Pictorial Atlas of Fossil and Extant Horseshoe Crabs, With Focus on Xiphosurida

Horseshoe crabs are an iconic group of extant chelicerates, with a stunning fossil record that extends to at least the Lower Ordovician (~480 million years ago). As such, the group has retained significant biological and palaeontological interest. The sporadic nature of descriptive and systematic research into fossil horseshoe crabs over the last two centuries has spread information on the group across more than 200 texts dating from the early nineteenth century to the present day. We present the most comprehensive pictorial atlas of horseshoe crabs to date to pool these important data together. This review highlights taxa such as Bellinurus lacoei and Limulus priscus that have never been documented with photography. Furthermore, key morphological features of the true horseshoe crab (Xiphosurida) families—Austrolimulidae, Belinuridae, Limulidae, Paleolimulidae, and Rolfeiidae—are described. The evolutionary history of horseshoe crabs is reviewed and the current issues facing any possible biogeographic work are presented. Four major future directions that should be adopted by horseshoe crab researchers are outlined. We conclude that this review provides the basis for innovative geographic and geometric morphometric studies needed to uncover facets of horseshoe crab evolution.


INTRODUCTION
Chelicerates, a group that includes arachnids (spiders, scorpions), eurypterids (sea scorpions), and Xiphosura (the so-called horseshoe crabs) have a stunning and extensive fossil spanning the early Palaeozoic to today and an exceptional modern diversity (Dunlop, 2010). Of these taxa, extant horseshoe crabs have been subject to detailed anatomical (van Der Hoeven, 1838;Owen, 1872;Lankester, 1881;Shuster, 1982;Shultz, 2001;Bicknell et al., 2018b,c,d), biochemical (Kaplan et al., 1977;Botton and Ropes, 1987), physiological (Sokoloff, 1978), morphological (Lee and Morton, 2005;Chatterji and Pati, 2014;Jawahir et al., 2017), and population dynamic (Botton, 1984;Brockmann, 1990;Gerhart, 2007) studies over the past two centuries. Furthermore, the impressive fossil record of this group, and apparent morphological conservatism that allowed survival of all five big mass extinctions, have driven extensive palaeontological interest in the group Rudkin and Young, 2009;Sekiguchi and Shuster, 2009;Krzeminski et al., 2010;Briggs et al., 2012;Dunlop et al., 2012;Lamsdell, 2013;Błazejowski, 2015;Lamsdell and Mckenzie, 2015;Bicknell et al., 2018bBicknell, 2019; Figure 1). Despite this extensive FIGURE 1 | The geological and morphological history of horseshoe crabs across the Phanerozoic. Number of named species is presented as well as suggested palaeoenvironment (Tables 1-7). A major transition to freshwater conditions occurred between the Devonian and Carboniferous. This was concurrent with a decrease in synziphosurine taxa and an increase in xiphosurids. Limulids had a diversification event in the Triassic and there was a transition back to dominantly marine conditions in the Jurassic. Dashed lines represent ghost lineages.
research, numerous avenues for further research remain for horseshoe crabs, and we highlight three here. Firstly, the evolutionary relationship between synziphosurines (the so-called "Synziphosura") and Xiphosura (Lamsdell, 2013(Lamsdell, , 2016Legg et al., 2013;Garwood and Dunlop, 2014). To help clarify this relationship, Lamsdell (2013) removed synziphosurines from Xiphosura and arrayed them within Prosomapoda and Planaterga. Secondly, there are a number of specimens that have been described in open terminology (Haug et al., 2012;Lamsdell et al., 2020) and despite the recent effort to bring taxa into recognized families, and genera, and erect new groups where appropriate (Bicknell, 2019;Bicknell et al., 2019e;Lamsdell et al., 2020), there remain an array of individuals that require taxonomic revision. Lastly, some genera appear to have been extensively over-split (Dunbar, 1923;Størmer, 1972;Fisher, 1984;Anderson, 1994;Haug et al., 2012;Kin and Błazejowski, 2014;Haug and Rötzer, 2018b). We therefore present a pictorial review of horseshoe crabs to aid current and future researchers in (1) the morphology and re-evaluation of taxa, (2) the determination of evolutionary relationships, and (3) the confirmation of species validity (Waterston, 1985;Selden and Siveter, 1987).
The palaeontological and evolutionary histories, broad taxonomy of families (Størmer, 1955;Novozhilov, 1991), and phylogenetic relationships (Lamsdell, 2013(Lamsdell, , 2016 of horseshoe crabs has often been reviewed (Bergström, 1975;Selden and Siveter, 1987;Anderson and Selden, 1997;Anderson and Shuster, 2003;Rudkin and Young, 2009). However, a document illustrating all horseshoe crab taxa has not been presented since Woodward (1866Woodward ( , 1867Woodward ( , 1879, Pringle (1929, 1930), Eller (1938b), and Raymond (1944). We have therefore collated images of all species considered horseshoe crabs (see taxa Dunlop et al., 2019), in a vital step toward understanding the true diversity and extent of Xiphosura (Lamsdell, 2013). We also present taxonomic descriptions of the facets that define members of xiphosurid families and consider of lifestyle and diversity of each group. We have focused on Xiphosurida as there are more taxa in this group than stem xiphosurids and synziphosurines. Nonetheless, synziphosurines and nonxiphosurid xiphosurans (previously considered Kasibelinuridae) are also briefly considered. It is vital to note that a thorough taxonomic revision of all species is beyond the intended scope of this review-namely the depiction and discussion of major horseshoe crab groups-but the images and details here represent the basis for such future work. The ultimate goal of this work is to depict all taxa in an open-access environment for future researchers to use as a reference point to continue research into this somewhat enigmatic group of chelicerates.

TERMINOLOGY
The following definitions are provided to clarify terminology used in descriptions. See Figure 2 for a depiction of these features. Somite: Fundamental unit or division that construct arthropod bodies (Lamsdell, 2013;Dunlop and Lamsdell, 2017).
Ophthalmic ridge: Ridge above the lateral compound eye that extends anteriorly and posteriorly relative to the compound eye (Størmer, 1955).
Thoracetron: Posterior section of Xiphosurida that is a fused solid plate. Shultz (2001) also suggested the termed tergum for this feature. The section may have expressed tergites.

Non-xiphosurid Xiphosura
First appearing in at least the Upper Ordovician of Canada and potentially the Lower Ordovician of Morocco the group contains taxa that have been considered stem-xiphosurids (Tables 1, 4, Figures 10-12; Rudkin and Young, 2009). There are eight genera and 10 species in this group. Two genera-Maldybulakia and Willwerathia-lack a family and the remaining six genera are considered stem-xiphosurids (formerly Kasibelinuridae, although this family was considered unhelpful by Bicknell et al., 2019c as it is a paraphyletic group). Non-xiphosurid xiphosurans are defined as chelicerates with a cardiac lobe extending to the anterior prosomal shield (Lamsdell, 2013). Species of this group  Rudkin et al., 2008;Rudkin and Young, 2009;Dunlop, 2010;Young et al., 2013; Figure 10B "Belinurus" alleghenyensis Eller, 1938b Stem xiphosurid Chadakoin Formation, New York State, USA

Xiphosurida
True horseshoe crabs are an extant order that first appeared in the Devonian (Figure 1). Key characteristics of true horseshoe crabs are a large, keeled, crescentic cephalothorax with anteriorly located lateral compound eyes, a thoracetron of fused tergites containing one or two sections, and a styliform telson (Anderson and Selden, 1997;Rudkin et al., 2008;Briggs et al., 2012;Lamsdell, 2016

Limulina
This sub-order comprises the superfamily Limuloidea, the families Paleolimulidae and Rolfeiidae, and the genus Bellinuroopsis. Limulina has a fossil record ranging from the Devonian to Recent. The diagnostic feature that separates Limuloidea from Belinurina is the fusion of the two most posterior thoracetronic tergites (sensu Lamsdell, 2016).

Paleolimulus kunguricus
Naugolnykh, 2017 Paleolimulidae Philippovian Formation, Russia Permian Marine Naugolnykh, 2017, 2018 Figure 22G ?Paleolimulus Bellinuroopsis rossicus Chernyshev, 1933 Unspecified Lebedjan Formation, Russia Devonian Marine Chernyshev, 1933;Eller, 1938b;Størmer, 1955;Novozhilov, 1991 Figure 23A Paleolimulid species were mostly marine taxa and their morphologies, similar to extant horseshoe crabs, reflect this life mode. They may have therefore variably explored swimming and burrowing life modes, with these ecological inferences related to the presence of movable thoracic spines . Paleolimulus woodae lacked thoracetronic movable spines and may have been capable of swimming, while Xaniopyramis linseyi, adorned with large thoracetronic spines, would have likely burrowed Lerner et al., 2016). The diversity of Paleolimulidae has previously been overstated and Paleolimulus is now considered a paraphyletic group (Lamsdell, 2016;Lerner et al., 2017;Bicknell, 2019). Many paleolimulid forms are now considered to be austrolimulids (discussed below), so continued research into these taxa is needed to uncover the true disparity of forms within this family and diversity of both austrolimulids and paleolimulids (Bicknell, 2019).

Rolfeiidae
This monospecific family consists of Rolfeia fouldenensis and is known from the Carboniferous-aged Cementstones Group, Scotland (Table 6, Figure 23). The cephalothorax is domed, exhibiting small genal spines, and a thick cephalothoracic margin. The species has a cardiac lobe narrows anteriorly and ophthalmic ridges that cross the lateral compound eyes, converging at the cardiac lobe (Waterston, 1985). The thoracetron is fused with visible tergal divisions and the opercular tergite is fully expressed. Large fixed and small moveable thoracetronic spines are known from R. fouldenensis (Waterston, 1985;Selden and Siveter, 1987;Lamsdell, 2016) and the telson is styliform. Lamsdell (2016) suggested that transverse cephalothoracic ridge nodes were characteristic of the family; however, as the holotype considered here lack these features, this feature may be treated tentatively. Presently, no appendages are known from this group (Waterston, 1985).
Limulids are thought to represent bradytelic evolution and exhibit strong morphological conservation between extant and fossil taxa. As such, they have been the focus of evolutionary and morphological research (Fisher, 1984;Bicknell and Pates, 2019b;. The limited morphological difference between the 148 Mya Jurassic "Limulus" darwini (Kcynia Formation, Poland) and modern juvenile L. polyphemus has been used to assert stabilomorphism; the "relative morphological stability of organisms in time and spatial distribution, the taxonomic status of which does not exceed genus level" (Błazejowski, 2015, p. 11). The conservation may reflect habitation of similar marine conditions, or convergence on an effective morphology.
Extant limulids have distributions across the east coast of the USA and Asia, with their common names reflecting said distribution (Shuster, 2001;Bicknell and Pates, 2019a): the American, or Atlantic, horseshoe crab, Limulus polyphemus; the Indonesian horseshoe crab, Carcinoscorpius rotundicauda; the Chinese horseshoe crab, Tachypleus gigas; and the Japanese horseshoe crab, T. tridentatus (Figures 35, 36; Itow et al., 2003;Zhou and Morton, 2004;Sekiguchi and Shuster, 2009). The ontogeny and morphology of these taxa has been documented thoroughly across the past two centuries (Shuster, 1982;Haug and Rötzer, 2018a) and the morphological similarities are depicted in Figures 35 and 36. Extant limulids occupy many environmental conditions and can exploit brackish, freshwater, shallow water, and fully-marine conditions . Limulus polyphemus, T. gigas, and T. tridentatus are mostly shallow marine, bottom-dwelling taxa that spawn on beaches and inhabit a combination of marine sub-habits during ontogeny (Fisher, 1984). Conversely, C. rotundicauda migrates into completely freshwater (Størmer, 1952;Fisher, 1984;Crônier and Courville, 2005;Sekiguchi and Shuster, 2009;Lamsdell, 2016). Despite representing the descendants of a long fossil lineage, they now face an extinction event. Extensive harvesting of specimens for their blood, and as a food source, as well as habitat modification have majorly impacted populations (Botton, 2001;Hsieh and Chen, 2009;Shin et al., 2009;Akbar John et al., 2011;Cartwright-Taylor et al., 2011;Carmichael and Brush, 2012;Nelson et al., 2015;Kwan et al., 2016;Fairuz-Fozi et al., 2018). Measures therefore need to be taken to prevent this group from an extinction event. To this end, L. polyphemus and its kin have now been suggested as world heritage species (Tanacredi et al., 2009) and T. tridentatus was recently listed as an endangered taxon (Laurie et al., 2019) Frontiers in Earth Science | www.frontiersin.org showing that progress is being made in preventing the humandriven extinction of Xiphosurida.

Valloisella
This monospecific genus from the Carboniferous Coal Measures in England and France (Figure 36) was originally considered a belinurid (Anderson and Horrocks, 1995) but has since been placed at the base of Limuloidea by recent phylogenetic analyses (Lamsdell, 2016). The genus is defined by an almond-shaped cephalothorax, genal spines that extend almost to the thoracetron terminus, and a flange located along the thoracetronic margin (Anderson and Horrocks, 1995). The fused thoracetron is trapezoidal with expressed tergal divisions, contrasting most other species in Limuloidea. No appendages are known from this genus.

HORSESHOE CRAB EVOLUTIONARY HISTORY AND DIVERSITY
Horseshoe crabs experienced three major evolutionary events across the Phanerozoic (Figure 1). The Palaeozoic horseshoe crab record was the most exploratory anatomically and evolutionarily (Błażejowski et al., 2017). The rise of synziphosurines began in the Lower Ordovician (Rudkin et al., 2008;Rudkin and Young, 2009;Dunlop, 2010;Van Roy et al., 2010, 2015. Across the Silurian and Devonian, the marine and marginal marine forms were abundant and represent the first evolutionary radiation of this group, before the diversification of Xiphosurida (Størmer, 1955). Synziphosurine diversity declined heavily, reducing to one taxon in the Carboniferous, when they subsequently went extinct (Selden and Drygant, 1987;Selden and Siveter, 1987;Babcock et al., 1995;Anderson and Selden, 1997;Moore et al., 2007;Lucas et al., 2014). Non-xiphosurid xiphosurans also arose in the Upper Ordovician, potentially even the Lower Ordovician, and are unknown after the Devonian (Bicknell et al., 2019c). Xiphosurida arose in the late Devonian with Bellinuroopsis (Moore et al., 2007). After this, at least four xiphosurid families arose in the Carboniferous: the Belinuridae, Limulidae, Paleolimulidae and Rolfeiidae (Selden and Drygant, 1987;Selden and Siveter, 1987;Babcock et al., 1995;Anderson and Selden, 1997;Lucas et al., 2014;Bicknell, 2019;Bicknell and Pates, 2019b;Bicknell et al., 2019e), with evidence suggesting that Austrolimulidae may also have arisen at this time (Lamsdell, 2016). Carboniferous Coal Measures and Konservat-Lagerstätten record the highest specific diversity and first radiation of Xiphosurida (Anderson, 1997;Moore et al., 2007;Rudkin and Young, 2009). Exploitation of brackish and freshwater conditions by the late Palaeozoic Xiphosurida may reflect adaptation to inconsistent coastlines and fluctuating shallow-marine conditions (Błażejowski et al., 2017). Xiphosurid diversity apparently decreased drastically during the Permian, reflecting the closure of exceptional preservation windows and an increase in xiphosurids inhabiting marginal environments that are poorly preserved in the geological record (Rudkin and Young, 2009). At the end of the Carboniferous, there is no further record of Rolfeiidae, while the first definite austrolimulid species arose in the Permian (Bicknell, 2019). The Permian-Triassic "Great Dying" drove belinurids and paleolimulids to extinction, while austrolimulids and limulids survived into the Mesozoic (Bicknell and Pates, 2019b). The Triassic was a period of extensive exploration in morphology and the second radiation of xiphosurids and the third evolutionary pulse in horseshoe crabs (Bicknell and Pates, 2019b;Bicknell et al., 2019e). An aspect of this radiation was size increase: Mesozoic taxa were much larger (30-60 cm long, including telson) than the Palaeozoic counterparts (3-5 cm) (Størmer, 1955;Bicknell and Pates, 2019b). Austrolimulid diversity peaked in the Triassic (Figure 1) but then decreased into the Cretaceous, during which time the group went extinct. Limulid diversity peaked in the Triassic with 12 species and decreased to five during the Cretaceous (Bicknell et al., 2019e). Only limulids survived into the Tertiary with one named Cenozoic species: the Eocene Limulus decheni (Rudkin and Young, 2009;Schimpf et al., 2017), a suggested "missing link" between extant Asian and American taxa . This evolutionary history is one of generally low generic diversity, such as in the four extant species (Anderson and Selden, 1997;Anderson, 1999;Sekiguchi and Shuster, 2009;Dunlop et al., 2012). However, the habitation of marginal environments with poor conditions for exceptional preservation of un-biomineralised exoskeleton cuticle also may have impacted this observed low diversity (Babcock, 1998;Anderson, 1999;Lamsdell and Mckenzie, 2015).

GEOGRAPHICAL DISTRIBUTION OF XIPHOSURAN MATERIAL
Distribution of horseshoe crab fossils is uneven in space and time; reflecting historical biases in collecting that favored North America and Western Europe. The UK has the highest number of taxa (n = 35), followed by the USA (n = 23) and Germany (n = 22). Other areas with much larger landmasses have far fewer known taxa: South America (n = 1), Australia (n = 7), Asia (n = 5), and Africa (n = 6). This uneven geographical sampling also partly reflects uneven temporal sampling (e.g., 25 UK taxa are Carboniferous, and eight are Silurian and 11 of 22 German taxa are Triassic). Within countries, well-explored horizons or formations also provide apparent diversity peaks. Notably the South Wales Coal Measures formations (South   Fisher (1977) and Anderson (1994) and discussed in Haug et al. (2012) Wales, UK) where six belinurids are known from the South Wales Lower Coal Measures Formation and three belinurids from the South Wales Upper Coal Measures Formation. These nine taxa, within a limited geographic and temporal sample, provide an apparently high Carboniferous diversity skewing the understanding of overall belinurid diversity and geographic spread as well as reflecting an over-splitting of the group. To address these sampling issues (which are by no means limited to horseshoe crabs) further exploration needs to be targeted to under-sampled regions (Africa, Asia, South America) and time periods (Jurassic and younger), as well as reassessing the apparent high diversity of taxa that have not been recently studied. Such efforts, combined with a concerted effort to redescribe and refine horseshoe crab taxonomy will allow ranges of different groups to be compared without the current underlying biases.

FUTURE DIRECTIONS
Horseshoe crabs are an iconic group of chelicerates and, as depicted here, have been thoroughly, if somewhat sporadically, scientifically explored over the past two centuries. However, in conducting this review we highlighted four main research areas that should be addressed. To conclude this review, potential future directions for horseshoe crab research are presented.
(1) Bicknell (2019), andBicknell et al. (2019e) highlighted that the traditional views that horseshoe crabs represent evolutionary conservatism, stasis, and bradytelic evolution (Fisher, 1984;Selden and Siveter, 1987;Rudkin et al., 2008) is overstated. In reality, the group experienced three major changes across the Phanerozoic: increased size, thoracetronic fusion, and restriction to marine habitats (Størmer, 1955;Crônier and Courville, 2005;Bicknell and Pates, 2019b). Lamsdell (2016) thoroughly explored the record of habitat change, but the remaining two points should be considered. Thoracetronic fusion has been attributed to a change in ecology, from enrolment to burrowing, but this remains fairly unexplored (Fisher, 1977(Fisher, , 1981(Fisher, , 1982Waterston, 1985;Lamsdell, 2016;Błażejowski et al., 2017). A study considering when complete fusion developed in the context of palaeoenvironmental and palaeoecological conditions may confirm this hypothesis. Size change is likely associated with exploitation of different niches: smaller Xiphosurida likely preferred freshwater conditions, reflected today in the smallest taxon-Carcinoscorpius rotundicauda (Hauschke and Wilde, 1991;Dunlop et al., 2012). A study considering shape and size change through time would allow this hypothesis to be tested. In addition, modern descriptive and statistical tools, such as multivariate geometric morphometrics, semilandmark, and landmark analyses could be employed to explore this topic in more detail (Bicknell, 2019;Bicknell and Pates, 2019b;Bicknell et al., 2019e).
(2) Rates of morphometric change in horseshoe crabs have not been thoroughly explored (Fisher, 1984). The same morphometric data outlined above could be used to address possible evolutionary rates and quantify whether the group, especially limulids, represent arrested evolution. Time series analyses can also be conducted with these data to study modes and models of evolution (Hunt and Carrano, 2010;Hunt et al., 2015;Bicknell et al., 2018a). (3) As Tables 6 and 7 outline there are many specimens have been identified as xiphosurids but not formally (re)described in light of recent progress in the field (Lamsdell et al., 2020). Formally describing these specimens would thoroughly aid understanding patterns of horseshoe crab diversity through time. Similarly, new collecting efforts should be focussed on under-represented parts of the globe such as Asia, Africa and South America, as well as Jurassic and younger deposits, where knowledge of this group is hindered by a lack of specimens. (4) Computer tomography (CT) scanning to document fossil and extant species has become a major tool over the past decade, which has started to positively impact horseshoe crab research. Schimpf et al. (2017) CT scanned Limulus decheni specimens to accelerate digital transfer of important morphological information (Figure 35). Zuber et al. (2017) used CT scans and augmented laminography to document muscle detail in a Limulitella sp. specimen (Figure 24), and Bicknell et al. (2018b) conducted micro-CT scans of iodine stained appendages to show L. polyphemus muscles in situ. Scanning and 3D reconstructions of specimens are still developing and therefore ripe for research, especially for documenting and disseminating information on holotypes.

CONCLUSIONS
The atlas presented here is the first comprehensive collation of named taxa and other unnamed specimens considered horseshoe    crabs. The work builds on research presented during the early-to middle-twentieth century and, its presentation in an open-access environment will allow all researchers interested in horseshoe crabs access to key anatomical information needed for new taxonomic studies. Brief notes detailing the characteristic features and supposed life modes of families within Xiphosurida are presented, synthesizing other key works on the group. A brief evolutionary history of horseshoe crabs is presented, which outlines diversity changes from the Lower Ordovician to today. Finally, we highlight four major avenues for future research: most notably analyses of morphometric data of horseshoe crabs to mathematically probe the evolutionary history of the group. These same data may represent an important step toward reconciling synziphosurines with true horseshoe crabs. (E) Allan Lerner.
Frontiers in Earth Science | www.frontiersin.org Frontiers in Earth Science | www.frontiersin.org

AUTHOR CONTRIBUTIONS
RB designed the study and made the figures, with input from SP. RB and SP photographed material and wrote the manuscript.