., p. 208.

Etymology. The generic name Pickettia is used in honour of John Pickett and for his contributions to invertebrate palaeontology, specifically with focus on horseshoe crabs, across his career.

Type and only species. Belinurus carteri.

Emended diagnosis. Distinguished from other stem xiphosurids by an anterior-most opisthosomal tergite overdeveloped into a free lobe, and opisthosomal tergites that terminate in pleural spines.

Holotype specim en. BMSC E 9644.

Locality, horizon and age. Hanley Quarry, lower Cattaraugus Formation, northwestern Pennsylvania, Late Devonian (Fammenian).

Preservation. BMSC E 9644 is preserved as an internal mould on yellow-red sandstone. The slightly vaulted nature of the exoskeleton suggests the taxon was vaulted in life. No counterpart reported.

Description of holotype. BMSC E 9644 is an articulated prosoma, opisthosoma, and partial telson (Fig. 1). A small degree of three-dimensional relief is present. Specimen is 39.8 mm long, including partial telson. Prosoma is semi-circular and only partly preserved; most of the left side and posterior section of right side are preserved. Distance between the posterior margin of left prosomal side and prosomal midline is 17.8 mm. Prosomal rim is preserved and has a maximum width of 0.57 mm. No prosomal doublure is visible. Left ophthalmic ridge is slightly preserved and not parallel to cardiac lobe (Fig. 1A). Ophthalmic ridge is straight and 7.47 mm long. No lateral compound eyes can be confidently identified. Cardiac lobe is convex, cone-shaped, and both sides of the posterior section are preserved (Fig. 1A). Cardiac lobe is 8.36 mm wide posteriorly, tapering to ∼4.5 mm anteriorly, although the anterior-most section is not preserved. Lobe is bordered by two interophthalmic ridges. Left interophthalmic ridge is 9.79 mm long and right interophthalmic ridge is 5.5 mm long. No obvious cardiac ridge is preserved. Ocelli are not observed. A subtriangular triradiate node appears to be present on the left side of the posterior boarder of the prosomal shield. The node is located 8.7 mm to the left of the prosomal midline and 5.02 mm from the left lateral prosomal border. Node leads into what appears to be a 2.8 mm long cheek ridge (Fig. 1E). Left genal spine extends 9.6 mm posterolaterally from the prosomal-opisthosomal hinge to anteroposterior midpoint of the opisthosoma. Lateral extent between preserved genal spine tip and opisthosoma is 6.4 mm. Angle between genal spine and opisthosoma is 57.8°. Genal spine is curved to maintain a horseshoe shape (Fig. 1A–C). Inner margin of genal spines is curved anteriorly. No prosomal appendages are preserved.

Preservation of the opisthosoma is better than that of the prosoma. Opisthosoma is trapezoidal, 17.2 mm long and 21.8 mm wide (where both sides are preserved), tapering to 6.3 mm posteriorly. Opisthosoma is segmented with the expression of tergites VIII–XV. Most anterior tergite (VIII) is a prominent free lobe (tergal expression of somite VIII) and extends laterally out to the genal spine tip (Fig. 1C). Only the left side of the free lobe is preserved. The most distal point of the left VIII tergite is 14.61 mm from the opisthosomal midline. The subsequent tergites are smaller: XV is 30% width of VIII. Tergites VIII–XIV are 1.8 mm long, and XV is 3.03 mm long, suggesting that it is a pretelson section (Fig. 1A). Opisthosomal axis is lobe-shaped: 17 mm long, 9.9 mm wide anteriorly, tapering to 4.76 mm posteriorly. No apodemes are noted. Left pleural lobe is flat, lacks relief and is segmented. Excluding overdeveloped VIII tergite, the left plural lobe is 6.7 mm wide anteriorly, tapering to 0.88 mm posteriorly. Marginal rim is poorly defined but when noted is ∼0.5 mm wide. Right pleural lobe is incompletely preserved. Tergites VIII to IX are partly preserved. Tergites X–XV are completely preserved. Right side of the right lobe is flat, lacks relief and is segmented. Lobe is 5.01 mm wide anteriorly (where complete), tapering to 0.69 mm posteriorly. The right marginal rim is slightly better preserved than the left side and is up to 0.5 mm wide. Tergites IX–XIV terminate with triangular pleural spines. On the left opisthosoma side, pleural spines are triangular and the length of posterior spine side decreases from 3.84 mm (tergite IX) to 2.19 mm (tergite XIV). The distance between spine tips to left opisthosomal border decreases from 3.6 mm (tergite IX) to 2.6 mm (tergite XIV). On the right opisthosomal side, only tergites X and XI have spines preserved. Length of posterior spine side is 3.8 mm for both spines. Distance of distal spine tips to right opisthosomal border is 3.5 mm. A cololite (fossilised gut contents) is preserved along the opisthosomal axis, spanning tergites VIII–XI and is 6.5 mm long (Fig. 1F). The telson is lanceolate and articulated with the posterior opisthosoma margin. A sub-circular hole is present at the opisthosoma-telson joint is noted (Fig. 1D). Telson is incomplete; the preserved portion is 11.5 mm long. No axial ridge is present noted along telson. Anterior section of telson is 3.39 mm wide tapering to 1.28 mm. Posterior tip of telson has a slight transverse ridge.

Remarks. The triradiate node and the cheek ridge were not mentioned in Eller (1940). These features are known for Paleolimulidae (Siveter and Selden, 1987), but in this case may reflect taphonomic distortion of the specimen; more specimens are needed to determine if this is indeed a real feature. The hypertrophic free lobe is interpreted as the first opisthosomal tergite. The movable spines described by Eller (1940) are reinterpreted here as pleural spines. The pretelson is substantially more pronounced than Eller (1940) suggested. We also propose that the ophthalmic ridges would have probably converged into an “M”-shaped double arch (Fig. 1B), as is observed in other stem xiphosurids from the Late Devonian of the USA (Babcock et al., 1995).

The phylogenetic analysis resulted in 18 most parsimonious trees of length 738 (CI: 0.472; RI: 0.879). The topology of the strict consensus tree is similar to that presented in Lamsdell (2016) and Selden et al. (2015); see Fig. 2) in hypothesising a monophyletic Xiphosurida, with synziphosurines represented as polyphyletic grade (sensu Lamsdell, 2013) forming the basal members of the sister taxon of, Xiphosurida + stem xiphosurids. Pickettia carteri resolves within a polytomy containing "Kasibelinurus" randalli (Beecher, 1902) and Lunataspis aurora Rudkin et al., 2008 and immediately crownward of K.  amicorum Pickett, 1993). All four species form successive outgroups to Xiphosurida. The implied weighting analysis resulted in a single most parsimonious tree under both concavity constants and differed only in "K". randalli and P. carteri forming a polytomy with L. aurora + Xiphosurida (k = 3), and the complete resolution of the polytomy of "K". randalli, L. aurora and P. carteri as successive outgroups of Xiphosurida (k = 12).

Pickett (1993) suggested that Pickettia carteri belonged in Kasibelinuridae. Here, we consider P. carteri a stem xiphosurid as Kasibelinuridae is a paraphyletic group: a notion supported by Lamsdell (2016). As stem horseshoe crabs, these taxa have features observed in Paleolimulidae and Belinuridae. Pickettia carteri has possible evidence of a triradiate node and cheek ridge diagnostic of Paleolimulidae (Lamsdell, 2016 and Siveter and Selden, 1987). Furthermore, P. carteri lacks movable spines, a derived character observed in Limulina and unknown from Belinurina. Finally, the pronounced pretelson section in P. carteri is a feature known from at least Paleolimulus woodae Lerner et al., 2016 and Bellinuroopsis rossicus Chernyshev, 1933. In light of the character combination presented here, there is no justification for placing this taxon within Belinuridae. The taxon does not belong in the stem genus Kasibelinurus as P. carteri has a hypertrophic free lobe and genal spines extending posteriorly to the third opisthosomal tergite (X): both features not known from Kasibelinurus (sensu Pickett, 1993).

Stem xiphosurids are an apparently rare grade of horseshoe crabs and their early divergence and basal phylogenetic position have presented issues regarding a clear understanding of their place within Xiphosurida. Furthermore, few studies have considered which taxa might belong in this grade (Babcock et al., 1995, Lamsdell et al., 2013 and Pickett, 1993). A consideration of select Devonian taxa therefore seems appropriate and four Devonian-aged taxa in Dunlop et al. (2019) are considered (Fig. 3): “Bellinurus” alleganyensis (Eller, 1938b) from the Late Devonian (Famennian)-aged Chadakoin Formation, New York, USA (Babcock et al., 1995); Kasibelinurus amicorum from the Late Devonian (Famennian)-aged Mandagery Sandstone, Australia; “K” randalli from the Late Devonian (Famennian)-aged Venango and Chadakoin formations, Pennsylvania, USA (Babcock et al., 1995); and Pickettia carteri. Comparisons between these four taxa suggest that “K.” randalli is the same taxon as “B.” alleganyensis (Fig. 3B–G) as the two taxa have very comparable prosomal and opisthosomal morphologies. Furthermore, both taxa are found in the Chadakoin Formation. Building on Babcock et al. (1995), “B.” alleganyensis is considered a junior synonym of “K.” randalli. Additionally, “K.” randalli should be moved out of the genus Kasibelinurus. This change is suggested as the holotype of Kasibelinurus, K. amicorum, lacks the double arched, “M”-shaped, ophthalmic ridge described in “K”. randalli (compare Fig. 3A with Fig. 3B–D) (Lamsdell et al., 2013 and Selden and Siveter, 1987). Together, these data suggest that stem xiphosurid diversity has been overstated and thorough taxonomic revision is needed to uncover the true diversity of stem horseshoe crabs.

Although the life mode and the diet of basal horseshoe crabs have been considered briefly in Lamsdell et al. (2013), we will explore this topic in more detail here. Pickettia carteri was likely a marine organism (see paleoenvironmental discussion of the Cattaraugus Formation in Tesmer, 1967) and therefore similar to other pre-Carboniferous horseshoe crabs (Lamsdell et al., 2013 and Moore et al., 2007). The marine trace fossil Selenichnites sp. (Romano and Whyte, 1987) has been described from localities close to the Hanley Quarry (Eller, 1938a, Eller, 1940 and Williams, 1885). Selenichnites represents “partial burial of the anterior (exoskeletal) region (of a horseshoe crab) while working the sediment with its appendages, presumably for concealment or in the search for food” (Whyte and Romano, 2013, p. 205). Babcock et al. (1995) suggested that “Kasibelinurus” randalli may have produced the Devonian-aged Selenichnites trace fossils, and by extension we suggest that P. carteri may have made similar traces. Devonian horseshoe crabs were therefore potentially capable of burrowing and, by extension, finding and consuming prey in a similar fashion to extant horseshoe crabs (Bicknell et al., 2018a). Although appendage data is not known for P. carteri, it likely had an appendage set similar to extant and fossil taxa (Bicknell et al., 2019b, Bicknell et al., 2018a, Haug et al., 2012, Lamsdell and McKenzie, 2015 and Racheboeuf et al., 2002; Shultz, 2001). Pickettia carteri likely used gnathobases on prosomal appendages II–VI to masticate the abundant shelly and possible soft prey (Caster, 1930) with feeding mechanisms similar to extant horseshoe crabs (Bicknell et al., 2018b, Bicknell et al., 2018c, Botton, 1984 and Razak and Kassim, 2018). Further exploration of this topic would require determination of the contents of the preserved cololite using non-destructive three-dimensional imaging techniques such as computed tomographic scanning. Similar cololites have been documented in specimens of Tachypleus syriacus (Woodward, 1879) (Bicknell et al., 2019b), and identification of gut contents of a range of basal xiphosurans would allow significant insights into the evolution of predator – prey interactions within basal horseshoe crabs (Bicknell and Paterson, 2018). However, this is beyond the scope of the current study.

Stem xiphosurids have been placed into the group Kasibelinuridae (sensu Lamsdell, 2016). Due to this paraphyletic status, we have refrained from referring our new taxon to the group. It is possible that the group is invalid and, if so, should no longer be used; further phylogenetic studies are needed to determine if this is the case. A particularly acute problem for coding xiphosurans into phylogenetic matrices is the lack of representative taxa that preserve characters pertaining to the appendages. So far, these data are limited primarily to Alanops magnificus Racheboeuf et al., 2002, Euproops danae (Meek and Worthen, 1865), Tachypleus syriacus (Yunnanolimulus luopingensis Zhang et al., 2009 (Bicknell et al., 2019b, Haug and Rötzer, 2018, Haug et al., 2012, Hu et al., 2017 and Lamsdell and McKenzie, 2015). The discovery of more xiphosurans and xiphosurids with complete appendages would therefore contribute a plethora of new morphological data that can aid in refining phylogenetic hypotheses. Other sources of data aside from discrete characters, such as continuous and morphometric data, also offer potential avenues for resolving xiphosuran interrelationships. Continuous characters have implementations under both parsimony and Bayesian criteria (Goloboff et al., 2006 and Höhna et al., 2016) and both two- and three-dimensional morphometric data can be analysed as such in TNT 1.5 (Goloboff and Catalano, 2016). Morphometric data in particular have another use: they can be used to explore morphological evolutionary patterns of the group to uncover the tempo and mode of horseshoe crab evolution (Bicknell et al., 2019c). Finally, the involvement of genomic data to produce a total evidence phylogeny, building on the work of Ballesteros and Sharma (2019), will undoubtedly prove central to uncovering the timing and patterns of major diversification events in Xiphosura, and other chelicerate groups (Giribet and Edgecombe, 2019).