A revision of Prolimulus woodwardi Fritsch, 1899 with comparison to other highly paedomorphic belinurids

Xiphosurida is an ingroup of marine Euchelicerata often referred to as “living fossils”. However, this oxymoronic term is inapplicable for Paleozoic and early Mesozoic forms, as during these periods the group experienced notable evolutionary radiations; particularly the diverse late Palaeozoic clade Belinurina. Despite the iconic nature of the group, select species in this clade have been left undescribed in the light of recent geometric morphometric and phylogenetic considerations and methodologies. To this end, we re-describe Prolimulus woodwardi Fritsch, 1899 using new and type specimens to reveal more details on appendage anatomy and possible ecology. Furthermore, we present geometric morphometric and phylogenetic analyses that uncover relationships between P. woodwardi and other belinurids without genal spines. Both approaches highlight that a clade containing Prolimulus Fritsch, 1899, Liomesaspis Raymond, 1944, Alanops Racheboeuf, Vannier & Anderson, 2002 and Stilpnocephalus Selden, Simonetto & Marsiglio, 2019 may exist. While we do not erect a new group to contain these genera, we note that these genera exemplify the extreme limits of the Belinurina radiation and a peak in horseshoe crab diversity and disparity. This evidence also illustrates how changes in heterochronic timing are a key evolutionary phenomenon that can drive radiations among animals.

In the Plzeň Basin, the lower parts of the Nýřany Member contain the locally developed Main Nýřany Coal with intercalated beds of lacustrine sapropelic coal  Opluštil et al., 2016). Arrow indicates the stratigraphic location of the studied material. Abbreviations: Fm., Formation; Mb., Member; S. Fm., Slaný Formation; T. Fm., Týnec Formation.

Specimen photography
Museums where Prolimulus woodwardi specimens are housed were contacted and photographs of specimens were either requested from the collection managers or made by the authors, or colleagues. Most specimens were photographed with SLR cameras under normal light. Select specimens were submerged in alcohol prior to photography to enhance contrast; however, this could not be conducted for all specimens due to collection constraints.

Phylogenetic analyses
The phylogenetic analysis was conducted to determine where Prolimulus woodwardi and the morphologically comparable Stilpnocephalus pontebbanus are located in tree space. These species were coded into the Bicknell, Lustri & Brougham (2019) matrix, derived from Lamsdell (2016). The analysis was performed under equally weighted parsimony in TNT 1.5 (Goloboff & Catalano, 2016) following Bicknell, Lustri & Brougham (2019) and Lamsdell (2016). Further, implied and equal weighted produced highly comparable trees. Five replications of a ''New Technology'' tree search was run using random sectorial searches, 1,000 iterations of the parsimony ratchet, 50 cycles of drifting and 5 rounds of tree fusing, holding a maximum of 10 trees per replication (Supplementary Information 1). All multistate characters were unordered (Lamsdell, 2016;Bicknell, Lustri & Brougham, 2019).

Geometric morphometric methods
Following Bicknell et al. (2019), a morphometric dataset of landmarks and semilandmarks from 91 specimens across 19 species was collected to explore Belinurina morphospace. Landmarking and semilandmarking was conducted using the Thin-Plate Spline (TPS) suite (http://life.bio.sunysb.edu/morph/index.html). The TPS file was constructed using tpsUtil64 (v.1.7). The TPS file was imported into tpsDig2 (v.2.26), which was used to place four landmarks across the prosoma and thoracetron and 40 semi-landmarks along the right prosomal shield ( Fig. 2; Table 1). Semilandmarks were placed in a clockwise direction along the most anterior section of the prosomal shield, coinciding with the first landmark, ending at the third landmark: the most lateral prosomal-thoracetronic articulation point. Points were digitised as xy coordinates. When the right side was poorly preserved, the left side was used, and these data were mirrored. These data populated the TPS file (Supplementary Information 2). TPS file was imported into R. The 'geomorph' package (Adams & Otárola-Castillo, 2013) was used to conduct a Procrustes Superimposition and Principal Components Analysis (PCA) of the data (Data S3). Only the first two Principal Components (PCs) were considered as they explained 87% of the variation in the data. The examined species were representatives of Alanops, Belinurus, Euproops, Liomesaspis, and Prolimulus. We were unable to include Anacontium and Stilpnocephalus as opisthosomal sections are not known from these genera. We had initially used generic assignment of Bicknell & Pates (2020) for this analysis. However, during the course of peer review, Lamsdell (2020b) presented a revision of Xiphosurida and proposed that Belinurina consisted of 14 genera. To compare, contrast, and explore the distribution of these newly erected groups with the more conservative perspective of Bicknell & Pates (2020), we presented the distribution of genera suggested in Bicknell & Pates (2020) and Lamsdell (2020b). Amended diagnosis: Belinurid with a round prosoma that is slightly wider than long.
No eyes, cardiac lobe, or ophthalmic ridges are present. Thoracetron is completely fused, without traces of segmentation, often showing a thoracetronic doublure. Thoracetrontelson articulation is 'U'-shaped. Telson is keeled.

Remarks:
The prosoma and thoracetron shape of Prolimulus woodwardi is morphologically comparable to other belinurids without genal spines (Figs. 12 and 13). However, P. woodwardi has a unique telson insertion morphology. In P. woodwardi the insertion is a 'U'-shaped indentation in the thoracetron, while Alanops and Liomesaspis lack this feature (Fig. 14). Furthermore, Alanops and Liomesaspis (Fig. 13) possess a 'thoracetronic boss', a bulge present on thoracetron over the insert of the telson. In P. woodwardi there is no evidence of this morphology. Finally, Stilpnocephalus has two notable grooves along the prosoma, not observed in P. woodwardi.

Phylogenetic results
The phylogenetic analysis produced 3 trees of length 746. The strict consensus tree resultant from these trees have a comparable topology to other publications that have used the same matrix (Lamsdell, 2016;Bicknell, Lustri & Brougham, 2019;Bicknell, Naugolnykh & Brougham, 2020;Bicknell & Pates, 2019;Fig. 15). The main difference is the grouping of species within Belinurina. Prolimulus woodwardi resolves within a polytomy with Liomesaspis birtwelli (Woodward, 1872) and Alanops magnificus Racheboeuf,Vannier & Figure 15 Results of the phylogenetic analysis. Strict consensus of the three trees produced by analyzing Supplementary Information 1. Prolimulus woodwardi is presented in bold. Topology of the outgroups, Synziphosura, Chasmataspidida, Eurypterida, and Arachnida collapsed as they are not considered here and are unchanged from other studies that used this dataset. Full-size DOI: 10.7717/peerj.10980/ fig-15 Anderson, 2002. Stilpnocephalus pontebbanus resolves in a polytomy containing L. laevis Raymond, 1944 andL. leonardensis Tasch, 1961 and the branch leading to L. birtwelli, A. magnificus and P. woodwardi. The autapomorphies that characterize this clade are the reduction or absence of the genal spine, a round thoracetron with limited to no expression of tergal boundaries, and the lack of movable or fixed thoracetronic spines.

Evolutionary framework of Prolimulus and kin
Belinurids represent the most successful Carboniferous and Permian xiphosurid group that explored freshwater niches (Lamsdell, 2016;Shpinev & Vasilenko, 2018;Shpinev, 2018;Bicknell, Pates & Botton, 2019). The group also has an exceptional diversity and disparity, which is unusual when compared to the Late Mesozoic and Cenozoic forms (Bicknell, 2019;Bicknell et al., 2019a;Bicknell et al., in press-a). Several attempts to colonize freshwater environments likely drove the Belinurina to evolve features that contrast the 'typical' xiphosurid morphology, and added to their extreme diversity and disparity (Lamsdell, 2016;Lamsdell, 2020a). Furthermore, freshwater environments can only sustain small populations (Wang et al., 2019) compared to marine conditions and are susceptible to isolating small populations (see DeWoody & Avise, 2000). Allopatric speciation clearly played a central role in the belinurid radiation (Lamsdell, 2016;Lamsdell, 2020a) and permitted innovative characters to be fixed within newly established populations. The importance of heterochrony during xiphosurid evolution has recently been considered by coding such characters in a phylogenetic framework (Lamsdell, 2020a). This work demonstrated (among other points) that belinurid evolution generally reflects paedomorphosis. While Lamsdell (2020a) did not assess Prolimulus, paedomorphic evolution no doubt drove the development of species that are located in the same tree space in Fig. 15. The reduced body size, short or vestigial genal spines, rounded thoracetron, and absence of pre-telson epimera (terminal thoracetronic spines) are considered paedomorphic characters in Belinurina (Lamsdell, 2020a). The reduction or absence of genal spines and a prosoma:thoracetron ratio of ∼1:1 observed in Prolimulus, Alanops, and Liomesaspis are also comparable to Limulus polyphemus (Linnaeus, 1758) trilobite stages  (Fritsch, 1899;Prantl & Pribyl, 1955;Haug & Rötzer, 2018a). However, the fossil genera display fully developed telson spines, unknown to early postembryonic L. polyphemus stages (Haug & Rötzer, 2018a). As such, the lack of genal spines and prosoma:thoracetron ratio are likely phylogenetically significant anatomical similarities, and not aspects of ontogeny. This hypothesis is supported by the Racheboeuf, Vannier & Anderson (2002) dataset that illustrated that the main ontogenetic modification in A. magnificus is increased size. The presence of juvenile characters in adult individuals of Prolimulus and its kin therefore represent a heterochronic event (Gould, 1977;Klingenberg, 1998). Given this unique combination of characteristics, one might consider erecting a clade to house these notably paedomorphic species. Indeed, Raymond (1944) had erected Liomesaspidae to contain these forms; however, this group is not used anymore. Furthermore, given the convoluted relationships between members of Belinurina, it seems unwise to re-introduce terminology. When phylogenetic and taxonomic relationships within the Belinurina are organized, it may then be pertinent to re-erect a higher order group. Lamsdell (2020b, p. 17) suggested that Prolimulus ''strong[s] affinity to Alanops and Pringlia, and there could be an argument for synonymizing Prolimulus with one of these genera''. We disagree with this suggestion based on our observations here. The morphology of the Prolimulus thoracetron-telson articulation differs from Alanops and Pringlia (here considered synonymous with Liomesaspis, following the more conservative Anderson & Selden, 1997, and amount of overlap in morphospace;Figs. 13, 17). Furthermore, a 'thoracetronic boss' is not observed in Prolimulus and the telson likely inserted in the thoracetronic doublure, through a telson notch (Figs. 6, 12 and 14). Regardless, if any synonymy were valid, Alanops or Liomesaspis would be synonymized with Prolimulus (not vice versa) as the Czech material has taxonomic priority.

Ethology
Clusters of extinct arthropods in the fossil record were considered evidence of biological activities (such as gregarious behavior) or traces of digestive processes (bromalite), as opposed to taphonomic artifacts (Speyer & Brett, 1985;Karim & Westrop, 2002;Paterson et al., 2008;Brett et al., 2012;Brett, 2015;Bicknell, Pates & Botton, 2019). Specimens on the sample NM Me 108 (Fig. 7) may represent such gregarious behavior. The lacustrine nature of the sapropelic coal suggest minimal physical disturbance; the individuals were therefore likely not accumulated by currents or other physical factors. As the assemblage is monospecific and has a uniform size distribution, defensive behavior can also be excluded. Interpreting this assemblage as a bromalite is also less parsimonious as there is no evidence of digestion, nor does the cluster conform to the morphology of regurgitalites, coprolites, or cololites (Hunt, 1992). We therefore suggest that either a moulting or mating event best explains the cluster. Horseshoe crab clustering is well documented in extant species (Shuster Jr, 1982;Brockmann, 1990;Brockmann, 2003;Brockmann, Nguyen & Potts, 2000;Brockmann et al., 2015); however, exceptionally rare in the horseshoe crab fossil record. Indeed, the only evidence is one possible Euproops danae (Meek & Worthen, 1865) cluster (Ambrose & Romano, 1972;Fisher, 1979;Bicknell, Pates & Botton, 2019). NM Me 108 (Fig.  7) therefore illustrates that clustering was potentially more common than previously thought and was employed by multiple belinurid genera.

Epibiotic organism associated with Prolimulus
Adult extant xiphosurids often experience interactions with epibionts (Patil & Anil, 2000;Shuster Jr, Botton & Keinath, 2003), while immature individuals often lack evidence of epibiotic fauna (Allee, 1923;Shuster Jr, 1957). This difference reflects frequent moulting by younger individuals, an event that removes any communities attached to the exoskeleton (Shuster Jr & Sekiguchi, 2003). Conversely, moulting events decrease drastically when the animals reach the sexual maturity, such that adult horseshoe crabs may have as few as one moult per year (Carmichael, Rutecki & Valiela, 2003). This infrequency of moulting events allows ectocommensal organisms to colonize the dorsal exoskeleton of adult horseshoe crabs. The presence and distribution of epibionts in the fossil record could therefore be used to infer developmental stages in fossil xiphosurids. Possible parasitic interaction between Prolimulus woodwardi (host) and Spiroglyphus vorax (parasite, serpulid annelid, or microconchid Taylor & Vinn, 2006) has been suggested (Prantl & Pribyl, 1955). The abundance of S. vorax on the studied specimens (Figs. 4A, 4B, 4D; 6A, 6B) suggests that P. woodwardi individuals had reached the sexual maturity and the examined population therefore represented fully adult individuals. Such evidence adds to the growing record of potential epibiotic and parasitic relationships preserved within the fossil record (see Conway Morris, 1981;Huntley & DeBaets, 2015;Klompmaker & Boxshall, 2015;Leung, 2017;Zhang et al., 2020).
Comparing the distribution of genera Bicknell & Pates (2020) with Lamsdell (2020b) allows the taxonomic framework based on phylogenetic topology to be examined and scrutinized. The position of Macrobelinurus Lamsdell, 2020b andAndersoniella Lamsdell, 2020b specimens in morphospace separate from the other clusters strongly supports the validity of these genera. Conversely, the overlap of Parabelinuris Lamsdell, 2020b with Belinurus and Euproops suggests that Parabelinuris represents over-splitting of the traditional genera (sensu Bicknell & Pates, 2020). Finally, the large spread of Koenigiella and Prestwichianella across Belinurus and Euproops suggests that either these new genera have large morphological variation, or are congeneric with Belinurus and Euproops. This over-splitting may represent the unfortunate compartmentalization of ontogenetic stages as Belinurus and Euproops taxa are may record the same ontogenetic trajectory (Haug & Haug, 2020). Regardless, more specimens of all genera are required for this morphospace to be more completely understood and to test the phylogenetic hypotheses of Lamsdell (2020b). Furthermore, a thorough taxonomic revision of the group is needed; a work that should illustrate the range of genera, comparable to Bicknell et al. (in press). Finally, and most importantly, a novel phylogenetic matrix should be constructed in tandem with such a treatise to document independently the convoluted taxonomic record of Belinurina.

CONCLUSION
Revision of Prolimulus woodwardi, coupled with phylogenetic and geometric morphometric analyses of belinurids, highlighted a diverse clade within Belinurina. These species without genal spines all share highly accentuated paedomorphic characters, such as vestigial genal spines, and are representative of paedomorphic evolution. A slab of multiple P. woodwardi individuals demonstrates new evidence for Carboniferous horseshoe crab ecology, revealing possible gregarious behavior, and further data on the deep origin xiphosurid clustering. Taken together, the examination presented here demonstrates the morphological variation and ecological conditions that permitted successful colonization of freshwater environments by Carboniferous horseshoe crabs.