1. Introduction

Euarthropods, more specifically members of the clade Artiopoda, dominated diversity in Palaeozoic animal communities (Daley et al., 2018). This group, which includes trilobites and their non-biomineralizing relatives, are united by a morphology consisting of a headshield covering a pair of uniramous antennae and at least three pairs of biramous limbs, followed by a dorsoventrally flattened body with a series of biramous appendages (Stein and Selden, 2012; Giribet and Edgecombe, 2019). The origins of key morphological features facilitating the diversity of trilobites can be found in their non-biomineralizing artiopodan relatives (e.g. Du et al., 2019; Schmidt et al., 2022). The cephalon, as an informative and functionally specialized body region, affords pivotal data for resolving the phylogenetic relationships and evolutionary history of arthropods, artiopodans, (Budd and Telford, 2009; Yang et al., 2013; Strausfeld et al., 2022) and trilobites in particular (e.g. Stubblefield 1936; Lieberman 2002; Paterson et al. 2019).

The presence, shape, and character of sutures in the trilobite cephalon have been central to classification schemes of the group for nearly 200 years (e.g. Emmrich 1839; Salter 1864; Beecher 1897a, b; Stubblefield 1936; Rasetti 1945; Whittington et al. 1997), while the similarity in the cranidial outlines of Cambrian suture-bearing groups supported a single origin or strong evolutionary constraint on its origins (Foote 1991; Hughes 2007). These sutures provided lines of weakness along which the cephalon split during moulting, creating an anterior ecdysial gape through which the animal exited the old exoskeleton (e.g. Henningsmoen 1975; Whittington 1990; Daley and Drage, 2016). Sutures may have served additional purposes (Stubblefield 1936) and were under additional functional demands, including feeding (Fortey & Owens 1999) and burrowing behaviours (Esteve et al. 2021). Different trilobite clades display distinct suture patterns, with two sets, the circumocular and marginal sutures, relevant for the creation of this ecdysial gape (Henningsmoen 1975). For the earliest diverging trilobites, olenellines (Palmer & Repina 1993; Lieberman 2002) which a have been recovered as a paraphyletic grade rather than a monophyletic group (e.g. Paterson et al. 2019) circumocular sutures facilitated shedding of the cornea during ecdysis (e.g. Ramskold & Edgecombe 1991) while a marginal suture (Fig. 1e) facilitated anterior egression through separation of the lateral cephalic doublure from the dorsal cephalon (e.g. Stubblefield 1936; Henningsmoen 1975). In later diverging trilobites the circumocular and marginal sutures combined into a single system (Fig. 1d). Here the cornea is fused to the free cheek (e.g. Stubblefield 1936; Ramsköld & Edgecombe 1991) and these facial sutures – the fused circumocular and marginal sutures - facilitated both ecdysis of the visual surface and anterior egression of the exoskeleton following withdrawal from the free cheeks (e.g. Henningsmoen 1975).

Fig. 1

The origin of these fused dorsal ecdysial sutures – facial sutures – has traditionally been considered to fall within Trilobita (e.g. Fortey and Whittington, 1989; Edgecombe and Ramsköld, 1999; Lieberman, 2002). However, the presence of dorsal ecdysial sutures in earlier diverging artiopodans – the Protosutura (Du et al. 2019) and eye slits in Petalopleura (Chen et al., 2019) raised questions of the homology and origins within Artiopoda (e.g. Hou et al. 2017; Du et al. 2019). An alternative hypothesis for the origins of the dorsal cephalic sutures emerged: that these had a deep root within Artiopoda and were subsequently lost in some groups including multiple times within Trilobita (Hou et al., 2017a; Du et al., 2019), rather than the traditional view that dorsal cephalic sutures in trilobites were derived within the clade, and thus these eye slits and facial sutures were acquired independently. Support for the deep root hypothesis comes from the variability of facial sutures in trilobites, and the recognition that convergence and loss of features is common when they have a function or allow adapation to a particular niche (e.g. Moore & Wilmer 1997). Further evidence for the variability of suture morphology comes from ontogenetic studies, such as the fusion of dalmanitinid facial sutures during ontogeny (Drage et al., 2018), and the loss of this feature in other trilobite such as ‘Cedaria’ woosteri (Hughes et al. 1997). This indicates that there is scope for a facial suture to have been lost repeatedly in artiopodan groups earlier diverging than Trilobita, should this feature prove to have a deeper origin within Artiopoda (Hou et al. 2017; Du et al. 2019). A deep root for facial sutures within Artiopoda would have repercussions for the importance of the facial sutures in determining trilobite relationships (e.g. Jell 2003) and the position of a grade of olenelline trilobites as the earliest diverging members of the group (e.g. Paterson et al. 2019). To date, a third possibility – that the ventral plates of Acanthomeridion are homologous to the doublure of olenellids, and thus Acanthomeridion and olenelline trilobites share presence of an unfused marginal suture – has not received broad attention in the literature, but should also be considered in light of the similarities in position of the ventral plates and olenelline doublure, and the marginal rather than dorsal position of the suture in Acanthomeridion..

Resolving the complex history of dorsal ecdysial sutures evolution within Artiopoda, requires the nature of cephalic sutures in artiopodans such as Acanthomeridion, protosuturans, and petalopleurans to be resolved, and their phylogenetic position confidently determined (Hou et al., 2017a; Du et al., 2019). However, the morphology of many earlier diverging artiopodans is not completely understood (Ortega Hernández et al., 2013; Lerosey-Aubril et al., 2017), contributing a significant amounts of missing data to character matrices used in phylogenetic analyses. Up to now, Acanthomeridion was arguably the least well known of these early diverging artiopodans, with its ventral anatomy and non-biomineralized structures very poorly known (Hou and Bergström, 1997; Hou et al., 2017a).

Here we describe new specimens of Acanthomeridion serratum, using micro-CT to reveal details of the ventral anatomy, eyes, and appendages for the first time, and reconstruct the changing shape of the body during ontogeny. We use these new data to show that A. anacanthus is a junior synonym of A. serratum. We then assess the phylogenetic position of the species considering two hypotheses relating to the dorsal ecdysial sutures – as homologous to the dorsal ecdysial sutures (facial sutures combined of a fused marginal and circumocular sutures) of trilobites (Fig. 1d), that they are homologous to marginal sutures of olenellines (Fig. 1e), and that these structures are unique to Acanthomeridion (Fig. 1f), to evaluate the different hypothesis of origination within Artiopoda.

2. Materials and methods

New specimens of Acanthomeridion were collected from Jiucun town, Chengjiang (Fig. 2) and accessioned at the Management Committee of the Chengjiang Fossil Site World Heritage (CJHMD 00052– 00062), and Research Center of Paleobiology, Yuxi Normal University (YRCP 0016–0020). New specimens of Wutingaspis tingi (YRCP 0021) and Xandarella spectaculum (YRCP 0022) were collected from Haikou town, Kunming (Fig. 2) and housed at Research Center of Paleobiology, Yuxi Normal University. The holotype of Zhiwenia coronata (YKLP 12370) is deposited at the Key Laboratory for Palaeobiology, Yunnan University which was excavated from Xiaoshiba area, Kunming (Fig. 2). The specimens were photographed by a LEICA DFC 550 digital camera mounted to a Stereoscope LEICA M205 C, and scanned with a Zeiss Xradia 520 Versa X-ray Microscope and Nikon XTH 225 ST micro-focus X-ray tomography machine. All micro-CT images were processed with the software Drishti (Zhai et al., 2019). Figures were processed with CorelDRAW X8, Inkscape 1.0 and Illustrator.

Fig. 2

Given that specimens with thorax longitudinally curved and specimens with flattened thorax are considered as two species of Acanthomeridion, despite having the same exoskeletal morphology, we measured their axial length and attempted to determined their true maximum width in all specimens, including those described in this study and previously documented measurable specimens. The data was initially analyzed using a bivariate regression fitting function in Microsoft Office Excel. Subsequently, the software Past (Hammer et al. 2001) was employed for testing and validation, obtaining the same function and conducting a rational analysis of the function.

The character matrix used for the phylogenetic analyses was adapted from Schmidt et al., (2022), itself adapted from Du et al. (2019) and Chen et al. (2020). Olenellus getzi was added, to facilitate comparison with olenelline trilobites, while Anacheirurus adserai, Cryptolithus tesselatus and Olenoides serratus were added to provide more trilobite terminals. Olenellus getzi was chosen as it represents the olenelline with best known soft tissues (antennae only; Dunbar 1925). Trunk appendages are known in A. adserai, C. tesselatus and O. serratus (Walcott, 1911; Whittington, 1975), with descriptions by Bicknell et al., 2021; Losso and Ortega-Hernández 2022 for coding the matrix. Our updated dataset includes 70 taxa (Acanthomeridion anacanthus was removed, four trilobites were added). Three matrices were used to test the alternative hypotheses of sutures, one with 100 characters considering the ventral plates unique, and not homologous to trilobite librigenae (‘ventral plates under head’ as a new character in the dataset) and two with 99 characters. The first of these considered the ventral plates homologous to trilobite librigenae, and the second to the doublure of olenellines (see Morphobank project). We added the character state ‘ringed distribution lamellae’ to Ch. 15 in all matrices (Chen et al., 2019; Du et al., 2019), and altered the characters relating to the suture pattern (Ch. 31, 32) to include character states suitable for Olenellus getzi. Given the difficulties of identifying circumocular sutures in non-biomineralizing artiopodans, Ch. 31 indicated the presence or absence of ecdysial sutures in the cephalon, while Ch. 32 indicated the type of suture. This could either be marginal and/or circumocular, but unfused (state 0) or marginal and circumocular fused into a facial suture (state 1). This facilitated differentiation between the state in Olenellus getzi and the other trilobites in the dataset, and thus allowed the two hypotheses relating to homology between the ventral plates and the doublure or free cheeks of trilobites to be tested. Eight additional characters relating to the trilobite glabellar region were added, taken from Paterson et al. (2019), in order to provide data to distinguish internal trilobite relationships. These were Ch. 5, 6, 7, 10, 11, 12, 14 and 15 in Paterson et al. (2019), which are Ch. 37 – 44 in all three matrices used for this study. A further character (calcified thorax separated into prothorax and opisthothorax) was also added (Ch. 99 or 100 in matrices used for this study). Matrices used in phylogenetic analyses are provided through morphobank (Project #P4290. Go to morphobank.org, then Log In. Use credentials: Email address: P4290, Password: Acanthomeridion2023).

Phylogenetic analyses were performed with MrBayes (Bayesian Inference) (Ronquist et al., 2012). Both the ‘maximum information’ and ‘minimum assumptions’ strategies of ref (Bapst et al., 2018) were utilised on an earlier version of the matrix, in order to confirm that model choice was not a major driver in differences for the results (Du et al. 2023). For the final version, the ‘maximum information’ strategy was used, as model choice did not play a large influence on the topologies recovered. Bayesian analyses were run for 20 million generations using four chains, every 1000^th sample stored and 25% burn-in (Lewis, 2001; Ronquist et al., 2012). Convergence was diagnosed using Tracer (Rambaut et al., 2018).

For each coding strategy, two sets of Bayesian analyses were run. The first was unconstrained, whereas for the second all trilobites except for Olenellus getzi were constrained as a monophyletic group to recover the relationships of trilobites (Olenellus as the earliest diverging member) comparable to the comprehensive trilobite phylogenetic analysis of Paterson et al. (2019). Resulting posterior tree samples were compared by comparisons in the consensus trees, and in multidimensional treespace (using R code and method from Pates et al. 2022). The number of trees where Acanthomeridion serratum was sister to trilobites was quantified for each coding strategy and for constrained and unconstrained analyses, and the area of treespace occupied by the posterior samples – and trees where A. serratum was sister to triobites within those samples – were visualized.

3. Results

3.1 Systematic palaeontology

ARTIOPODA Hou and Bergström, 1997

Acanthomeridion Hou et al., 1989

Type and only species

Acanthomeridion serratum Hou et al., 1989

Diagnosis

Non-biomineralized artiopodan with elongate dorsal exoskeleton that gives body an elliptical outline. Subtriangular head shield with deep lateral notches that accommodate stalked elliptical eyes, rounded genal angles of the head shield, an axe-like hypostome, and paired teardrop-shaped plates on either side of hypostome. Head bears large eyes and long multi-segmented antennae consisting of over 40 podomeres anterior to three pairs of small cephalic limbs. Trunk composed of 11 tergites bearing expanded tergopleurae with well-developed distal spines, and a terminal spine. The ninth tergite bears a pleural spine more elongate than others, eleventh tergite reduced and expanded into leaf-like outline. Each tergite bears a pair of biramous limbs with long and dense spines on the endopods, and slender stick-like exopods with long and dense bristles. Modified from Hou et al., 1989, 2017a.

Acanthomeridion serratum Hou et al., 1989 Syn. nov.

Figs. 3–5; 7g-i; S1–6

1989 Acanthomeridion serratum Hou et al., pl. Ⅲ, figs. 1-5; pl. Ⅳ, figs. 1-5; p. 46, text-figs. 3, 4.

1996 Acanthomeridion serratum Chen et al., p. 157, figs 199-203.

1997 Acanthomeridion serratum Hou and Bergström, p. 38.

1999 Acanthomeridion serratum Luo et al., p. 51, pl. 6 fig. 5.

1999 Acanthomeridion serratum Hou et al., p. 130, fig. 188.

2004 Acanthomeridion serratum Chen, p. 280, fig. 442.

2004 Acanthomeridion serratum Hou et al., p. 176, fig. 16.63; p. 177, fig. 16.64.

2017a Acanthomeridion serratum, A. anacanthus, and Acanthomeridion sp in Hou et al., p. 734, figs. 1-4.

2017b Acanthomeridion serratum Hou et al., p. 204, fig. 20.34; p. 205, fig. 20.35.

Type material

Holotype, CN 108305. Paratype, CN 108306–108310.

New material

Sixteen new specimens were collected from Jiucun (Fig. 2), Chengjiang County and housed in the Management Committee of the Chengjiang Fossil Site World Heritage (CJHMD 00052–00062), and Research Center of Paleobiology, Yuxi Normal University (YRCP 0016–0020).

Occurrence

Yu’anshan Member, Qiongzhusi Formation, Wutingaspis–Eoredlichia biozone, Cambrian Stage 3. Jiucun and Maotianshan, Chengjiang County, Yunnan, China; Mafang and Ercaicun, Haikou town, Kunming, Yunnan, China (Fig. 2).

Diagnosis

As for genus, by monotypy.

Description

The dorsal exoskeleton of Acanthomeridion serratum displays weak trilobation and elliptical outline. Specimens measure from 20 to 75 mm along the sagittal axis (Figs. 3a, c; 4a; 5; S1–6). The head shield is subtriangular in dorsal view, with rounded anterior margin and small spines on posterior margin (Figs. 3a, c; 4a; 5; S6a, c). A pair of lateral posterior notches accommodate stalked elliptical eyes. The lateral margin of the head shield is a suture that attaches paired ventral plates. Ventral plates have a teardrop outline, occupy the entire length of the cephalon and terminate in a posterior spine (maximum length 3.4 mm; Fig. S2b) that projects into the thorax as far as the anteriormost tergite (Fig. S1a). The medial margin is curved towards the axially line and sits adjacent to a conterminant axe-shaped hypostome (maximum dimensions 9.6 mm x 11.6 mm; Fig. 3a, b). Four pairs of appendages are present in the head (Figs. 3f–j; S5a). A pair of antennae with at least 43 articles (Figs. 4a, b; S2a, f; S4b) sits anterior to three pairs of small cephalic limbs (Fig. 3f–j). The cephalic appendages are poorly preserved and only the gross morphology can be discerned as no individual podomeres are visible. The protopodite of the cephalic limbs are subtriangular in outline. Endopodites and exopodites are not clearly preserved. Appendage three may include the endopodite, but no podomeres are visible.

Fig. 3

Fig. 4

The trunk is composed of 11 tergites which extend laterally into spinose tergopleurae. Tergites are all subequal in length. Tergites 8-11 curve towards the posterior, and reduce in width progressively, so that T11 is approximately 25% the width of T8 and 33% the width of T9 (Figs. 3a, c; 4a; S1a, c, d). Each tergite has small spines on its lateral and posterior margins (Figs. 6; S1a, c–f; S2i, k; S3a, j; S5b, e, f; S6a–f). The shape of tergites and the convexity of the trunk changes through ontogeny (Fig. 5). In the smaller specimens, the trunk appears slender from a dorsal view as the body displays a high convexity (Fig. 5t). Pleurae curve ventrally, and only slightly towards the posterior (Fig. 5a–k). In larger specimens, the trunk is less convex than for the smaller specimens (Fig. 5s), ventral curvature of pleurae is decreased, and posterior curvature increased (Fig. 5o–r). A linear relationship between trunk axial length and width is obtained (Fig. 6). Where trunk appendages are preserved, each tergite is associated with a pair of biramous trunk appendages (Figs. 3a, c, h, k; 4; S1a, b). The protopodite is sub triangular with a broad attachment to the body wall and short robust gnathobasal spines along the medial margin (Figs. 3h–j; S3a–f, i). The ventral margin of the protopodite has endites longer than the gnathobasal spins. The dorsal edge of the protopodite is poorly preserved and does not clearly show the attachment of the exopodite. Exopodites are slender, stick-like and long (8.3 mm, ca. 3x protopodite length), bearing two rows of lamellae (Figs. 3h, k; S1a, b). Exopods are shorter than the width of corresponding tergites (Figs. 3h, S1a, b). The lamellae appear short but may not be completely preserved. There is no evidence that the exopodite divided into lobes or has articulations. Endopodites are composed of seven trapezoidal podomeres. Podomeres 1–6 bear spiniferous endites arranged in rows, six on pd1, 2, four on pd3, 4, and three on pd5, 6. Podomere seven displays a terminal claw of three elongate spines (Figs. 3h, k, k; 4d). From the preserved appendages, no significant antero-posterior differentiation or dramatic size reduction can be discerned (Figs. 3, S1) One specimen shows evidence for paired midgut diverticulae (Fig. S4a, f). The body terminates in a long, slender spine, dorsal to a paired paddle-like structure (Figs. 4a, e; S5b, g).

Fig. 5

Fig. 6

Remarks

The eyes were previously identified as genal projections (see Fig. 1, Fig. 2A, and Fig. 3A in Hou et al., 2017a) because of the limited available specimens and the posterior location of the eyes. Acanthomeridion anacanthus Hou et al. 2017a is here proposed to be a junior subjective synonym of Acanthomeridion serratum Hou et al.1989, an interpretation supported by:

1. The paddle-like structure under the terminal spine (Figs. 4a, e; S5b, g) was previously interpreted as the twelfth tergite. The specimen YKLP 11115 (see Fig.4 in Hou et al., 2017a) which is a similar size to YKLP 11116 (holotype of Acanthomeridion anacanthus) and YKLP 11113 (see Fig.2A, C and Fig.3A, C in Hou et al., 2017a) show 11 tergites.

2. Specimens CJHMD 00052-55 and YRCP 0016 show the evidently pleural spines of T8 and T9 (Figs. 3a, c; 4a; S1a, c, d; S2a, f; S4b, g).

3. The orientation of pleurae makes some specimens appear narrower than others. Specifically, those where pleurae curve ventrally look more slender than those where pleurae are curved posteriorly. Specimen YRCP 0017 preserves right ventrally curving pleura and left flat pleura, which lead to its right pleura narrower than the left (Figs. 5j, o; S2i). Moreover, most small specimens (1cm–3cm) are preserved with ventrally curving pleurae (Figs. 5a–e; S3a–c; S4a; S5a-c; S6g, i) and these appear narrower than the few small specimens with flat pleurae (Figs. 5f; S4b. See also Fig.2A–D, and Fig.4 in Hou et al., 2017a). Larger specimens are typically preserved with flattened, posteriorly curving pleurae.

4. Twenty specimens show a good ontogenetic series of Acanthomeridion including some that have previously have been assigned to ‘Acanthomeridion anacanthus’ (Figs. 5, 6). Linear measurements do not support distinct species of different sizes, nor with different proportions of thorax axial length to width. The length and width data fit well with a linear equation; most of the data fall within the 95% forecast (Fig. 6b). Data fall close to the fitted line, as indicated by the R² value of 0.95. Lastly, specimens previously assigned to ‘A. anacanthus’ are recovered as smaller A. serratum specimens with similar thoracic shape.

The new specimens and CT data presented herein allow the description of cephalic and trunk limbs, and a conterminant hypostome adjacent to ventral plates in Acanthomeridion for the first time. The presence of a terminal spine on these ventral plates is identified. The ventral plate is separated from the dorsal part of the cephalon by a suture that runs close to the stalked eyes. The possible affinities of these ventral plates, and the phylogenetic implications of these, are discussed below (Subsections 3.2, 3.3) The presence of four appendages in the head (or three in the head and one underneath the articulation) is comparable to Zhiwenia (Du et al., 2019), as well as other artiopodans including trilobites (Edgecombe and Ramsköld, 1999). However, the morphology of the three non-antenniform cephalic limbs is difficult to determine with only the protopodite clearly visible. The lack of observed endopodites and exopodites on these three appendages may be a preservational artefact. Possibly the absence of clear endopodites and exopodites represents specialization of the cephalic appendages with the structures being significantly reduced. Other Cambrian artiopodans have been shown to have appendage specialization (Losso and Ortega-Hernández, 2022) which is frequently found in the cephalon (Chen et al., 2019; Schmidt et al., 2022), but none show reduction of both the exopodite or endopodite. In Sinoburius lunaris, the first two post-antennal appendages have reduced endopodites with elongated exopodites (Chen et al., 2019), and in Pygmaclypeatus daziensis the exopodite is significantly reduced in the cephalic appendages (Schmidt et al., 2022).

Recently, the head of Retifacies abnormalis was interpreted to develop three pairs of uniramous post-antennal appendages and following by one biramous appendage pair (Zhang et al., 2022). It would be unclear the function of an appendage with both distal elements being significantly reduced until more clear appendages are found.

The exopodites of Acanthomeridion also differ substantially from those known in other artiopodans. Specifically, the stick-like nature of the exopodite, the rows of lamellae combined with the apparent lack of segmentation, differ from the paddle shaped exopodites thought to characterise the remainder of the group (e.g. Schmidt et al., 2022, and fig. 4 in Ortega Hernández et al., 2013).

3.2 Possible affinities of the ventral plates in Acanthomeridion

3.2.1. Ventral plates as homologues of trilobite free cheeks

The ventral plates of Acanthomeridion have previously been suggested to be homologues to the free cheeks of trilobites, with the suture between these features and the remainder of the cephalon interpreted as a facial suture (Fig. 1d) (e.g. Hou et al. 2017). Morphological similarities are apparent between the ventral plates and free cheeks in many trilobites (Fig. 7a, b) (Whittington et al., 1997) as well as the path of suture separating them from the rest of the cephalon. Our new data provide additional support for interpreting the ventral plates as homologues of the librigenae. The ventral plates of Acanthomeridion are here shown to be transversely broad, tapering posteriorly to a spine, in a position directly comparable to the genal spine of many trilobites (Figs. 1d; 7a, b). Furthermore, the ventral plate is separated from the rest of the cephalon by a suture that passes near the compound eye. This path is similar to the facial suture (fused circumocular and marginal sutures) of many trilobites especially those bearing opisthoparian facial sutures, which separate the free cheeks from the cranidium passing through the visual surface (Fig. 7a, b) (Whittington et al., 1997). The homology of eye notches (e.g., Luohuilinella and Zhiwenia (Fig. 7e, f)), and eye slits (e.g. Phytophilaspis (Ivantsov, 1999) of petalopleurans, dorsal ecdysial sutures in trilobites and Acanthomeridion has been suggested previously (Du et al. 2019). These eye slits join the dorsal eye with the anterolateral margin of the head shield, possibly forming a single suture that facilitated release of the visual surface and created an anterior gape for the animal to leave the old exoskeleton during ecdysis. However, the evolution from eye notches to eye slits and finally to the loss of the eye slit can been seen, calling into question the affinity between these structures and sutures (Chen et al. 2019). This hypothesis was tested explicitly in the first matrix, where the ventral plates of Acanthomeridion were considered homologous to the free cheeks of trilobites. In this analysis, the dorsal slits and sutures of other artiopodans were also treated as homologous ecdysial sutures in the cephalon (Fig. 8a).

Fig. 7.

Fig. 8.

3.2.2. Ventral plates as homologues to doublure in olenelloid trilobites

In the above scenario, the ventral plates of Acanthomeridion would be homologous to the free cheeks both dorsally and ventrally in trilobites, including the doublure. However it is possible that the ventral plates are only homologous to the doublure, and thus that the suture represents the marginal suture, rather than a fused facial suture. The difficulty lies in the marginal position of the eyes, which could support the interpretation that these are fused sutures, but might also represent a position that facilitated ecdysis through a single, unfused, suture. Considered homologous only to the doublure, the suture in Acanthomeridion separating the ventral plates from the cephalon would be homologous to the unfused, marginal suture in olenelloids rather than the fused facial suture of most other trilobites (Fig. 1e) (e.g. Stubblefield 1936).

Few morphological features can be drawn to support this hypothesis beyond the ventral position of the plates. In particular, the lateral position of the eyes in Acanthomeridion, and the path of the suture close to these eyes, suggests that the suture functioned to release both the ventral plate and the eye. Although some artiopodans are known with expansive doublure such as Squamacula buckorum (Paterson et al., 2012) this structure usually follows the outline of the margin and lacks spines (Whittington et al., 1997).

This hypothesis was tested in the second matrix, where the ecdysial suture of Acanthomeridion was considered unfused (Fig. 8b). In this scenario, the presence of eye slits and other dorsal sutures in other artiopodans would not represent homologous structures, as the proposed evolution is from a ventral-lateral structure to a dorsal one. Thus, the eye slits of the petalopleurans who bear them were not considered as ecdysial sutures for this analysis.

3.2.3 Ventral plates and eye slits of artiopodans as unique structures without homologues in trilobites

Eye slits of petalopleuran artiopodans as homologues to facial sutures of trilobites has drawn support from comparisons such as Loganopeltoides (Rasetti 1948), where a suture charts a path from the dorsal eye to the anterolateral margin of the head shield. However, in Loganopeltoides this suture represents the vestiges of the free cheeks (Edgecombe & Ramksold 1999), a derived state within Trilobita. Given that the features in support of homology between the ventral plates of Acanthomeridion and trilobite free cheeks, and implicit within this a deep root of the facial suture within Artiopoda rather than Trilobita, may not represent true homologues (e.g. spines are often evolved convergently, and the dorsal eye slits of artiopodans have previously been considered unique characters – see e.g. Edgecombe & Ramskold 1999; Chen et al., 2019) it is important to consider the hypothesis that these features are unique to Acanthomeridion. Thus for this third analysis, the ventral plates of Acanthomeridion were treated as their own character in the matrix (Fig. 8c).

3.3 Phylogenetic analyses

All six Bayesian analyses produced the same broad relationships within Artiopoda, with the exception of the position of Acanthomeridion (Fig. 8). Petalopleurans were recovered sister to Nektaspida, Australimicola sister to Conciliterga, and Vicissicaudata was not completely resolved. The positions of Bailongia, Emeraldella, Kwanyinaspis, Squamacula and Zhiwenia were not resolved, and the internal relationships of Artiopodans were also not well resolved. Acanthomeridion was recovered as sister to Trilobita when the ventral plates were considered homologous to the free cheeks of trilobites, but otherwise its position was not well resolved. Within Trilobita, when the analysis was unconstrained, Olenellus and Eoredlichia formed a group sister to the other trilobites, or these two taxa were recovered in a polytomy with a clade containing the other trilobites (SFigs. 8—10). When the group Anacherirurus, Cryptolithus, Eoredlichia, Olenoides and Triarthrus was constrained (with the exclusion of Olenellus), then Olenellus was recovered as sister to all other trilobites (SFigs. 8—10). This result is important for considering the hypothesis that the ventral plates of Acanthomeridion were homologous to the doublure of trilobites.

When visualised in treespace, constrained and unconstrained analyses produce very similar results (Fig. 9). Trees in the posterior sample of the analysis where the ventral plates of Acanthomeridion were treated as homologous to the free cheeks of trilobites fall in the area of treespace supporting Acanthomeridion as sister to trilobites. This result is corroborated by the raw counts in the full sample (Table 1), and by the support for this node as seen in the consensus tree (Fig. 9). Posterior samples of analyses treating ventral plates as homologous to the doublure and those treating them as non-homologous with any trilobite feature overlapped in treespace, over a much broader region. These display a very similar confidence ellipse to the sample of trees where Acanthomeridion is not resolved as sister to trilobites, but instead in a different position in the tree (Fig. 9). Again, these treespace results are corroborated by the raw counts in the full sample (Table 1) and by the poorly resolved position of Acanthomeridion in the consensus trees (Fig. 9).

Figure 9.

Table 1:

4. Discussion and conclusion

All consensus trees, of parsimony and Bayesian results from both coding strategies, support multiple origins of cephalic sutures in Artiopoda (Fig. 8, SFigs. 8—10). When the sutures of Acanthomeridion and trilobites are considered as homologous (e.g., Hou et al., 2017), as well as to those of other artiopodans (e.g., Du et al., 2019), terminals with ecdysial sutures in the cephalon are found in multiple groups in Artiopoda (Fig. 8a), although Acanthomeridion is recovered sister to trilobites. If instead the sutures in Acanthomeridion are considered marginal, and homologous to those of olenelline trilobites, Acanthomeridion and trilobites do not form a clade (Fig 8b, Table 1), and trees occupy a different region of the treespace (Fig. 9). Indeed, the posterior samples of analyses treating the ventral plates as not homologous to any other artiopodan feature (Fig 8c) are extremely similar to those treating them as homologous to the doublure of trilobites, as illustrated by their broad overlap in treespace (Fig. 9) and similar support for nodes in the consensus trees (SFigs. 8—10). Thus, two of the three coding strategies receive some support from the phylogenetic results. Either Acanthomeridion shares the presence of facial sutures with trilobites, and these form a clade, but the sutures in petalopleurans and Zhiwenia are distinct (multiple origins of dorsal ecdysial sutures) or the ventral lateral plates of Acanthomeridion are a distinct feature to the free cheeks of trilobites (and petalopeluran and Zhiwenia features are also distinct; multiple origins of sutures). The position of Acanthomeridion as not sister to trilobites in the analyses treating the ventral plates as homologous to doublure mean that this hypothesis did not gain support, given the assumptions and character matrices of this study.

Support for Acanthomeridion as sister to trilobites comes from morphological similarity between the spinose termination of the ventral plates of Acanthomeridion (similar to trilobite genal spines) and the suture that follows an oblique line and passes near the eye both for Acanthomeridion (similar to trilobite opisthoparian facial sutures; Whittington et al., 1997). Taken together with trilobate exoskeleton of large Acanthomeridion specimens, the non-biomineralised Acanthomeridion bears many morphological similarities to biomineralised trilobites that further support a sister relationship. However, this sister relationship rests on the treatment of the ventral plates as free cheeks. If this interpretation is not favored, then these similarities can be considered convergent, and the position of Acanthomeridion within Artiopoda is not well resolved by these analyses. For this interpretation of the ventral plates, the path of the suture between the ventral plates and cephalon passing close to the eye may represent a way to minimise damage to the eye during ecdysis, which is comparable to what is observed in trilobites with facial sutures.

Other features observed in Acanthomeridion for the first time – uniramous deutocerebral appendages, an axe-shaped conterminant hypostome, two librigenal-like plates and three additional head appendages – are consistent with a phylogenetic position for the species as an early diverging artiopodan, or as sister to trilobites. Multipodomerous uniramous deutocerebral antennae are known in nearly all other members of the clade (Zhang et al., 2007; Du et al., 2019; Zhai et al., 2019), and thus this feature is not very informative. Similarly, a conterminant hypostome was already known in numerous other artiopodans (Zhang et al., 2004; Stein and Selden, 2012). As this type of hypostomal attachment is considered ancestral within Trilobita (Fortey, 1990), its presence in the possible sister to trilobites (if dorsal sutures are considered homologous) is consistent with these expectations from trilobite anatomy. If Acanthomeridion is instead an early diverging artiopodan, the presence of natant hypostomes in both trilobitomorphs and vicissicaudates (Fortey, 1990; Chen et al., 2019) demonstrates that a natant hypostome has most likely evolved repeatedly within Artiopoda.

In summary, new fossils and CT-scan data reveals the ventral anatomy and appendages of Acanthomeridion serratum for the first time, and additional specimens demonstrate that A. ‘anacanthus’ represents a junior synonym of A. serratum, with some features thought to distinguish the two instead ontogenetic in origin. The presence of a posteriorly orientated spine on the cephalic ventral plates of Acanthomeridion might be homologous to the genal spine of trilobite librigenae, providing additional support for a.sister group relationship between these taxa. Phylogenetic analyses interrogating this hypothesis recover a sister relationship between Acanthomeridion and trilobites, but still require multiple origins of cephalic sutures within the group. If the ventral plates are instead treated as homologous to the doublure of trilobites, or as not homologous to any trilobite feature, then no close relationship is recovered, and the morphological similarities between Acanthomeridion and trilobites such as the path of the suture separating the ventral plates from the cephalon, and details of the ventral plates, should instead be treated as convergent. Thus, regardless of how the ventral plates of Acanthomeridion are interpreted, multiple origins of cephalic sutures in artiopodans are recovered by phylogenetic analyses.

Conflict of interest

The authors declare that they have no conflict of interest.

Acknowledgements

The comments and suggestions of three anonymous reviewers greatly improved the manuscript. We thank Jian Han and Jie Sun for scanning the fossils and Javier Ortega-Hernández for helpful discussion. This work was supported by the National Natural Science Foundation of China (grant numbers 42262004, 42202003, 41662003), State Key Laboratory of Palaeobiology and Stratigraphy (Nanjing Institute of Geology and Palaeontology, CAS) (No. 193104). Stephen Pates acknowledges support from a Herchel Smith Postdoctoral Fellowship (University of Cambridge) and NERC Fellowship NE/X017745/1.

CRediT authorship contribution statement

Kunsheng Du: Conceptualization, Data Curation, Investigation, Methodology, Visualization, Writing – original draft, Writing – review and editing. Jin Guo: Resources, Data curation, Validation. Sarah R. Losso: Conceptualization, Data Curation, Formal analysis, Writing – original draft, Writing – review and editing. Stephen Pates: Conceptualization, Data Curation, Investigation, Formal analysis, Methodology, Visualization, Writing – original draft, Writing – review and editing, Supervision. Ming Li: Methodology, Data curation. Ailin Chen: Resources, Funding acquisition, Project administration, Visualization, Supervision.