Naraoiids are non‐biomineralized euarthropods characterized by the complete fusion of post‐cephalic tergo‐pleurae into a single shield, as well as an extensively ramified digestive tract. Ranging from the early Cambrian to the late Silurian (Pridoli), these arthropods of simple appearance have traditionally been associated with the early diversification of trilobites and their close relatives, but the interrelationships and affinities of naraoiids within Artiopoda remain poorly characterized. Three new species from the Burgess Shale (middle Cambrian, Stage 5) of British Columbia, Canada, are described here: Misszhouia canadensis sp. nov., from Marble Canyon (Kootenay National Park), the first species belonging to the genus Misszhouia outside of China; Naraoia magna sp. nov., from Marble Canyon and also from the Raymond Quarry (Yoho National Park), the largest species of Naraoia described thus far, reaching up to 9 cm in length; and Naraoia arcana sp. nov., from two sublocalities on Mount Stephen (Yoho National Park), defined by its unusual combination of spines. This new material shows that gut morphology is no longer a reliable character to distinguish Misszhouia from Naraoia. We demonstrate that Naraoia and Misszhouia can instead be discriminated morphometrically, based on simple metrics of the dorsal exoskeleton. Our quantitative results also help with inter‐specific discrimination and illustrate possible cases of sexual dimorphism. Phylogenetically, the inclusion of morphometric data adds resolution to our cladogram, although parsimony and likelihood treatments provide somewhat different evolutionary scenarios. In all cases, liwiines are nested within Naraoiidae, resolved as the most derived clade of trilobitomorph arthropods.
Members of Naraoiidae, a family of non‐biomineralized trilobitomorph arthropods within the Order Nektaspida (Euarthropoda, Artiopoda), are known mainly from Cambrian Burgess Shale‐type deposits (Whittington 1977; Chen et al. 1997; Zhang et al. 2007; and other references), but were present in early Palaeozoic seas until at least the late Silurian (Caron et al. 2004). Naraoiid fossils are especially abundant in both the Burgess Shale (Whittington 1977; Caron & Jackson, 2008) and Chengjiang deposits (Chen et al. 1997; Zhang & Hou 1985; Zhang et al. 2007). These euarthropods are distinct among trilobitomorphs in having a dorsal exoskeleton divided by a single articulation into two shields with smooth surfaces, as well as in displaying well‐developed and ramified protrusions of the cephalic alimentary tract (Whittington 1977). All but one of the eight species described so far (Fig. 1) belong to either Naraoia or Misszhouia, the two main genera. Pseudonaraoia hammanni, reported from the Ordovician of the Czech Republic (Budil et al. 2003), remains poorly known and the validity of this third genus is difficult to ascertain. The monospecific genus Misszhouia, originally established from Naraoia longicaudata (Chen et al. 1997), is currently distinguished by the presence of axially‐confined cephalic diverticula (Zhang et al. 2007). Misszhouia was known only from the Gondwanian supercontinent during the lower Cambrian, until 2014 when it was first reported, but not described, from the Burgess Shale (Caron et al. 2014).Figure 1 Open in figure viewerPowerPoint General morphological comparison of naraoiid species. Specimens from the Bertie Formation of Ontario (D), the Burgess Shale (A–C, E, F) or Chengjiang (G, H), preserved in dorsoventral aspect (except B and H), cephalic shield to the top. A–C, Naraoia compacta: A, ROMIP 64522, note reflective preservation of the gut diverticula (detailed in Fig. 4C, D); B–C, ROMIP 61033: B, full specimen preserved dorsolaterally; C, close‐up of anterior trunk appendages (framed area in B); note elongate lamellae along rod‐like exopod shafts and elongate articulated terminal podomeres. D, N. bertiensis, ROMIP 56013; note preservation of a marginal rim along both cephalic and trunk shields. E, N. ‘halia’, ROMIP 62130; note short genal spines and marginal rim. F, N. spinifer, ROMIP 61146, full specimen; note prominent genal and trunk spines, and marginal rim. G, N. spinosa, ROMIP 57961, full specimen; note lateral trunk spines and well‐developed pair of terminal spines, and marginal rim. H, M. longicaudata, ROMIP 57957, full specimen preserved in ventral aspect; note bases of biramous appendages and sternites, and marginal rim. Digital single‐lens reflex (DSLR) images taken using direct light (A, D, G, H) or cross‐polarized light (B, C, F, E), under dry conditions (B–E, G, H) or under water (A, F). Abbreviations: ant, antennule; ca, digestive trunk caeca; cd, pair of cephalic diverticula; dou, doublure; edl, distal segment of exopod; end, endopod; exo, exopod; hec, hemolymph cavity and associated tissues; hyp, hypostome; gs, genal spine; gut, intestinal tract; lam, exopod lamellae; lfo, lateral frontal organ; lsp, lateral spines; ltp, lateral trunk spine; mri, rounded marginal rim (anterior or posterior shield); pts, posterior trunk spine; st, sternite. Scale bars represent 5 mm (A–B, D–F, H); 2 mm (C, G).
There have been longstanding questions about the morphology, taxonomy and phylogenetic relationships of naraoiids and, more broadly, nektaspids (Chen et al. 1997; Zhang et al. 2007). The erection of the genus Misszhouia (Chen et al. 1997) has been contested (Luo et al. 1999; Chen et al. 2002), but was later reinstated based solely on the difference in the morphology of the cephalic diverticula and the ‘elongation’ of the trunk shield, since all of the other original diagnostic characters were either common to Naraoiidae or shared with species of Naraoia (Zhang et al. 2007). Gut morphology, however, is sensitive to ecological variations across biozones, and, thus, particularly variable (Vannier et al. 2014), which may not make it a reliable diagnostic trait at supra‐specific levels. As for trunk shape, the definition lacked quantification, and the extent to which it could overlap the existing variation within Naraoia has remained unknown.
As with other soft‐bodied Cambrian arthropods (e.g. García‐Bellido & Collins 2007), the morphological differences observed between naraoiid morphotypes have also been discussed in the context of sexual dimorphism (Whittington 1977; Zhang et al. 2007). The current state of sexual versus specific differences among Naraoia species is somewhat confusing. Zhang et al. (2007) argued for the validity of the species N. halia, erected by Simonetta & Delle Cave (1975), based on taphonomic and morphometric differences between two subgroups of ‘N. compacta’, each exhibiting two morphs interpreted as separate cases of sexual dimorphism, instead of the single sexual dimorphism proposed by Whittington (1977). However, the alleged morphometric differences were neither shown graphically nor statistically tested. Zhang et al. (2007) also mentioned possible size differences for morphs of N. spinosa lacking spines, and likewise recognized different types of dimorphism, but favoured a single dimorphic subdivision within N. spinosa. A broad clarification, based in part on quantitative data, was therefore long overdue.
Since it was established by Dzik & Lendzion (1988), the validity of the Liwiinae has not been questioned. Elevated to the family Liwiidae by Budd (1999), this group has, in fact, been retrieved as monophyletic and outside Naraoiidae in all phylogenetic analyses that include a wide sample of nektaspids (e.g. Edgecombe & Ramsköld 1999; Caron et al. 2004; Cotton & Braddy 2004; Paterson et al. 2010, 2012; Budd 2011; Legg et al. 2013; Ortega‐Hernández et al. 2013; Stein et al. 2013; Edgecombe et al. 2017; Lerosey‐Aubril et al. 2017). However, current character states considered to be apomorphies for Liwiidae (see in particular Edgecombe & Ramsköld 1999; Cotton & Braddy 2004) are questionable: the overlap of the first trunk somite is shared with naraoiids and other trilobitomorphs, as shown below, and a wide cephalic doublure is also present in N. bertiensis (Caron et al. 2004). Under these circumstances, the phylogenetic position of Liwiidae relative to naraoiids is challenged and needs to be reassessed.
In this paper, we describe two new species of Naraoia, including the largest species known so far, and the first species of Misszhouia from outside China. We also make new observations about the type species, N. compacta, proposing a new interpretation of preservation patterns, which leads, in particular, to a revised definition of the doublure along both the anterior and posterior shields. Compiling all data currently available, we use morphometric and phylogenetic tools to clarify the interrelationships of these new species within Naraoiidae and within the greater context of Artiopoda.Geological setting and taphonomy
The bulk of the new fossil material studied comes from the upper part of the thick Stephen Formation in northern Kootenay National Park, around Marble Canyon and Mount Whymper, British Columbia, Canada (Caron et al. 2014; Mayers et al. 2018, appendix S1). New material collected in 2015 and 2016 from the eastern side of Tokumm Creek, just north of the original 2012 Marble Canyon discovery site, expands the fossiliferous outcrops up to 8.5 km to the north‐west of the main quarry (Fig. 2). No naraoiid specimens were retrieved in excavations at the main Marble Canyon discovery site, although some specimens were collected on talus slopes immediately below it (e.g. the specimen shown in Fig. 14A–B), which could be related to the relatively small collecting effort at that site, where only a few metres of strata were sampled (Caron et al. 2014). An overall survey of the Royal Ontario Museum collections led to the discovery of 10 additional specimens of Naraoia magna sp. nov., which had been previously excavated by Royal Ontario Museum parties immediately to the north of the original Raymond Quarry in Yoho National Park and in layers that are likely to be stratigraphically equivalent. The three specimens of Naraoia arcana sp. nov. occur on Mount Stephen in Yoho National Park (about 40 km north‐west of Marble Canyon) in the lower part of the thick Stephen Formation (Kicking Horse and Campsite Cliff Shale Members; see Fletcher & Collins (2003) and Fig. 1).Figure 2 Open in figure viewerPowerPoint Locality map. Most specimens in this study come from Marble Canyon and surrounding areas (thick Stephen Formation), in northern Kootenay National Park, along the British Columbia‐Alberta border (Canada). Colour online.
Fossils are preserved as carbonaceous films replicated in whole or in part by clay minerals, and embedded within a finely laminated and fine‐grained claystone, typical of Burgess Shale‐type preservation (e.g. Orr et al. 1998; Butterfield et al. 2007; Gaines et al. 2008). A substantial proportion of specimens of both Naraoia magna sp. nov. and Misszhouia canadensis sp. nov. are preserved at an oblique angle (19% of specimens; n = 23/122) (Fig. 3B), or disarticulated (17%; n = 21/122), consistent with slump features regularly observed along the upper part of the thick Stephen Formation (Caron et al. 2014). Across both genera, 62% (n = 75/122) of specimens are preserved quasi‐parallel to the bedding plane. Although the precise stratigraphical origin of most specimens is unknown, we interpret these taphonomic data as evidence for a range of burial conditions, with low energy conditions in a majority of cases. Most of the fossils from the Marble Canyon area lack counterparts and many are highly weathered (Fig. 3A–D), having been collected from talus slopes. Some of the more weathered specimens preserve only the outline of the cephalic and trunk shields; others, however, have retained details of internal structures. The gut and gut caeca tend to be preserved in slight relief (Fig. 3A) or as reflective films (Fig. 4B), like those found in other Burgess Shale species (Butterfield et al. 2007; Aria et al. 2015).Figure 3 Open in figure viewerPowerPoint Size comparison between Naraoia magna sp. nov., Misszhouia canadensis sp. nov. and N. compacta. All specimens to scale, preserved in dorsoventral aspect (except B), cephalic shield to the top. A–B, largest specimens recorded to date: A, Naraoia magna sp. nov., paratype ROMIP 64417; note extent of gut diverticula, doublure and marginal rim along cephalic shield; B, Misszhouia canadensis sp. nov., ROMIP 64438; note preservation of trunk appendages. C–D, smallest specimens recorded to date: C, Naraoia magna sp. nov., ROMIP 64423; D, M. canadensis sp. nov., ROMIP 64447. E, N. compacta, lectotype USNM 57687 for comparison, showing putative preservation of hemocoelic cavity, not previously reported (see larger images Fig. 4A, B). Digital single‐lens reflex (DSLR) images taken using cross‐polarized light (A, B, E) or direct light (C, D), under dry conditions. Abbreviations: ant, antennule; ca, digestive trunk caeca; cd, pair of cephalic diverticula; def, decay fluids; dou, doublure; end, endopods; exo, exopods; gut, intestinal tract; hec, hemolymph cavity and associated tissues; mes, indeterminate mesodermic tissues; mri, rounded marginal rim (anterior or posterior shield); ta, trunk appendages. Scale bar represents 10 mm. Colour online. Figure 4 Open in figure viewerPowerPoint Preservation of cuticle and body cavities in Naraoia compacta from the Burgess Shale (Walcott Quarry, Yoho National Park). A–B, lectotype USNM 57687; specimen shows faint carbonaceous line surrounding endodermic digestive tissues, probably corresponding to the outer margin of the hemocoelic cavity; the coelomic area is surrounded by a broad carbonaceous patch distinctly separated from the trunk margin by a wide, featureless rim (double arrows in A and B); this wide rim possibly accommodates an extensive trunk doublure; inset in B is an interpretative diagram of internal composition based on the fossils; the central mesoderm surrounding the alimentary tract is possibly preserved as a faint dark (A) or reflective (B) trace. C–D, ROMIP 64522; cephalon showing probable doublure (double arrows in C and D) delimited by a faint mesial rim distinct from the very narrow marginal rim (mri in C); the mesial edge of the doublure may broadly correspond to the extent of the hemocoelic cavity surrounding the expansive diverticula (hec in D). Digital single‐lens reflex (DSLR) images taken using cross‐polarized light (A, C) or direct light (B, D), under water. Abbreviations: ant, antennule; ca, digestive trunk caeca; cd, pair of cephalic diverticula; dou, doublure; edm, endoderm; gut, intestinal tract; hec, hemolymph cavity and associated tissues; lfo, lateral frontal organ; mes, indeterminate mesodermic tissues; mri, rounded marginal rim (anterior or posterior shield); ta, trunk appendages. Scale bars represent: 10 mm (A, B); 5 mm (C, D). Colour online. Material and method
The material studied comprises 29 specimens of Misszhouia canadensis sp. nov. (plus 1 suspected but unconfirmed specimen), 77 confirmed specimens of Naraoia magna sp. nov. (plus 15 partial or fragmentary individuals), 3 specimens of Naraoia arcana sp. nov., as well as 5 previously unpublished specimens of Naraoia spinifer (including Fig. 1F), the lectotype (Figs 3E, 4A–B) and other unpublished specimens of Naraoia compacta (Figs 1A–C, E (N. ‘halia’); 4C, D), as well as the holotype and only known specimen of Naraoia bertiensis (Fig. 1D; see also Mayers et al. 2018, appendix S1). Detailed observations of the fossils were made using both Nikon SMZ 1500 and Leica M205C stereomicroscopes. High‐resolution digital images were taken using a Canon EOS 5Dr camera under different lighting conditions, including cross‐polarized lighting. Ammonium chloride (NH4Cl) sublimate was used to emphasize details preserved three‐dimensionally, with specimens photographed using low angle lighting.
Among the new taxa described, only specimens preserved parallel to the bedding plane in dorsoventral aspect, complete and sufficiently well‐preserved, were measured and used in the morphometric analysis, that is, c. 34% (26/77 confirmed specimens) of the Naraoia magna sp. nov. and c. 37% (11/30 confirmed specimens) of Misszhouia canadensis sp. nov. The three specimens of Naraoia arcana sp. nov., the five new specimens of Naraoia spinifer, and the holotype of Naraoia bertiensis were all measured. For specimens not present in the Royal Ontario Museum collections, measurements were taken from publications using imageJ image processing freeware (Schneider et al. 2012). ImageJ was used to measure Naraoia taijiangensis (Peng et al. 2012, fig. 4a, d) and N. ‘halia’ (Zhang et al. 2007, figs 3.2, 4.2, 4.7, 4.8). The freeware Webplot digitizer (Rohatgi 2018) was used to retrieve plot data for Misszhouia longicaudata (Zhang et al. 2007, fig. 31.1, 31.3), Naraoia spinosa (Zhang et al. 2007, fig. 17.1, 17.3) and Naraoia compacta (Whittington 1977, fig. 2c, e).
Quantitative analyses were performed using R (R Core Team 2017) complemented with the packages ape, mixtools, PBSmapping, RColorBrewer, sm, splancs and vegan. Boxplots of total length were produced for each species (Fig. 5) and comparisons of Kernel density plots were performed using the sm.density.compare function (package sm) (Figs 6, 7). Measurements of cephalic length, cephalic width and trunk length of all naraoiid specimens analysed were compared in raw units (mm) and log form (Figs 8D, 9). We also used the ratio of cephalic length/cephalic width as a proxy for cephalic shape. The detection test of the intermediate cluster in Figure 8D based on expectation–maximization (EM) normal mixtures involved both an Akaike's information criterion (AIC) comparison of log likelihoods (obtained from the normalmixEM function in the package mixtools) and a parametric bootstrapped comparison of mixture models (boot.comp function with B = 1000, mix.type = ‘normalmix’ and epsilon = 1e−4); hull coordinates were determined by locating the range of x values containing the intermediate normal distribution in the k = 3 model. In Figures 8D and 9, convex hull intersections were determined using the function joinPolys (package PBSmapping). Round‐up tests of pairwise significance (Table 1) were performed using two‐sided Mann–Whitney tests between taxa. A phenogram of an Unweighted Pair Group Method with Arithmetic Mean (UPGMA) hierarchical clustering was computed based on relative cephalic width and relative cephalic length using a Jaccard dissimilarity index (Fig. 10).Figure 5 Open in figure viewerPowerPoint Boxplots of total length for naraoiid species or morphs. Abbreviations: Chen., Chengjiang; Haik., Haikou; N, Naraoia; ns, no genal spines; M, Misszhouia; gs, with genal spines. The grey boxplot within N. magna represents specimens from the Raymond Quarry. Figure 6 Open in figure viewerPowerPoint Kernel density plots of Naraoia species or morphs for the following metrics: A, total length; B, trunk length; C, cephalic width; D, cephalic length; E, relative cephalic length; F, relative trunk length; G, relative cephalic width; H, cephalic shape. N. magna specimens from the Raymond Quarry were not included here to focus on morph distribution in the Mount Whymper/Tokumm Creek area. Abbreviations: gs, with genal spines; ns, no genal spines. Y‐axis is density; X‐axis is the value of the corresponding metric. Figure 7 Open in figure viewerPowerPoint Kernel density plots of Misszhouia species or morphs for the following metrics: A, total length; B, trunk length; C, cephalic width; D, cephalic length; E, relative cephalic length; F, relative trunk length; G, relative cephalic width; H, cephalic shape. Y‐axis is density; X‐axis is the value of the corresponding metric. Figure 8 Open in figure viewerPowerPoint Morphometric analyses of naraoiid species or morphs. A, cephalic length vs trunk length; aR2 = 0.87, N. compacta; aR2 = 0.95, N. spinosa (gs and ns); aR2 = 0.89, M. longicaudata (Chengjiang); aR2 = 0.94, M. longicaudata (Haikou). B, cephalic width vs relative trunk length; vertical dashed line corresponds to the value 0.65 for relative trunk length, separating most specimens (except one) into two groups: Naraoia species (left) and Misszhouia species (right). C, cephalic width vs cephalic length. D, log of relative cephalic width vs log of cephalic shape (i.e. cephalic width/cephalic length ratio); greyed area corresponds to value interval (−0.882 < x < −0.532) of log of relative cephalic width for the intermediate distribution found when running an EM algorithm of normal mixtures. Solid lines in A and C are plotted linear regressions (grey for N. spinosa, N. compacta and both morphs of M. longicaudata in A; black for all Misszhouia specimens and grey for all Naraoia specimens in C); parallel dotted lines are corresponding confidence intervals (95%); dashed lines show range of predicted values. Figure 9 Open in figure viewerPowerPoint Cephalic shape vs log total length plotted for specimens of Naraoia species or morphs. Solid contours are the convex hulls of the corresponding morphotypes: gs, N. spinosa with genal spines; ns, N. spinosa without genal spines; h, N. ‘halia’; c, N. compacta. Table 1. Results (p‐values) of pairwise Mann–Whitney tests for difference in total body length (A), relative cephalic width (B) and relative cephalic length (C) between naraoiid species or morphs Naraoia Misszhouia arcana bertiensis compacta (ns) halia magna spinifer spinosa (ns) spinosa (gs) taijiangensis canadensis longicaudata (C) longicaudata (H) A Total body length arcana 1.000 0.500 0.977 1.000 0.012 0.036 0.001 0.197 0.017 0.038 0.007 0.507 bertiensis NA 0.258 0.500 0.500 0.667 0.077 0.232 0.250 0.333 0.160 0.857 compacta (ns) 1.000 0.985 0.000 0.001 0.000 0.006 0.000 0.000 0.000 0.014 ‘halia’ 1.000 0.001 0.010 0.000 0.078 0.001 0.002 0.000 0.198 magna 1.000 0.264 0.000 0.000 0.000 0.942 0.742 0.004 spinifer 1.000 0.000 0.001 0.003 0.038 0.045 0.791 spinosa (ns) 1.000 0.015 0.656 0.000 0.000 0.000 spinosa (gs) 1.000 0.050 0.000 0.000 0.000 taijiangensis 1.000 0.000 0.000 0.000 canadensis 1.000 0.522 0.018 longicaudata (C) 1.000 0.015 longicaudata (H) 1.000 B Relative cephalic width arcana 1.000 0.500 0.001 0.017 0.001 0.036 0.002 0.012 0.017 0.005 0.001 0.000 bertiensis NA 0.581 1.000 0.583 0.333 1.000 0.591 0.250 0.167 0.095 0.057 compacta (ns) 1.000 0.259 0.783 0.069 0.089 0.718 0.719 0.000 0.000 0.000 ‘halia’ 1.000 0.174 0.030 1.000 0.312 0.209 0.000 0.001 0.000 magna 1.000 0.061 0.040 0.640 0.598 0.000 0.000 0.000 spinifer 1.000 0.003 0.056 0.202 0.003 0.051 0.000 spinosa (ns) 1.000 0.126 0.054 0.000 0.000 0.000 spinosa (gs) 1.000 0.533 0.000 0.000 0.000 taijiangensis 1.000 0.000 0.000 0.000 canadensis 1.000 0.000 0.667 longicaudata (C) 1.000 0.000 longicaudata (H) 1.000 C Relative cephalic length arcana 1.000 0.500 0.131 0.267 0.003 0.036 0.019 0.015 0.017 0.005 0.001 0.000 bertiensis NA 0.065 0.250 0.083 0.333 0.154 0.120 0.250 0.167 0.095 0.057 compacta (ns) 1.000 0.608 0.000 0.019 0.012 0.000 0.001 0.000 0.000 0.000 ‘halia’ 1.000 0.000 0.010 0.061 0.002 0.001 0.000 0.000 0.000 magna 1.000 0.413 0.054 0.057 0.144 0.000 0.000 0.000 spinifer 1.000 0.448 0.717 1.000 0.000 0.000 0.000 spinosa (ns) 1.000 0.259 0.281 0.000 0.000 0.000 spinosa (gs) 1.000 0.897 0.000 0.000 0.000 taijiangensis 1.000 0.000 0.000 0.000 canadensis 1.000 0.000 0.176 longicaudata (C) 1.000 0.000 longicaudata (H) 1.000
In addition to the 10 naraoiid species (including the three new morphotypes described here), and the 4 liwiine species, we included a variety of artiopod species representative of the main taxonomic groups (based on Edgecombe & Ramsköld 1999; Cotton & Braddy 2004; Caron et al. 2004; Paterson et al. 2010; Ortega‐Hernández et al. 2013; Lerosey‐Aubril et al. 2017), as well as the cheiromorph Yawunik as the outgroup (Mayers et al. 2018, appendix S1). Pseudonaraoia is too poorly known and was not used. We used a morphological matrix composed of 34 taxa and 45 discrete characters, with 2 additional continuous characters being added for a separate analysis based on our morphometric results (Mayers et al. 2018). Inapplicable states were coded as uncertainties and characters remained unordered and unweighted in all analyses. Parsimony analyses (Fig. 11A, B) were performed with TNT v.1.5 (Goloboff et al. 2003; Goloboff & Catalano 2016) as we made use of the ability of this program to include characters with continuous states (Fig. 11B); the searches were heuristic under an unweighted tree bisection and reconnection algorithm (uTBR), 1000 replicates with 10 trees saved for each replicate. The Bayesian analysis (Fig. 12) was run using MrBayes v.3.2.6 (Ronquist et al. 2012) with parameters set to follow the Mkv+Γ model (Lewis 2001), as well as 5 000 000 generations, 4 runs, 4 chains, a tree sampled every 1000 generations and burn‐in of 20%. Character state optimization was investigated using Mesquite 3.31 (Maddison & Maddison 2017).Figure 11 Open in figure viewerPowerPoint Cladograms describing naraoiid relationships. A, frequency‐based consensus cladogram of 70 MPTs obtained from a tree bisection reconnection heuristic analysis with 1000 replicates, based on 34 taxa and 45 discrete characters. B, frequency‐based consensus cladogram of 10 MPTs obtained from a tree bisection reconnection heuristic analysis with 1000 replicates, based on 34 taxa, 45 discrete and 2 continuous characters. Numbers next to nodes show frequency of occurrence (%); unmarked nodes were retrieved in all MPTs. Figure 12 Open in figure viewerPowerPoint Maximum clade credibility cladogram from a Bayesian analysis using an Mkv+Γ evolutionary model, based on 34 taxa and 45 discrete characters (same matrix as for Fig. 11A). Letters at nodes correspond to the following main synapomorphies: (a) 2 Fusion of trunk tergites: present; (b) 10 Lateral compound eyes: very reduced or absent, 14 Hypostome attachment: natant, 28 Overlap of unfused trunk tergites: extensive articulation; (c) 30 Fused pleurae forming lateral spines: absent; (d) 7 Distinct pygidium: absent, 16 Frontal organs within hypostome (hypostome complex): present, 22 Cephalic shield overlapping first trunk sternite, pair of trunk appendages, or trunk caeca: present; (e) 6 Degree of effacement of medial tergite boundaries on fused trunk portion: high‐fused tergo‐pleurae form a homogeneous plate, 20 Shape of cephalon: ovoid truncated, 21 Shape of posterior cephalic margin: convex, 25 Trunk tergites taper anteriorly relative to cephalic shield: present, 42 Ramification/branching of cephalic secondary digestive structures: present. Numbers next to nodes are posterior probabilities < 100.
ROMIP, Royal Ontario Museum Invertebrate Palaeontology, Toronto, Canada.Morphometric results
Species groups emerge from patterns of total length data (Fig. 5): Misszhouia species are distinguished by greater sizes, along with Naraoia spinifer and N. magna sp. nov., while in contrast, N. spinosa (morphs with and without genal spines, referred to as gs and ns respectively) and N. taijiangensis display lower values within the naraoiid size range. Total length and relative cephalic width are efficient descriptors of the various morphs, showing statistical significance for a large number of pairwise comparisons (Table 1A, B). In this context, M. longicaudata from Chengjiang is shown to differ significantly in overall size from the same species from the Haikou area, and N. spinosa (gs) is significantly different in size from N. spinosa (ns); no significant difference is found, however, between N. compacta and N. ‘halia’ (Table 1A, B). Relative cephalic length may be an even better descriptor than relative cephalic width to discriminate Misszhouia species from Naraoia (it is significant for N. spinifer vs M. longicaudata (Chengjiang)) but provides a different inter‐specific signal among Naraoia species (Table 1C). Relative cephalic length outlines a difference between the groups (arcana, compacta and ‘halia’) and (magna, spinifer, both morphs of spinosa, and taijiangensis).
N. magna, N. spinosa (gs) and M. longicaudata (Haikou) show particularly broad ranges of total length (Fig. 5). These ranges reveal bimodal distributions for at least M. longicaudata (Haikou) and N. magna specimens from the Marble Canyon area (Fig. 6). In these two species, however, the bimodality is not clearly reflected in relative size and shape measurements, showing that this polymorphism is isometric. By contrast, N. spinosa (ns) and N. spinifer only display bimodal distributions for relative length ratios (cephalic and trunk lengths), while the values of N. halia’ are bimodal in most cases, but not for relative lengths. M. canadensis sp. nov., in turn, is possibly bimodal for raw length values as well as for relative cephalic width and cephalic shape. N. spinosa (gs) and M. longicaudata (Chengjiang) species show uneven distributions in certain cases, but no clear bimodality (Fig. 7).
Misszhouia species are distinguished morphometrically from Naraoia species by the lengths of their trunks relative to the lengths of their cephala. Misszhouia morphotypes possess longer trunks relative to the lengths of their cephala, which becomes apparent when plotting cephalic length against trunk length (Fig. 8A), cephalic width against relative trunk length (Fig. 8B) or even relative cephalic width against cephalic shape (Fig. 8D). Relative trunk length (or, symmetrically, relative cephalic length) is found to be the best quantitative descriptor to separate Misszhouia from Naraoia, distributing the specimens into two very distinct clusters. A relative trunk length of 0.65 reads as an optimal boundary between specimens of both genera, with Misszhouia specimens representing the upper bound (Fig. 8B, with a single N. spinosa (gs) as outlier).
Naraoia and Misszhouia species also differ in the shape of their heads, particularly as observed between larger specimens (Fig. 8C, D). Although all naraoiids follow a general trend of proportional increase of cephalic width with cephalic length, Misszhouia species tend to develop more ovate heads, as their width slightly exceeds their length, while Naraoia species have slightly more elongate heads (Fig. 8C, D). The subpopulation of very large specimens within N. magna sp. nov. seems to occupy the whole range of variation from elongate to wide cephala (Fig. 8C, D; see also Systematic Palaeontology).
A morphometric discrimination at the species level is more difficult to outline graphically. Although relative cephalic width is most of all a good generic descriptor, we found that it also carries some inter‐specific signal. A log of relative cephalic width appears to isolate a subpopulation of Misszhouia specimens, mainly composed of M. longicaudata (Chengjiang). When a convex hull is applied to the range of points corresponding to this intermediate distribution after identification with an EM normal mixture model (−0.882 < x < −0.532, see Material and Method) a sector of the graph is isolated that indeed corresponds to the location of nearly all M. longicaudata (Chengjiang) points (Fig. 8D). However, AIC and bootstrapped likelihood tests of normal mixture models do not return an optimal subdivision into three normal distributions for the relative cephalic width data. Relative cephalic width is therefore not a strong enough factor to identify subgroups within Misszhouia at present, but it provides a possible morphological basis for separating the two morphs of M. longicaudata. This is also supported by the patterns displayed in Figure 8A–B.
When plotting cephalic shape against the log total length, which summarizes the spread of data points two‐dimensionally based on both shape and raw size, N. ‘halia’ is entirely included within the N. compacta polygon, while N. spinosa (ns) is entirely included within N. spinosa (gs; Fig. 9). Hence the bimodal distributions detected across some metrics in N. ‘halia’ and N. spinosa (ns) do not translate into subpopulations that extend beyond the morphological range of N. compacta and N. spinosa (gs), respectively. Apart from N. taijiangensis, which is fully nested within N. spinosa (ns) at the lowest values of cephalic width, other Naraoia species are plotted along the upper boundary of log total length, at the edge of N. compacta's and N. spinosa's morphological areas.
Owing to the taxonomic signal detected with relative cephalic width and relative cephalic length values, we tested their impact on the phenetic resolution of naraoiid species. The phenogram obtained (Fig. 10) isolates Misszhouia species from Naraoia species, as expected from the strong signal characterizing those two metrics. Morphs of N. compacta and N. spinosa are grouped together, whereas N. spinifer, N. taijiangensis and N. magna form another cluster. N. arcana and N. bertiensis represent morphometric intermediaries, with N. arcana between Misszhouia and Naraoia, and N. bertiensis between the two main Naraoia clusters. These results are in overall accordance with and constitute a good synthesis of the morphometric and statistical analyses presented above.
In light of the morphometric results described above, we consider the ‘halia’ morphotype to be the dimorphic equivalent of N. compacta bearing genal spines, much as there is a sexual morph of N. spinosa with genal spines. Also based on these results, M. longicaudata from Chengjiang and from Haikou could probably be considered as two different species, although a detailed revision of this material is beyond the scope of this study.Phylogenetic results
Our consensus parsimony cladogram obtained from discrete characters only is surprisingly well resolved compared to previous phylogenetic analyses of trilobitomorphs, with most nodes being retrieved in all most parsimonious trees (MPTs). Naraoiid relationships, however, are the exception. In spite of using a specific character to differentiate relative trunk length between Naraoia and Misszhouia, Misszhouia canadensis sp. nov. is here included within a broader Naraoia clade (Fig. 11A). Importantly, we do not obtain a monophyletic Liwiidae. Instead, Liwia is allied with N. spinosa, while Soomaspis and Tariccoia are grouped with N. bertiensis (albeit in only 6% of MPTs). The only somewhat recurrent result concerning relationships between Naraoia species is the grouping of N. spinifer and N. arcana across 46% of the MPTs. In a broader context, the Conciliterga sensu Hou & Bergström (1997) (including Helmetiidae) are resolved as part of a larger clade that includes the trilobites. Emucarididae resolve as sister group to a clade inclusive of naraoiids and petalopleurans sensu lato (Xandarella, Cindarella and Phytophilaspis), which would imply that Petalopleura is in fact a part of Nektaspida.
The implementation of some of the continuous data from the morphometric analyses (i.e. relative cephalic length and relative cephalic width) adds resolution to naraoiid relationships relative to the use of discrete characters only (Fig. 11B). Naraoia is now monophyletic in all MPTs (notwithstanding the inclusion of liwiines), while Misszhouia resolves as a paraphyletic outgroup. The latter result is caused by the morphological proximity of the digestive system between M. canadensis and Naraoia. Liwiine affinities are comparatively less resolved, these taxa being included in a polytomy with N. compacta, N. spinosa and N. bertiensis. N. arcana and N. spinifer remain grouped together by the presence of a median terminal spine. N. magna is found to be the sister taxon of all other Naraoia species, and thus the closest to Misszhouia.
The Bayesian tree (Fig. 12) is relatively less resolved than parsimony‐based results, with most higher rank nodes being supported by very low posterior probabilities, although some new phylogenetic signal is found for internal naraoiid relationships. However, unlike the parsimony‐based topology using discrete characters only, but similar to the analysis including continuous data, the Bayesian tree search found some support for the monophyly of Naraoia + liwiines. Although N. magna and N. taijiangensis were resolved differently, this means that the addition of continuous data to the discrete morphological matrix increased somewhat the congruence between parsimony and likelihood results. The fact that the Bayesian analysis did not include the continuous data implies that the phylogenetic signal for a monophyletic Naraoia + liwiines was already present in the discrete‐only dataset (albeit very weakly), but could not be resolved with a traditional uTBR.
In the Bayesian topology, Misszhouia remains a grade to Naraoia, but other naraoiids are now split between a N. bertiensis + liwiines group and a group containing the remaining Naraoia species. As with parsimony, N. spinifer is the closest relative of N. arcana, whereas N. compacta and N. magna are tentatively grouped together with N. taijiangensis. This scenario implies that N. bertiensis is part of a ‘Naraoia’ lineage with wide doublures that survived beyond the Cambrian and has been identified so far as the Liwiinae.
Relationships between artiopodan terminals under a Bayesian treatment remain similar to those recovered using parsimony, but a different arrangement of higher nodes across Trilobitomorpha appears. Under this scheme, trilobitomorphs previously grouped in a clade with trilobites are spread out as a grade. The basal positions of Buenaspis, Emucarididae and Petalopleura sensu lato with respect to Naraoiidae remains unchanged. A notable difference with parsimony is that Saperion and Tegopelte are here resolved together as the sister clade of naraoiids. Trilobitomorpha itself is supported by the presence of some fusion of trunk tergites, which calls into question the exact position of Kwanyinaspis as a basal member. A better‐constrained clade is defined by the ancestor of trilobites and remaining trilobitomorphs, which is characterized by the absence of a telson.
The overlap of the cephalic shield over the anterior portion of the trunk (ch. 22) and the type of exopod branch (ch. 37) contribute to group a broader Petalopleura Hou & Bergström, 1997, which also includes Phytophilaspis Ivantsov, 1999, Xandarella and Cindarella, along with Tegopelte, Saperion and Naraoiidae. This configuration conflicts with the parsimony trees, in which Saperion remains a more classical concilitergan, a placement supported by the presence of a rounded anterior sclerite similar to those of Kuamaia and Skioldia. Whether Saperion and Tegopelte form a natural group, and thus whether their wholly‐fused morphology has a common origin, can be answered only through a more thorough morphological description of these still poorly‐known taxa.
The retrieval of xenopodans as basal (within Vicissicaudata Ortega‐Hernández et al. 2013) in spite of using a leanchoiliid rather than a stem chelicerate as the outgroup (Aria & Caron 2017), independently supports the hypothesis that the chelicerate‐like features of Sidneyia and Emeraldella (Størmer 1944; Legg 2014) are due to an early divergence from the chelicerate lineage defining the groundplan of the entire Artiopoda.Discussion
Bimodal distributions within morphotypes are patterns that could result from different controlling factors: anagenetic speciation events under directional selection across stratigraphic levels, phenotypic plasticity in response to changing environments, or sexual dimorphism. Stratigraphic data are essential to discriminate between these potential causes, since sexual morphs are more likely to occur in the same beds independently of time‐driven variations. Unfortunately, since most naraoiid specimens, including those from the Burgess Shale, were mainly retrieved from talus slopes or from induced time‐averaged assemblages (e.g. the entire Phyllopod bed) we lack the necessary information to make these distinctions. We can, however, constrain the significance of these morphometric patterns based on the consistency of discrete morphological differences, as well as on the difference between changes in raw size and changes in shape.
Whittington's (1977) view that N. halia is a junior synonym of N. compacta and probably a sexual morph is largely supported by our analysis. This also applies to N. spinosa, which likewise includes a morph with genal spines. In both cases, absolute and relative measurement values overlap between morphs, and no statistical difference can be demonstrated using the best morphometric descriptors. However, those morphs also show bimodal distributions for several metrics, indicating the presence of additional subpopulations. In particular, strong bimodalities for shape values in N. ‘halia’ and N. spinosa (ns) suggest discrete spatial variations: possibly between the Maotianshan locality and other localities for N. spinosa, and between the Canadian and Chinese specimens for N. ‘halia’. The lack of bimodality for the same values in the supposed corresponding sexual morph remains difficult to explain, although is it possible that these variations were subtler or isometric, as seems to be the case for N. spinosa (gs).
By contrast, M. longicaudata from Haikou and N. magna from the Marble Canyon area are clearly composed of two different isometric populations. The example of N. magna, with specimens analysed here coming only from Tokumm Creek and Mount Whymper, suggests that these differences could instead be temporal, and hence represent anagenetic changes and/or phenotypic adaptations to environmental variations. Sexual dimorphism seems less likely in this case because such strong patterns are not detected in other naraoiid morphotypes. Similar results would be expected to some degree as a consequence of the conservation of sexual mechanisms in this small group, especially among sister species, respectively N. compacta and M. longicaudata (Chengjiang).
Ecologically, our newly collected material shows that the gut architecture of N. magna and M. canadensis was very similar to that of N. compacta and N. spinosa, suggesting that they had similar feeding habits. Following Vannier & Chen (2002), the extensive anterior diverticula seen in N. compacta and N. spinosa probably served to produce digestive enzymes and provided space to store food. The difference in gut morphology between M. longicaudata and M. canadensis sp. nov. is an intriguing change in structure, as the anterior digestive tracts in M. longicaudata are completely confined to the axial region (Zhang et al. 2007). M. longicaudata was suggested by Chen et al. (1997) to have fed more regularly than other naraoiids and may not have needed the same gut volume to store food, nor the same volume of digestive enzymes, hence its simplified gut architecture. The change in the gut architecture of M. canadensis may point to a shift in ecology from a regularly feeding organism to an opportunistic feeder or scavenger, probably in the same environment as N. magna. Having accumulated evidence that Misszhouia specimens are not sexual morphs of Naraoia, it is not clear how these two species may have shared the same niche. This question should be answered with a more detailed community analysis of Marble Canyon and surrounding outcrops, currently in progress.
The transfer of characters previously thought to be diagnostic of Liwiidae into the expanded pool of naraoiids + liwiines translates phylogenetically into the integration of liwiines within Naraoiidae. As such, both parsimony and Bayesian likelihood consider liwiines to be naraoiid taxa that experienced reversals of their complete fusion of trunk tergo‐pleurae. This may not be so surprising when considering that variations in the fusion of post‐cephalic segments characterize the entire Trilobitomorpha, potentially leading, as our Bayesian tree suggests, to an explosion of the Conciliterga concept. However, this solution also strongly relies on the fact that limb morphology is not known in liwiines, being optimized on our cladograms as similar to the types observed in Naraoia.
The paraphyly of the genus Misszhouia is surprising, but not inconsistent with our morphometric results insofar as the change in shape would characterize the common ancestor of Naraoia species sensu lato (that is, inclusive of liwiines). In this case, the taxonomy would only approximate the phylogeny, but would appear as a more practical solution than erecting a new monospecific genus for M. canadensis. Notwithstanding, the placement of M. canadensis is essentially due to the expanded cephalic diverticula it has in common with Naraoia, the phylogenetic significance of which has been questioned above; if the morphology of the diverticula is indeed put in doubt, the monophyly of Misszhouia remains secured by its cephalon‐to‐trunk proportions.
Our parsimony‐based results highlight the potential value of implementing morphometrics into phylogenetics, in particular for small groups that provide few discrete characters. Although naraoiid relationships are not yet fully resolved and remain partially incongruent when compared to a (discrete‐only) Bayesian topology, parts of the tree have been readily improved or confidently established, such as the monophyly of Naraoia sensu lato. and the close relationship between N. spinifer and N. arcana. While it is possible that our Bayesian topology yields some false precision relative to parsimony (O'Reilly et al. 2016), it is also capable of detecting signals present in the matrix that uTBR does not, or does more scarcely, as shown by the same examples above. Even a clade supported by an abysmal posterior probability of 7, such as Naraoia senso lato, can be congruent with the same fully‐supported node in a parsimonious consensus obtained with additional data (Fig. 11). This analysis aligns with the view that parsimony may be a suboptimal method for retracing evolutionary relationships (Wright & Hillis 2014), especially with datasets containing a large number of homoplastic characters and/or inadequately‐handled inapplicable states (Aria et al. 2015). When working with palaeontological evidence and substantial amounts of homoplastic characters, and if using simplified evolutionary models and Bayesian statistics can detect weaker phylogenetic signals in these data, the results from this method are worth considering as a whole and should be discarded only after further investigations based on improved data or methodology.Systematic palaeontology
Class Trilobita Walch, 1771; Class Nektaspida Raymond, 1920 (emended here), Subclass Conciliterga Hou & Bergström, 1997; Genus Arthroaspis Stein et al., 2013; Genus Kwanyinaspis Zhang & Shu, 2005; Genus Retifacies Hou et al., 1989.
Lerosey‐Aubril et al. (2017) recently elevated the Class Artiopoda Hou & Bergström, 1997 to the rank of Subphylum; this conflicts with the current status of Trilobitomorpha (part of Artiopoda) as a Subphylum. In consistency with the taxonomy of the other main artiopodan clade, Vicissicaudata Ortega‐Hernández et al., 2013 now established as a Superclass (Lerosey‐Aubril et al. 2017), we likewise change the status of Trilobitomorpha to a Superclass. Only the taxa reanalysed here were considered for the composition of Trilobitomorpha below, and monospecific higher‐level taxa (e.g. Order Retifaciida, Family Retifaciidae) were not used.
Subclass Petalopleura Hou & Bergström, 1997; Family Naraoiidae, Walcott, 1912 (emended here); Family Emucarididae Paterson et al., 2010, Genus Tegopelte Simonetta & Cave 1975; Genus Saperion Hou et al., 1991; Genus Buenaspis Budd, 1999; Genus Phytophilaspis Ivantsov, 1999.
Trilobitomorph artiopodans with the following characters as their ground pattern: lateral eyes reduced or absent; hypostome with natant attachement; extensive articulation overlap between unfused trunk tergites; fused pleurae not forming lateral spines. (Emended from Raymond 1920.)
Parsimony and Bayesian‐based results do not coalesce with respect to the position of Arthroaspis. Because of this uncertainty, we do not include this genus within our emended definition of Nektaspida. However, we transfer to the diagnosis all synapomorphies present at the node already containing Arthroaspis in our Bayesian topology (node b in Fig. 12), for otherwise the only synapomorphy of nektaspids would be (plesiomorphically) the absence of pleural spines along the fused portions of the trunk (ch. 30, node c in Fig. 12).
The phylogenetic scenario presented here implies a number of reversals, in particular that of the reduction of lateral eyes and the fusion of trunk tergites, which are respectively developed and free in Xandarella and Cindarella. However, the placement of Petalopleura as sister group to naraoiids (notwithstanding the position of Tegopelte and Saperion) is a relatively robust result, congruent between parsimony and Bayesian analyses, and is supported by a wealth of characters (node d in Fig. 12). Such reversals may therefore be part of a secondary trend within this group towards adaptations better suited to ecologies that defined ancestral trilobitomorphs.
Due to its expansion as a main subdivision of the Superclass Trilobitomorpha, the Order Nektaspida is hereby elevated to a Class.
Nektaspid trilobitomorphs with the following characters: cephalon with ovoid truncated shape; convex posterior cephalic margin; presence of ramifications on secondary digestive structures; anterior tapering of trunk relative to cephalic shield; complete effacement of tergite boundaries within fused section of the trunk. (Emended from Zhang et al. 2007.)
As we discuss in this paper, we find liwiines to be derived naraoiids. This implies that the characteristic trunk fusion of Misszhouia and Naraoia was secondarily lost, probably to allow more flexibility. We thus de‐elevate the Family Liwiidae to its original status of Subfamily in order to include it within Naraoiidae.
Naraoia compacta Walcott, 1912.
Naraoiid nektaspids with the following characters: trunk length/total length ratio: <0.65; cephalic and trunk margins smooth or spinose. (Emended from Zhang et al. 2007.)
The concept of doublure in naraoiids has been debated. Chen et al. (1997) rejected the interpretation of a narrow cephalic doublure by Whittington (1977). Zhang et al. (2007) later defended the original interpretation based on a re‐examination of the Walcott collection at the Smithsonian NMNH and a large sample of specimens from the Chengjiang Lagerstätte and similar early Cambrian deposits of China. Chen et al. (1997) did recognize the recurrent presence of inner borders in the head shield, but proposed that such rims are rather ‘somewhat strengthened portion[s] of the ventral cuticle’. However, as pointed out by Zhang et al. (2007), the presence of a wide doublure in N. bertiensis (Caron et al. 2004) calls for a reconsideration of these rims as true doublures. In N. bertiensis, the doublure was identified in the holotype mostly thanks to the three‐dimensional preservation of the specimen, giving access to the ventral side of the cephalon; a doublure is otherwise difficult to identify in soft‐bodied arthropods, except indirectly or with the help of cuticular impressions in the cephalic shield. We find that, in naraoiids, and taking N. compacta as an example, internal preservation patterns may indirectly circumscribe the extent of the doublure in both head and trunk.
A first important distinction must be made between the rim proper and the doublure. The rim of the head or trunk shield is narrow and rounded, as can be found in many trilobites, and is associated with the folding of the tergal sclerite on the ventral side creating the doublure per se. This is often what was meant by Whittington when describing very narrow doublures, and is also generally what Zhang et al. (2007) called a doublure. The question raised by Chen et al. (1997) is whether this folding also extends to a wider doublure.
This becomes likely when analysing patterns of internal preservation. Certain specimens of N. compacta (like numerous other naraoiid specimens, see Zhang et al. (2007) and below) allow for a distinction between the digestive tract and its ramifications, an area outlined at the edge of the gut ramifications, a broader carbonaceous area, and, delimited at a specific distance from this area, the margin of the cuticle with circular mesial lines and a distinct rim (Fig. 4). The ramified digestive tract corresponds to the endoderm, which is expected to be surrounded by a thin layer of mesoderm and contained within the hemocoelic cavity. This cavity should thus correspond to the outlined area surrounding the ramified gut; we construe that the muscle layer surrounding the gut proper is not identifiable. This means that the broader area sometimes visible as a faint carbonaceous trace corresponds to the mesoderm closing the hemocoel (Fig. 4A, mes). Muscle tissues are virtually never preserved in Burgess Shale arthropods, despite being large organic masses, implying that these traces are remains from other mesodermic tissues. We think, however, that this patch more or less corresponds to the extent of the mesoderm, for there is no reason to think that the mesoderm would be partitioned proximodistally. This is turn implies that the rim lying outside of this mesoderm is ectodermic, that is, cuticular. In the head, the outer limit of the diverticula overlaps more or less with the extent of the hemocoelic cavity and mesoderm, but the inner edge of the doublure remains visible (Fig. 4C, D), defining a featureless ‘fringe’ that does not contain internal tissues. We think that, in the case of both the cephalon and trunk, this fringe corresponds to a thin doublure forming a tight section of the exoskeleton that does not house mesodermic tissue.Figure 13 Open in figure viewerPowerPoint Naraoia magna sp. nov. from the Burgess Shale (Tokumm Creek, Kootenay National Park). All specimens preserved in dorsoventral aspect, cephalic shield to the top. A, C, holotype ROMIP 64455; A, full specimen, showing major morphological features, including a well‐rounded cephalic shield, extensive cephalic diverticula, ramified gut and rod‐shaped lamellate exopods; C, close‐up of medial area (framed area in A), showing phosphatized folds of the putative oesophagus and three cephalic exopods. B, D, paratype ROMIP 64481; B, full specimen, with split through the trunk shield revealing sternites; D, close‐up of medial area (left, framed area in B; right, counterpart), showing possible u‐shaped oesophagus with multiple phosphatized folds. E, paratype ROMIP 64482; full specimen and second largest specimen recorded to date, showing margins of both cephalic and trunk shields. F–G, paratype, ROMIP 64484; F, full specimen, showing large hypostome preserved with putative traces of oesophagus, note long endopods partly preserved over the trunk shield; G, close‐up of axial area (framed area in F). Digital single‐lens reflex (DSLR) images taken using direct light (A, C) or cross‐polarized light (B, D–G), under dry conditions (A–C, E–G) or under water (D). Abbreviations: ant, antennule; ca, digestive trunk caeca; cd, pair of cephalic diverticula; dou, doublure; end, endopod; exo, exopod; gut, intestinal tract; hec, hemolymph cavity and associated tissues; hyp, hypostome; mes, indeterminate mesodermic tissues; mri, rounded marginal rim (anterior or posterior shield); oes, oesophagus; st, sternite; ta, trunk appendages. Scale bars represent 5 mm (A, B, E–G), 2 mm (C, D). Colour online. Figure 14 Open in figure viewerPowerPoint Naraoia magna sp. nov. from the Burgess Shale (Marble Canyon (A, B) and Tokumm Creek (C–F), Kootenay National Park). All specimens preserved in dorsoventral aspect (except E, F), cephalic shield to the top. A–B, paratype ROMIP 62970, full specimen: A, part; B, counterpart; showing large paired organs at the front of the hypostome area. C–D, paratype ROMIP 64443; C, full specimen; D, close‐up of branching cephalic diverticula (framed area in C). E–F, paratype ROMIP 64508, full specimen showing the margins of the cephalic and trunk shields, as well as the long biramous appendages extending laterally; note that the specimen is buried with the cephalon dipping into the plane of bedding. Digital single‐lens reflex (DSLR) images taken using cross‐polarized light (A–D, F) or direct light (E), under dry conditions (A–D) or under water (E, F). Abbreviations: ant, antennule; anu, anus; ca, digestive trunk caeca; cd, pair of cephalic diverticula; def, decay fluids; dou, doublure; exo, exopod; gut, intestinal tract; hyp, hypostome; lfo, lateral frontal organ; mri, rounded marginal rim (anterior or posterior shield). All scale bars represent 5 mm. Colour online. Figure 15 Open in figure viewerPowerPoint Naraoia magna sp. nov. from the Burgess Shale (Raymond Quarry, Yoho National Park). All specimens preserved in dorsoventral aspect, cephalic shield to the top. A, ROMIP 64552; incomplete specimen showing well‐developed diverticula, including secondary ramifications below the cephalic shield. B, ROMIP 64557; complete specimen with traces of exopods and antennae. C, ROMIP 64559; specimen showing well‐developed cephalic diverticula. Digital single‐lens reflex (DSLR) images taken using cross‐polarized light (A, C) or direct light (B), under water. Abbreviations: ant, antennule; cd, pair of cephalic diverticula; dou, doublure; gut, intestinal tract; hec, hemolymph cavity and associated tissues; lfo, lateral frontal organ; mri, rounded marginal rim (anterior or posterior shield); ta, trunk appendages. All scale bars represent 10 mm. Colour online.
From the Latin magnus meaning ‘large, great’ in reference to the uncharacteristically large size of the specimens relative to all other described naraoiid species to date.
Holotype: ROMIP 64455 (Fig. 13A, C). Paratypes: ROMIP 62970 (Fig. 14A, B); ROMIP 64417 (Fig. 3A); ROMIP 64443 (Fig. 14C, D); ROMIP 64481 (Fig. 13B, D); ROMIP 64482 (Fig. 13E); ROMIP 64484 (Fig. 13F, G); ROMIP 64508 (Fig. 14E, F).
ROMIP 64403–64407 (N = 5); 64410 (N = 1); 64412–64416 (N = 5); 64419–64423 (N = 5); 64426–64427 (N = 2); 64430–64436 (N = 7); 64439–64442 (N = 4); 64444–64445 (N = 2); 64452–64453 (N = 2); 64456–64459 (N = 4); 64461–64462 (N = 2); 64464 (N = 1); 64466–64470 (N = 5); 64472 (N = 1); 64476–64481 (N = 6); 64483 (N = 1); 64485–64496 (N = 12); 64502–64506 (N = 5); 64508 (N = 1); 64550–64559, (N = 10).
Naraoia species with the following characteristics: cephalon rounded, wider than long; trunk shield up to 1.7 times longer than cephalic shield; c. 23–24 pairs of biramous limbs in trunk; genal spines absent; cephalic and trunk margins smooth; posterior trunk margin substraight to slightly convex.
Average length (excluding appendages): 52.1 ± 17.9 mm (n = 26); length of largest specimen: 84.8 mm (Fig. 3A); length of smallest complete specimen (note head shield slightly at an angle in sediment, and therefore this specimen was not used in the morphometric analysis): 14 mm (Fig. 3C).
Marble Canyon, Mount Whymper, Tokumm Creek, thick Stephen Formation, in the Canadian Rockies, Kootenay National Park, and Raymond Quarry Shale Member on Fossil Ridge (Yoho National Park), British Columbia.
Large, dorsoventrally flattened body composed of a subcircular cephalic shield that is slightly wider than long, trunk shield consisting of fused tergo‐pleurae that is longer than wide. Cephalic shield overlapping first trunk somite. Maximum widths of cephalic and trunk shields approximately equal (Fig. 13E). Trunk shield gently tapering posteriorly with convex posterior margin (Fig. 13B). Trunk shield is on average 40% longer than the cephalic shield.
Cephalic shield on average 34.6% wider than long; cephalic shield on average 28.6% shorter than the trunk shield; cephalic margins smooth and convex; posterior cephalic margin overlapping first trunk somite; genal angles rounded and spineless. Narrow to moderately wide cephalic doublure (Figs 13A, B, F; 15A, C), seemingly more extensive anteriorly, c. double the width (10–15% length of anterior shield), compared to the sides (e.g. Figs 13A, F; 15A, C). Presence of a thin, rounded marginal rim all around the cephalic shield, similar to other naraoiid species (Fig. 1A, D, E, H). Carbonaceous traces surrounding digestive organs possibly corresponding to indeterminate mesodermic tissues (Fig. 13E, F).
Frontalmost appendages a pair of antennules, attaching laterally to hypostome and projecting beyond anterior cephalic margin. Following antennules are three pairs of biramous appendages (Fig. 13C). Fourth pair of limbs (which is poorly preserved) is overlapped by both the cephalic shield and the trunk shield; we consider that they belong to the trunk tagma, as described by Whittington for N. compacta (Whittington 1977, fig. 96). Exopod probably rod‐shaped, bearing imbricate lamellae (Fig. 13C).
First pair of gut caeca (cd1) forming extensive, branching diverticula, projecting laterally and anteriorly from the oesophagus, and occupying the majority of the cephalic space (Figs 3A, 13A–B, 14C–D, 15A, C). The diverticula appear to ramify four to five times, ending in approximately 50 ± 2 blind narrow projections near the cephalic margins (Figs 14D; 15A, C). Individual diverticula of cd1 appear long and narrow; bifurcation nodes occur early along the length of caeca (Fig. 15A, C). Distalmost bifurcation occurs near cephalic margin; length of distalmost projections shorter than previous rami. Caeca cd2–cd4 of corresponding somites increasing in size towards the posterior cephalic margin (Figs 13B, E; 14C).
Alimentary canal in trunk is similarly lined with pairs of caeca that are reduced in size relative to cd2–cd4, also coinciding with limb pairs (Fig. 13E, ca); trunk caeca consistent in size along the anterior two‐thirds of the trunk, then decrease in size in tandem with trunk appendages, until they are highly reduced intestinal evaginations near the anus. The U‐shaped oesophagus shows internal phosphatized lining which probably represents compressional folds (Fig. 13C–D).
Thoracic tergo‐pleurae fused together forming a single, posterior shield separated by a single articulation from the cephalic shield. Trunk is on average 40% longer than the cephalon and tapers posteriorly to a convex margin. Maximum width of the trunk only slightly less than that of the cephalon. Trunk margins unadorned. Extent of doublure unclear, although putative mesodermic traces suggest it may be wide (Fig. 13E, F) as in N. compacta (Figs 3E, 4A–B).
At least 21 trunk appendage pairs with the same morphology as the post‐antennular cephalic appendages, albeit larger (Fig. 13A, F). Limbs seem to slightly increase in size over the first half of the trunk, then taper from the middle of the trunk posteriad.
From the Latin arcanus, meaning ‘mysterious’ in reference to its key morphological features, including relatively large genal spines and single posterior terminal spine, characteristics which are individually present in other naraoiid species, but not in combination.
Holotype: ROMIP 64516 (part and counterpart, Fig. 16A, B). Paratypes: ROMIP 64515, 64517 (Fig. 16C, D). Collins Quarry (ROMIP 64516, 64517) and Glossopleura Zone (ROMIP 64515) on Mt Stephen, middle Cambrian.
Naraoiid species with the following characteristics: cephalon ovate; trunk length subequal to cephalic length; narrow and elongate genal spines directed posterolaterally; trunk shield with convex posterior and lateral margins, ending in a short axial spine.
Average length (excluding antennae): 28.7 ± 2.5 mm (n = 3), average ratio of cephalic length and trunk length is 1:1; cephalon on average 54% wider than long. Trunk shield is 5–10% narrower than the cephalic shield.
Mount Stephen. Holotype and paratype specimens from fossil sites ‘G’ ROMIP 64515 (Fig. 16C) and ‘CQ’ ROMIP 64516 (Fig. 16A, B) and ROMIP 64517 (Fig. 16D) respectively (see Fletcher & Collins 2003, fig. 3), Kicking Horse Shale Member of the Burgess Shale Formation (Fletcher & Collins 1998) (= ‘thick’ Stephen Formation) in the Canadian Rockies, Yoho National Park, British Columbia.
Ovate cephalic shield, trunk shield of near equal width; cephalic shield slightly overlapping trunk shield; genal angles each bearing narrow elongated posterolaterally directed spine (Fig. 16A–D), trunk shield gently tapering posteriorly, terminating in a short posteromedial spine.
Cephalic shield ovate, on average 54.0% wider than long; slightly overlaps trunk shield; smooth and convex anterior margin and substraight posterior margin with convex medial projection; genal angles each bearing single elongate posterolateral spine (spines on average one‐third of the length of the trunk); maximum width at sagittal midline. Very thin marginal rim and narrow cephalic doublure (Fig. 16D).
Antennules elongate, multi‐segmented; widest at the base and tapering distally to a fine tip (Fig. 16D). At least 20 podomeres appear beyond anterior cephalic margin; exact number of podomeres unclear. Morphology of other cephalic appendages unknown.
Thoracic tergo‐pleurae fused together forming a single trunk shield separated by a single division from the cephalic shield. Trunk length subequal to cephalon, with width 5–10% less than the cephalon. Lateral margins smooth and convex; posterior margin convex, bearing a short axial spine.
Misszhouia longicaudata Zhang & Hou, 1985.
Naraoiid nektaspids with the following characters: trunk length/total length ratio: ≥ 0.65; cephalic and trunk margins smooth. (Emended from Zhang et al. 2007.)Figure 17 Open in figure viewerPowerPoint Misszhouia canadensis sp. nov. from the Burgess Shale (Tokumm Creek, Kootenay National Park). All specimens preserved in dorsoventral aspect, cephalic shield to the top. A, holotype ROMIP 64408, showing long biramous appendages and digestive system, including secondary ramifications of cephalic diverticula. B–C, ROMIP 64411; B, full specimen, showing putative preservation of hemolymph cavity; C, close‐up of hypostome (framed area in B), showing possible preservation of the lateral frontal organ. D, F, paratype ROMIP 64451; D, full specimen with well‐preserved limbs bearing exopod lamellae, antennules directed posteriorly, and digestive system as well as a putative hypostome organ preserved in relief; F, close‐up of digestive system (framed area in D). Note the greenish color potentialy representing chalcopyrite. E, paratype ROMIP 64450, full specimen showing three‐dimensional preservation of digestive system. Digital single‐lens reflex (DSLR) images taken using cross‐polarized light (A–D) or direct light with specimen coated by ammonium chloride (E), under dry conditions (A–E) or under water (F). Abbreviations: ant, antennule; anu, anus; ca, digestive trunk caeca; cd, pair of cephalic diverticula; def, decay fluids; exo, exopod; gut, intestinal tract; hec, hemolymph cavity and associated tissues; hyp, hypostome; lfo, lateral frontal organ; mri, rounded marginal rim (anterior or posterior shield); ta, trunk appendages. Scale bars represent: 5 mm (A–B, D–E), 2 mm (C, F). Colour online. Figure 18 Open in figure viewerPowerPoint Misszhouia canadensis sp. nov. from the Burgess Shale (Tokumm Creek, Kootenay National Park). All specimens preserved in dorsoventral aspect, cephalic shield to the top. A–D, ROMIP 64511; A, full specimen; B, close‐up of digestive caeca and trunk exopods (right framed area in A); C, close‐up of digestive tract within the trunk (middle framed area in A); note the greenish colour potentially representing chalcopyrite; D, counterpart of A, showing pair of frontal organs and posterior end of trunk with numerous tapering appendages. E–F, ROMIP 64510; E, full specimen with appendages preserved only on one side of the body; F, close‐up of the anterior section (framed area in E) showing possible location of hypostome. G–H, paratype ROMIP 64509; G, full specimen; H, close‐up of cephalic area (framed area in G), showing the first pair of digestive diverticula (cd1), which bifurcate repeatedly and expand laterally into the cheeks of the cephalon. Digital single‐lens reflex (DSLR) images taken using cross‐polarized light, under dry conditions (A, B, D–F) or under water (C, G, H). Abbreviations: ca, digestive trunk caeca; cd, pair of cephalic diverticula; def, decay fluids; edl, distal segment of exopod; exo, exopod; gut, intestinal tract; hec, hemolymph cavity and associated tissues; lfo, lateral frontal organ; mri, rounded marginal rim (anterior or posterior shield); ta, trunk appendages. Scale bars represent: 5 mm (A, D–E, G), 2 mm (B–C, F, H). Colour online.
From the Latin, meaning ‘from Canada’, referring to the country of discovery, which is also the first record of this genus outside of China.
Holotype: ROMIP 64408 (Fig. 17A). Paratypes: ROMIP 64411 (Fig. 17B, C); ROM 64438 (Fig. 3B); ROMIP 64450 (Fig. 17E); ROMIP 64451 (Fig. 17D, F); ROM 64509 (Fig. 18G, H); ROMIP 64510 (Fig. 18E, F); ROMIP 64511 (Fig. 18A–D).
ROMIP 62975 (N = 1); ROMIP 64497 (N = 1); ROMIP 64409 (N = 1); ROMIP 64424–64425 (N = 2); ROMIP 64428–64429 (N = 2); ROMIP 64437 (N = 1); ROMIP 64446–64449 (N = 4); ROMIP 64454 (N = 1); ROMIP 64460 (N = 1); ROMIP 64463 (N = 1); ROMIP 64465 (N = 1); ROMIP 64465 (N = 1); ROMIP 64471 (N = 1); ROMIP 64473–64475 (N = 3); ROMIP 64507 (N = 1); ROMIP 64512–64514 (N = 3).
Misszhouia species with the following characters: ramifying gut diverticula in cephalon with broad lateral extensions; trunk shield 2.2–3.2 times as long as cephalic shield; cephalic shield wider than long, c. 1/3 as wide as trunk.
Average length excluding antennae: 50.21 ± 13.6 mm (n = 11). Largest specimen ROMIP 64438 (Fig. 3B): at least 76.8 mm in length (this specimen is tilted and was not included in the morphometric analyses), smallest specimen ROMIP 64447 (Fig. 3D): 18.6 mm in length.
Marble Canyon, Mount Whymper and Tokumm Creek, thick Stephen Formation, in the Canadian Rockies, Kootenay National Park, British Columbia.
Large, dorsoventrally compressed body composed of rounded cephalic and subrectangular trunk shields, the latter consisting of fused trunk tergo‐pleurae. Greatest width in cephalic shield at sagittal midline. Trunk shield c. 2.2–3.2 times length of cephalic shield, tapering posteriorly.
Cephalic shield ovate to rounded in shape, on average 46% wider than long, 34% wider than trunk shield, 2.2–3.2 times shorter than trunk shield; overlaps first trunk somite; has rounded, spineless genal angles. Marginal rim visible, but extent of doublure unclear.
Hypostomal plate natant, located centrally within the cephalon, behind the anterior cephalic margin by approximately one‐third of the total cephalic length. Anteriorly, hypostomal plate associated with two small lobes (lfo; Figs 17C, F; 18D, F), similar to hypostomal organs in M. longicaudata and probably sensorial.
First pair of appendages elongate, multi‐segmented antennules attaching laterally to hypostome, projecting anteriorly, emerging from under anterior margin of cephalic shield with apparent lateral range of motion. Following antennules are three pairs of biramous cephalic appendages spreading anterolaterally, slightly protruding beneath cephalic shield. Endopod is a narrow, segmented shaft (Fig. 18E). Exopod is rod‐shaped, bearing thin tightly spaced lamellae along most of outer margin; lamellae inserted parallel to one another (Fig. 17A, D).
Presence of gut caeca in cephalon and trunk associated with somatization. First pair of gut caeca are extensively branching diverticula (cd1), extending laterally from the anterior portion of the gut 60–80% of the distance to the margin of the cephalon (Figs 17A; 18D, G, H). These diverticula bifurcate four to five times, ending in pairs of blind narrow projections in the cheeks of the cephalon (Fig. 18H). Posterior to cd1 are three smaller pairs of caeca (cd2–cd4), corresponding to the three post‐antennular somites, increasing in size posteriorly (Figs 17A, E; 18A, D, F). Trunk caeca similar to cd2–cd4. Posteriormost caeca taper gently before the anus (Fig. 18E).
Thoracic tergo‐pleurae fused together forming a uniform post‐cephalic subrectangular shield with smooth margins and effaced segmentation. Trunk is on average 2.74 times longer than the cephalon, and tapers posteriorly to a rounded terminus. The maximum width of the trunk is on average 25% narrower than the cephalon. Extent of doublure unclear.
Trunk bears at least 30 post‐cephalic appendage pairs with the same biramous structure as the post‐antennular cephalic appendages. Appendages are long and extend laterally beyond trunk shield margins (Figs 17A, D, E; 18A, B, D). Appendages gently taper posteriorly from mid‐trunk towards the anus (Fig. 18D, E). From the anterior region of the trunk the limbs project laterally, but are oriented in an increasingly posterior direction as they approach the posterior end of the trunk (Fig. 17A, 18D).
Misszhouia canadensis sp. nov. differs from M. longicaudata in the morphology of the cephalic diverticula. The exact number of appendage pairs was challenging to count accurately in most specimens as they decrease in size rapidly towards the posterior end, a difficulty also noted by Zhang et al. (2007). We provide here an estimation.Conclusions
We have described here Misszhouia canadensis sp. nov., Naraoia magna sp. nov. and Naraoia arcana sp. nov., increasing the diversity of the naraoiid group by almost 50%. Naraoia magna can be distinguished from N. compacta based on its larger body size, the morphology of the first cephalic diverticula and additional pairs of trunk limbs. The discovery of Misszhouia canadensis in the Burgess Shale marks the first time that Misszhouia has been found outside of China, significantly extending the longevity and spatial range of this genus from the first appearance of M. longicaudata during the lower Cambrian (Zhang et al. 2007).
The morphology of the cephalic diverticula can no longer be used as a discriminatory generic character for Naraoiidae. Instead, Naraoia and Misszhouia can be distinguished morphometrically from one another based on the lengths of their cephala relative to their trunks, with a relative trunk length of 0.65 being proposed as a new diagnostic character. Morphometrics otherwise support the attribution of N. ‘halia’ as a sexual morph of N. compacta, the sexual dimorphism among N. spinosa specimens, and the likely presence of two separate species within ‘M. longicaudata’. Size and shape metrics also hint at the existence of subpopulations within established morphotypes, suggesting more evolutionary complexity than previously thought.
The presence of a more robust digestive system in M. canadensis sp. nov. suggests that it could have been a scavenger, driven by opportunistic feeding habits. These changes in morphology point to a shift in the ecology of the Misszhouia genus over very large spatial and temporal ranges.
Based on our phylogenetic results, and, in particular, on the congruent solution between our Bayesian topology and a parsimony analysis of a mixed discrete + continuous dataset, we postulate that Naraoiidae sensu lato form the most derived clade of non‐trilobite trilobitomorphs (that is, nektaspidans), with representatives surviving in the Ordovician (some liwiines) and into the Silurian (N. bertiensis).Acknowledgements
Many thanks to the members of the Invertebrate Palaeobiology lab at the Royal Ontario Museum, including Maryam Akrami, Peter Fenton, Joe Moysiuk and Karma Nanglu, for their help and insights in many forms. The Burgess Shale material from the Royal Ontario Museum was collected under several Parks Canada Research and Collections permits. We are grateful to John Paterson and an anonymous reviewer for their comments and to Sara Scharf for proofreading and editorial suggestions. Gregory Edgecombe kindly provided the artiopodan miniatures used to illustrate our phylogeny. Major funding support for field work comes from the Royal Ontario Museum (Research and collection grants), the Polk Milstein Family, the National Geographic Society (2014 research grant to J‐BC) and the National Science Foundation (2016 EAR‐1556226 Award to Robert Gaines, Pomona College). We also thank Todd Keith from Parks Canada for technical support in the field. We would like to also acknowledge the Dorothy Strelsin Foundation. BM's research was supported by a 2015 National Science and Engineering Research Council (NSERC) Undergraduate Student Research Award through the University of Toronto (Department of Ecology and Evolution) and J‐BC's NSERC Discovery Grant (#341944). This is the Royal Ontario Museum Burgess Shale project number 76.
J‐BC, CA and BM designed research; J‐BC led fieldwork, prepared and imaged fossil material and assembled final plates of fossils; BM and J‐BC collected morphometric data; BM and CA compiled the morphological matrix and wrote character descriptions; CA did the statistical tests, ran the morphometric and phylogenetic analyses and assembled graphical displays of results; BM wrote the first draft of the manuscript; all authors contributed to the observation of the material and interpretation of the results and to the writing of the final version of the paper.