Newly discovered morphology of the Silurian sea spider Haliestes and its implications

The three‐dimensionally preserved Haliestes dasos from the Silurian (Wenlock) Lagerstätte is the most complete fossil sea spider and the oldest unambiguous pycnogonid known from the fossil record. The discovery of two new specimens to add to the holotype reveals new features including proximal annulations of the appendages and segmentation of the trunk end, critical details for comparison with pycnogonids from the Devonian (Emsian) Hunsrück Slate and for the interpretation of the evolutionary significance of Palaeozoic genera. There is some evidence of sexual dimorphism. Haliestes dasos was nektobenthic and its morphology indicates an unusual mode of feeding compared with living pycnogonids. The new morphological features of H. dasos are closely similar to those in Palaeoisopus problematicus from the Hunsrück Slate and it clearly belongs, together with that species, in stem Pycnogonida and not the crown group.

P Y C N O G O N I D S (sea spiders) are marine arthropods that typically show a distinct suite of morphological characters: a cephalosoma bearing an eye tubercle, a paired chelifore, palp and oviger, and a proboscis; a small trunk with four, rarely five or six, paired uniramous walking legs; and a reduced trunk end (Vilpoux & Waloszek 2003; referred to as abdomen in traditional terminology, e.g.Stock 1975;Bamber 2007;M€ uller & Krapp 2009).Pycnogonids have been considered either as the sister group to other euarthropods (e.g.Zrzav y et al. 1998) or, in recent studies, as the sister group to Euchelicerata within Chelicerata (e.g.Dunlop & Arango 2005;Giribet & Edgecombe 2013, 2019;Ballesteros et al. 2021).Some 1400 extant pycnogonid species (Bamber et al. 2023) occur globally from littoral to abyssal deep-sea environments down to at least 6000 m.Their diversity is highest in the Southern Ocean (Munilla & Soler-Membrives 2015) where most species with five or six pairs of legs occur, as well as giant examples with legs more than 240 mm long (King 1973;Arnaud & Bamber 1987).
The fragile nature of pycnogonids, combined with their thin, unbiomineralized cuticle, is reflected in their sparse fossil record.Cambropycnogon from the Cambrian Orsten deposits of Sweden is known from only seven larval specimens (Waloszek & Dunlop 2002), and the single Ordovician genus, Palaeomarachne from Manitoba, Canada, is based on four moults (Rudkin et al. 2013).The identification of these two genera as pycnogonids is still regarded as uncertain by some (Bamber 2007;Sabroux et al. 2019).The monotypic Haliestes from Herefordshire, UK, is the only known Silurian genus (Siveter et al. 2004).The greatest Palaeozoic diversity occurs in the Devonian (Emsian) Hunsr€ uck Slate of Germany, which has yielded five monotypic genera, Palaeoisopus, Palaeopantopus, Palaeothea, Flagellopantopus and Pentapantopus (Bergstr€ om et al. 1980;Poschmann & Dunlop 2006;K€ uhl et al. 2013).Mesozoic pycnogonids, which are more closely related to extant species (Pantopoda), are known from the Jurassic of La Voulte-sur-Rhône, France (Charbonnier et al. 2007) Sabroux et al. (2019Sabroux et al. ( , p. 1935) ) inferred that H. dasos and other fossil pycnogonids 'belong to early offshoots of Pycnogonida' and Sabroux et al. (2023, p. 17) argued that the assignment of H. dasos to the crown group (Order Pantopoda, to which all living pycnogonids belong) is problematic, noting that its abdomen may be segmented whereas a reduced unsegmented terminal tagma is diagnostic of Pantopoda.
Herefordshire Lagerst€ atte fossils are preserved in exquisite three-dimensional (3D) detail as calcitic infills in concretions in fine, marine-deposited volcaniclastics (Orr et al. 2000;Siveter et al. 2020).Two new specimens of H. dasos from there add to the evidence of the holotype (Siveter et al. 2004) and provide novel insights into the morphology and biology of Haliestes which bear on its affinity to living pycnogonids and their evolutionary history.

METHOD AND TERMINOLOGY
The two new specimens of Haliestes dasos were serially ground and photographed at 20 lm intervals.The images were aligned, edited and used to generate 3D virtual fossils with the SPIERS software suite (Sutton et al. 2014;Spencer et al. 2020) for on-screen stereoscopic investigation.The 3D models in VAXMLT/STL format of the holotype and the two new specimens, plus the datasets from the serial grinding, are housed in the Oxford University Museum of Natural History (OUMNH) and in the Dryad Digital Repository (Sutton et al. 2023).The terminology used follows that of Siveter et al. (2004) except that chelifore, article and claw are used instead of chelicera, appendage segment and subchela, respectively.The terms used for individual articles follow Bamber (2007;after Child 1979), with the introduction of metatibia for article 7 of walking legs 2-4 (WL2-4) of H. dasos.Holotype.OUMNH PAL-C.29571 (Fig. 2).
Diagnosis.As for the genus.

NEW DATA ON MORPHOLOGY
The two new specimens are slightly larger than the holotype.The sagittal length from the anterior of the cephalosoma to the posterior of the trunk end is 5.7 mm in OUMNH PAL-C.29661 (Fig. 3A) and 4.4 mm in OUMNH PAL-C.29656 (Fig. 4A) compared with 3.5 mm in the holotype (Fig. 2A).

Cephalosoma
A small boss-like feature is present anteroventrally on the cephalosoma, between the eye tubercle and the proboscis (Figs 2L, 3K, 4G); its nature is unknown.
The chelifore of both new specimens (Figs 3L, 4B) is like that of the holotype (Fig. 2C) (Siveter et al. 2004).The original description of the palp identified at least eight articles including the claw (Siveter et al. 2004).The evidence of one of the new specimens (Fig. 3B), however, prompted reconsideration of the holotype (Fig. 2F) and indicates that there were 10 articles including the claw, which is a unique feature on the palp of sea spiders.The remaining nine correspond to the configuration of the palp in most extant representatives of Ascorhynchidae and Colossendeidae, as well as the ammotheid genera Ammothea, Ammothella, Cilunculus and Nymphopsis.The oviger was originally considered to comprise nine articles (Siveter et al. 2004).The evidence of the two new specimens and reconsideration of the holotype indicates that 10 were present (Figs 2G,3G,4I).This contrasts with the typical number in living sea spiders, which is 10 plus the claw (Bamber 2007), although there is variability (e.g. in Endeidae, Phoxichildiidae and Pycnogonidae).
The two new specimens confirm that the lumen of the proboscis is triradiate in transverse section (Figs 3E, 4F;Siveter et al. 2004, fig. 1C), reflecting the arrangement of antimeres typical of extant pycnogonids (Miyazaki 2002).
Haliestes remains the only fossil pycnogonid to preserve this feature.

Walking legs
The new specimens show the pronounced, presumably in vivo, forward flexion of the walking legs between articles 1 and 2 evident in the holotype (Figs 2A, B, 3A, C, 4A, C), identifying the proximal article as coxa 1 (Siveter et al. 2004, supplementary note 1).The succeeding article, coxa 2, is conspicuously longer than coxa 1 and coxa 3 (Fig. 1D) as in a number of extant pantopods (e.g.Anoplodactylus madibenthos Sabroux et al. (2022, fig.22F, I)).The ventrodistal margin of coxa 2 in OUMNH PAL-C.29661bears a low protuberance (e.g.Fig. 3N, P, Q) similar to the site of the gonopore on coxa 2 in extant female specimens (e.g. A. madibenthos; Arnaud & Bamber 1987, fig.13).A single seta is present on at least one of these protuberances (Fig. 3R), although the significance of this is uncertain.
The new specimens confirm that WL1 comprises nine articles including the claw (Siveter et al. 2004), as in WL1-4 of extant pantopods (Fig. 1B, D), and it is reasonable to assume that the articles are homologous (Figs 1A,  B, D, 2K, 3P, 4D).The short tarsus (article 7) is distinguished from an arthrodial membrane by the presence on its inner ventral margin of paired setae (Fig. 2H, K).Nor does it represent a propodus heel (e.g.Arnaud & Bamber 1987, fig.8C, F), given that the outline is convex both dorsally and ventrally (Figs 2K, 3Q).
The number of articles in WL2-4 (Fig. 2E), each leg of which is successively longer than WL1, was undetermined in the holotype (Siveter et al. 2004).The new specimens show that there are 10 (Figs 3I, Q, 4D), in contrast to the nine in WL1.The most proximal four (coxa 1, 2 and 3, and femur) and distal three (tarsus, propodus and claw) are similar in WL1 and WL2-4 and presumably homologous.Thus article 7 of WL1 (Figs 2E, H, K, 3P) is homologous with article 8 of WL2-4 (Figs 3I, O, Q, 4D).A comparison of the length and form of the intervening articles in WL1 and WL2-4 indicates that article 7 in WL2-4 (between tibia 2 and the tarsus) is the additional one, which we term the metatibia.The metatibia may be the product of the introduction of a new articulation point during embryonic development (Boxshall 2004).
Annulations, which are a feature of the appendages of almost all Hunsr€ uck pycnogonids (Bergstr€ om et al. 1980), were originally considered to be lacking in Haliestes (Siveter et al. 2004).However, the new specimens show them on all walking legs, and re-investigation of the holotype also revealed them, albeit convincingly preserved only on WL1.The walking legs of Haliestes are borne by lateral processes projecting from the trunk.The process bearing WL1 is more uniform in width than those to which WL2-4 are attached, which flare in width distally (Figs 2A, 3A, J, M, 4A, K).Article 1 (coxa 1) of WL1 shows three annulations and article 1 of WL2-4 shows five.In addition, there is an annular structure at the end of the lateral processes that bear these limbs.The annulations associated with WL1 are relatively wider transversely than those of WL2-4.There is some evidence of tuberclelike structures on the annulations (Fig. 3J), but these may represent an artefact where annulations and the reconstruction slices intersect.et al. (2004) determined the short trunk end of the holotype to be composed of three elements that meet at a slight angle.It was not clear whether these elements represent three segments or whether the trunk was made up of a bent single segment as in some extant species (e.g.Eurycyde clitellaria; M€ uller & Krapp 2009, fig.30).The new specimens show further variation in the angles between the three elements.The trunk end in OUMNH PAL-C.29661 (Fig. 3D) is almost straight whereas the proximal element is inclined moderately upwards and the distal one strongly downwards at a right angle in OUMNH PAL-C.29656 (Fig. 4E).There is no evidence of taphonomic deformation.It is also unlikely that these differences correspond to individual non-segmental variability because the three specimens show marked variation in trunk end curvature, to a degree not observed in extant species.The evidence indicates that the three elements represent articulating segments.

Siveter
The posterior end of the middle segment bears an upwardly projecting node-like structure with two pairs of setae, one pair directed posterodorsally, the other pair posteroventrally (Figs 2I, 3D, 4E).This structure probably represents the type of dorsal ornamentation that occurs in some extant species (e.g.Eurycyde muricata; Child 1995, fig.5).The position of the anus is unknown.

COMPARISON WITH HUNSR € UCK SLATE PYCNOGONIDS
The greatest diversity of Palaeozoic pycnogonids occurs in the Hunsr€ uck Slate, in which the fossils are pyritized.Palaeoisopus problematicus, the largest and most abundant of them, has nine articles in WL1 and 10 in WL2-4 (Bergstr€ om et al. 1980, figs 10, 17b, 20), presumably homologous with those now resolved in Haliestes (and with reference to WL1, in pantopods) (Fig. 1A, C, D).Article 2 (coxa 2) is the longest of the three coxae in both.The proximal part of article 5 (tibia 1) is acutely pointed in Palaeoisopus (Fig. 1C; Bergstr€ om et al. 1980, figs 2, 10, 11, 16, 19) and there is some indication (Fig. 3P) that this is also the case in Haliestes.There are nine articles in the walking limbs of Pentapantopus vogetli, at least nine in those of Palaeopantopus maucheri, and the number in Flagellopantopus blocki and Palaeothea devonica is uncertain because their limbs are more incomplete (Bergstr€ om et al. 1980;Poschmann & Dunlop 2006;K€ uhl et al. 2013).

SEXUAL DIMORPHISM
The discovery of additional specimens enables the possibility of sexual dimorphism to be considered (Arnaud & Bamber 1987, pp. 3-23).The ovaries of extant sea spiders extend into the legs, where vitellogenesis occurs (King & Jarvis 1970;Jarvis & King 1972;Miyazaki & Bili nski 2006;Miyazaki & Makioka 2010;Brenneis et al. 2023).Oocytes are typically stored in the femora in mature female specimens, but in some instances an enlarged coxa 2 is used (e.g.Ascorhynchus ovicoxa Stock, 1975).Thus, the enlarged coxa 2 of WL2-4 in OUMNH PAL-C.29661 (Fig. 3C, I, N, P, Q) may identify it as female, differing from OUMNH PAL-C.29571 (Fig. 2) and OUMNH PAL-C.29656 (Fig. 4), which are male.The trunk of the female Haliestes is relatively stouter than that of the male specimens (Figs 2A, 3A, 4A), and the female specimen is the largest of the three, which is consistent with size differences in extant pycnogonids.The protuberance on coxa 2 of the walking legs in the female specimen (Fig. 3N, P, Q) may represent the position of the gonopore.A similar structure in at least one of the other specimens (Fig. 2K) may represent the male gonopore as in modern sea spiders (e.g.Arnaud & Bamber 1987, fig.14).There is no evidence of a cement gland aperture on the femur of the walking legs in the supposed male specimens, a feature present in extant species, but it might not be visible in this kind of preservation.
Female and male specimens of living species often show significant morphological differences in their ovigers, which can be more developed in the male sex (e.g. in Ammotheidae, Callipallenidae, Nymphonidae and Pallenopsidae) or totally absent in the female sex (e.g. in Endeidae, Phoxichilidiidae and Pycnogonidae).However, the available specimens of Haliestes, in common with species of Ascorhynchidae, Colossendeidae and Rhynchothoracidae, and most of Austrodecidae, do not show such dimorphism.Similarly, there are no obvious differences in the palps of the supposed female and male specimens, in contrast to several genera within Callipallenidae.Thus, Haliestes may be an example of a pycnogonid with very little sexual dimorphism, as in species of Colossendeidae.

FUNCTIONAL MORPHOLOGY
The appendages of Haliestes, like those of Palaeoisopus (Bergstr€ om et al. 1980, p. 32, fig. 20), are flattened with paddle-like articles distally, suggesting a more nektobenthic lifestyle than the benthic habit of most extant sea spiders.The long terminal claw on the walking legs indicates a raptorial function.They may have been used to cling onto surfaces, such as the body of sponges that are abundant in the Herefordshire fauna (Nadhira et al. 2019;Siveter et al. 2020, fig. 2a).
The enlarged coxa 2 of OUMNH PAL-C.29661 (Fig. 3) suggests that the ovaries of Haliestes extended into the legs as in extant species, and the location of the presumed gonopores is similar to that in modern forms.Oocytes are laid by the female of living species through the gonopores before the male fertilizes them (Nakamura & Sekiguchi 1980).The male then collects the eggs on the ovigers.However, the oviger of Haliestes differs from that of extant species in lacking a strigilis, the curved, hook-like feature formed by the most distal four articles (e.g.Child 1979, fig. 20).Additionally, the palp of Haliestes differs from that in living species in bearing a terminal claw, which in extant species (although not in all families) exists only in the oviger.The palp and oviger are alike in structure, are positioned surrounding the proboscis (Figs 3F, 4H) and do not display sexual dimorphism.Thus, unlike those in most extant species, we interpret their role as primarily related to feeding.Extant sea spiders feed with their proboscis on sources varying from biofilms and algae to anthozoans and molluscs (Dietz et al. 2018), and modern colossendeids can use their palps to constrain prey (Braby et al. 2009).However, most modern species use their ovigers primarily as egg-carrying appendages (Arnaud & Bamber 1987), which differs from the functional interpretation presented here for Haliestes.
chelifore in Haliestes could interact with the mouth at the end of the long proboscis.The chelifore may have been used for subduing prey or grasping sessile benthos such as sponges or crinoids when at rest.Haliestes may have used the palps and ovigers to manipulate or stabilize prey (Fig. 5).The triradial jaws of extant species grasp and abrade soft tissues and feed on, for example, the tentacles of sea anemones (Wyer & King 1974; Fig. 5A).Alternatively, the proboscis may have penetrated the body wall of its prey (Fig. 5B), in a similar fashion to Pycnogonum litorale (Wyer & King 1974).Whereas P. litorale uses its first two walking legs to facilitate tissue penetration by the proboscis, Haliestes may have used the palps and ovigers.In either case it is likely that Haliestes used its walking legs to remain stable on large prey (e.g.poriferans, bivalves or echinoderms; Arnaud & Bamber 1987).F I G . 5 .Diagrammatic representation of two non-exclusive hypotheses indicating the feeding strategy of Haliestes dasos.A, the prey is maintained at the tip of the proboscis by the palps and ovigers; the triradial jaws abrade the soft tissues of the prey to feed.B, the specimen stabilizes over the prey by using its palps and ovigers; after abrading the surface with its triradial jaws, the strength of its palps and ovigers is used to push the proboscis through the prey's body wall, into its tissues.

CONCLUSION
Even though Haliestes dasos was the most completely known fossil pycnogonid based on a single specimen (Siveter et al. 2004), two new examples add significant details regarding its phylogenetic position.The presence in Haliestes of coxal annulations, features previously known only in Devonian pycnogonids, is determined for the first time.Also, there is now clear evidence of a tri-segmented trunk end in this Silurian genus.These features imply a position in the pycnogonid stem.Thus, the age of H. dasos from the Herefordshire Lagerst€ atte is not an appropriate minimum for crown group Pycnogonida (i.
S I V E T E R E T A L .: S I L U R I A N S E A S P I D E R H A L I E S T E S 9 The new Haliestes specimens have a multi-segmented trunk end, as in other Palaeozoic taxa (Palaeoisopus, Palaeopantopus, Flagellopantopus), but distinct from Pantopoda.They also show that Haliestes and Hunsr€ uck Slate pycnogonids share other characters that are absent in Pantopoda, particularly annulations on the leg bases (Bergstr€ om et al. 1980).A re-assessment of the phylogenetic position of Haliestes requires a new analysis including all fossil and representative extant pycnogonids, integrating new morphological and genomic data yet to be obtained.In the meantime the new evidence presented here strongly indicates that Haliestes belongs in a clade with the Hunsr€ uck genera (excepting Palaeothea), which have been determined as stem forms (e.g.Poschmann & Dunlop 2006).
and Solnhofen, Germany (Sabroux et al. 2019).The discovery of Haliestes dasos in the Herefordshire Lagerst€ atte (Wenlock Series, c. 430 Ma; Siveter et al. 2020) represents a reference point in our understanding of sea spider evolution given that it is widely considered the oldest unambiguous record of Pycnogonida (e.g.Arango & Wheeler 2007; Bamber 2007; Sabroux et al. 2019) and is the most completely known fossil species (Siveter et al. 2004).Wolfe et al. (2016) and Ballesteros et al. (2021) used it as a calibration point (soft minimum) for the age of the crown group.The phylogenetic position of H. dasos, however, remains uncertain.Siveter et al. (2004, p. 980) obtained a poorly supported position 'near or in the pycnogonid crown group', which they regarded as preliminary.The analysis of Arango & Wheeler (2007), based on sequence and morphological data, recovered H. dasos and other fossil taxa within the crown group associated with some ammotheids.Bamber (2007) erected a new Order Nectopantopoda to accommodate H. dasos but did not justify this decision in a cladistic context.