One of the fundamental problems in the study of the history of life is the timing of the origins of the major groups of metazoans. Molecular clock studies (Runnegar, 1982; Hedges et al., 2004; Peterson et al., 1979; Erwin et al., 2011; dos Reis et al., 2015; Dohrmann and Wörheide, 2017) have placed the origins of the metazoan phyla within the Tonian (max.) to Ediacaran (min.) interval, dating key branching events, including the protostome-deuterostome split and the split between cnidarians and other eumetazoans, at various points within this broad time span. Standing in contrast to the results of molecular clock studies is present understanding of the Neoproterozoic fossil record. Specifically, Ediacaran body fossils currently interpreted as skeletonized or soft-bodied eumetazoans, for example Cloudina, Corumbella, and Kimberella (Fedonkin and Waggoner, 1997; Fedonkin et al., 2007; Bobrovskiy et al., 2018; Dunn et al., 2021), are less than 600 million years old, and thus at present there is a substantial gap between the ages of the apparent first appearances of eumetazoans in the body fossil record and the oldest molecular clock estimates of their times of origin.

Described in this article is the first documented Neoproterozoic conulariid, Paraconularia ediacara n. sp. from the terminal Ediacaran Tamengo Formation (upper Corumbá Group) of west-central Brazil. This conulariid, currently represented by a flattened partial periderm preserving such anatomical features as transverse ribs, nodes, and microlamellae, was originally identified (Van Iten et al., 2014a; Van Iten et al., 2016) as Paraconularia sp. Previously, the oldest known occurrences of Paraconularia were in strata of Middle Devonian age (Hergarten, 1985; Babcock and Feldmann, 1986), and therefore the presence of this genus in the topmost part of the Ediacaran System is truly noteworthy.

Conulariids in general are an extinct order (Conulariida) of marine cnidarians, last occurring in the topmost Triassic (Norian) Stage (Lucas, 2012; Barth et al., 2013), that may have been most closely related to scyphozoan cnidarians of the extant order Coronata (Van Iten et al., 2006a). Conulariids and coronates are united by the possession of a prominent, sessile polyp stage that produced/produces a multi-lamellar, steeply pyramidal or conical periderm bearing (in some species) internal projections along the perradii and interradii (Werner, 1966; Van Iten et al., 1996). Conulariids also exhibit similarities to staurozoans, including tetramerous radial symmetry and prominent gastric septa (Van Iten, 1992; Jerre, 1994; Marques and Collins, 2004; Van Iten et al., 2006a), and thus the hypothesis of a sister group relationship between conulariids and staurozoans (Marques and Collins, 2004) may merit further investigation. The oldest previously known conulariids are Baccaconularia meyeri and B. robinsoni from the Furongian (late Cambrian) Saint Lawrence Formation of southwestern Wisconsin and southeastern Minnesota, United States (Hughes et al., 2000; Van Iten et al., 2006b). Two other Ediacaran fossils, Vendoconularia triradiata (late Ediacaran, White Sea Coast, Russia) (Ivantsov and Fedonkin, 2002; Van Iten et al., 2005; Ivantsov et al., 2019), preserved as molds and casts, and conulariid-like carbonaceous compression fossils from the early Ediacaran Lantian Formation of South China (Yuan et al., 2011; Van Iten et al., 2013), have been interpreted as conulariids or have been compared with this group (Ivantsov and Fedonkin, 2000; Van Iten et al., 2005; Ivantsov et al., 2019). However, hypotheses of homology between these Ediacaran fossils and conulariids have been challenged (Grazhdankin, 2014; Wan et al., 2016; Dzik et al., 2017). In contrast, conulariids exhibit detailed similarities in gross morphology to carinachitiids and hexangulaconulariids, two families of small shelly fossils (SSFs) from the basal (Fortunian) stage of the Cambrian System (Morris and Menge, 1992), and at present there seems to be no better candidate for the nearest relatives of these SSF taxa than conulariids (Guo et al., 2020a; Guo et al., 2020b; Guo et al., 2021). Importantly, the possible presence of Cambrian conulariids or closely related forms immediately above the Ediacaran-Cambrian boundary itself suggests that conulariids may have originated during Neoproterozoic times.

The terminal Ediacaran genus Corumbella, currently known from localities in North and South America and Iran (Pacheco et al., 2015; Walde et al., 2015; Vaziri et al., 2018; Walde et al., 2019; Amorim et al., 2020), has also been interpreted as a polypoid scyphozoan closely related to conulariids (Van Iten et al., 2014a; Van Iten et al., 2016; Pacheco et al., 2015). It should be noted, however, that Walde et al. (2019, p. 335) hypothesized that Corumbella was a worm-like bilaterian.

The discovery of Paraconularia in strata of latest Ediacaran age not only demonstrates that conulariids crossed the crucial Ediacaran-Cambrian boundary, but it also corroborates the hypothesis that phylum Cnidaria originated during the Neoproterozoic. Additionally, the existence of this ancient scyphozoan, extant species of which engage in predation (Pearse et al., 1987), may provide additional support for the hypothesis (e.g., Hua et al., 2003; Schiffbauer et al., 2016) that the origin of predation and complex food webs predated the beginning of the Phanerozoic Eon and the Cambrian Explosion.

The late Ediacaran to earliest Cambrian (ca. 565-539 Ma; see Linnemann et al., 2019) Corumbá Group, named after the city of Corumbá in Mato Grosso do Sul State (west-central Brazil), crops out at the junction of the Amazon Craton, the northern Rio Apa Block, and the folded southern Paraguay Belt (Amorim et al., 2020) (Figure 1). It was initially deposited in an elongate rift basin that evolved into a passive continental margin hosting shallow to deep marine environments. The basin was deformed during the Brazilian Orogeny, which resulted in the formation of the southern part of the Paraguay Belt in southwestern Brazil (Almeida, 1968; Gaucher et al., 2003; Alvarenga et al., 2009; Boggiani et al., 2010; Warren et al., 2014; Amorim et al., 2020). The Corumbá Group exhibits a maximum thickness of about 400 m and is subdivided into five formations (Figure 2). The lowermost, or Cadiueus Formation, consists of conglomerate, sandstone, and shale, while the overlying Cerradinho Formation is composed of sandstone, shale, and carbonates (limestone and dolostone). Above this unit, the Bocaina Formation, composed of dolomite and subordinate shale, directly underlies the Tamengo Formation, which ranges from 80 to 100 m thick and consists predominantly of dark gray carbonaceous limestone and subordinate silty shale (Amorim et al., 2020). Both lithologies yield macroscopic body fossils, the most conspicuous of which are the skeletonized eumetazoans Cloudina lucianoi (in limestone) and Corumbella werneri (in silty shale). The Corumbá Group terminates with the Guaicurus Formation, a thick package of uniform shale which has yielded trace fossils of meiofaunal bilaterians (Parry et al., 2017).

High-precision dating of two volcanic tuffs situated a few meters below the top of the Tamengo Formation yielded mean U-Pb ages of 541.85 ± 0.75 Ma and 542.27 ± 0.38 Ma, respectively (Parry et al., 2017). Combined with an age of 555.18 ± 0.30 Ma for a tuff bed near the top of the underlying Bocaina Formation (Parry et al., 2017), these two dates indicate that the entire Tamengo Formation is latest Ediacaran in age. This conclusion is corroborated by the presence throughout the Tamengo Formation of Cloudina (Figure 2), a likely index fossil for the latest Ediacaran (Xiao et al., 2016). Importantly, Paraconularia ediacara occurred at a level situated well below the top of the Tamengo Formation, and therefore its age may be several million years older than 542 Ma, the approximate age of the second oldest tuff layer mentioned above.

The Tamengo Formation conulariid, formally diagnosed and described below, was collected from an outcrop designated as ELC (Ladário/Corumbá Escarpment) IV (Amorim et al., 2020) and located on the south bank of the Paraguay River near the village of Ladário (Figures 1, 2). The conulariid occurred within a 1.5-m-thick, non-bioturbated, finely laminated silty shale bed, at a level approximately 45 m below the top of the Tamengo Formation (Figure 2). The same bed, including the bedding plane on which the conulariid occurred, also yields Corumbella, vendotaenids, and macroalgae. The conulariid was revealed by splitting a hand-sized slab into two pieces, one bearing the part and the other the counterpart. Both are housed in the paleontological collections of the Department of Sedimentary and Environmental Geology of the Geosciences Institute of the University of São Paulo, under collection number 1T/2301a, b.

Assignment of the conulariid to a new species of Paraconularia is based on two lines of evidence: (1) recognition of a set of gross morphological features uniquely exhibited by conulariids in general (e.g., Hughes et al., 2000; Van Iten et al., 2008); and (2) comparisons with illustrations and reposited specimens of previously described Paraconularia from Devonian through Triassic rocks in North and South America (Driscoll, 1963; Babcock and Feldmann, 1986; Babcock et al., 1987; Babcock, 1993; Leme et al., 2004; Escalante-Ruiz et al., 2014), Africa (Van Iten et al., 2008), Europe (de Koninck, 1883; Slater, 1907; Hergarten, 1985; Barth et al., 2013; Lukševičs, 2020), Asia (Xu and Li, 1979; Zhu, 1985; Swami et al., 2017; Min et al., 2021), and Australia/New Zealand/Tasmania (Thomas, 1969; Waterhouse, 1979; Parfrey, 1982; Waterhouse, 1986). Imaging and Raman compositional analysis of the new conulariid were carried out in the Department of Sedimentary and Environmental Geology of the Geosciences Institute of the University of São Paulo, São Paulo State, Brazil. Low magnification optical examination and light photography were conducted using an OLYMPUS DSX stereomicroscope under low angle illumination. A small fragment of the periderm was coated with gold and examined using a LEO 440 scanning electron microscope (SEM).

Diagnosis: Large Paraconularia with circular nodes that are moderately coarse, widely spaced (10-11 per 5 mm), and extended adapertureward as a short, adaperturally tapered interspace ridge. Faces approximately equal in width. Transverse ribs in the apertural/upper middle region of the periderm low bell-curve-shaped and uninterrupted at the facial midline, numbering 10-11 per 10 mm. Apical angle approximately 10°. Corner sulcus subangular; corners and facial midline without internal thickening or carina.

Description: Part and counterpart of a thin (∼ 0.2–0.5 mm), strongly compressed (transversely) fragment of a steeply pyramidal, four-sided periderm measuring approximately 34 mm long and 24 mm wide and lying parallel to bedding. Exposed portion of the periderm consists primarily of two mutually adjacent, very gently tapered partial faces and the corner sulcus between them (Plate I, Figures 1, 2). Faces originally about equal in width. Nearly smooth inner surface of small portions of the two mostly covered faces visible at the broken apical end of the part (Plate I, Figure 1). Apical region entirely missing. Apertural margin may be partially preserved. Apical angles ∼10°. Length of the complete periderm exceeded 100 mm (as indicated by adapical extension of the truncated facial midlines). External surface of the two exposed faces exhibits 32 trochoidal, thickened, node-bearing transverse ribs separated by broad interspaces and numbering 10-11 per 10 mm. Transverse ribs adaperturally arcuate, approximately bell-curve-shaped (inflected circular curve geometry; Babcock and Feldmann, 1986), crossing both faces without interruption or diminution at the facial midline (Plate I, Figures 1, 2, 5, 7). Transverse ribs bent adapertureward on the shoulders of the subangular corner sulcus, within which they terminate, with the end portions of the transverse ribs from one face alternating with and slightly overlapping those from the adjoining face (Plate I, Figures 1, 3–6). Nodes moderately coarse, separated from each other by a gap that measures approximately 0.5–1.5 (rarely 2.0) node diameters in length, numbering from 10 to 11 per 5 mm (Plate I, Figures 5, 8). Many nodes extended adapertureward as a short, spine-like interspace ridge (adapertural spine; Babcock and Feldmann, 1986) (Plate I, Figure 5). Nodes of every other transverse rib form rectilinear files that are nearly parallel to the facial midline or the nearest corner. Corners and facial midlines without internal thickening or carina. Schott (apical wall) absent. Periderm very finely lamellar (Plate I, Figures 9, 10), apparently organic, though possibly with originally phosphatic microlamellae lost secondarily.

Derivation of name: ediacara, from Ediacaran, the age of the conulariid occurrence.

Type material: The holotype, reposited in the palaeontological collections of the Department of Sedimentary and Environmental Geology, University of São Paulo, São Paulo State, Brazil (1T/2301 a, b).

Occurrence: Thick silty shale bed in the middle part of the Tamengo Formation (upper Corumbá Group) at Locality ELC (Ladário/Corumbá Escarpment) IV (Amorim et al., 2020) on the south bank of the Paraguay River near the village of Ladário in Mato Grosso do Sul State, southwestern Brazil.

Age and horizon: Latest Ediacaran (no younger than 542 Ma), approximately 45 m below the top of the Tamengo Formation (upper Corumbá Group) (Parry et al., 2017).

Remarks and comparisons: Together with previously published anatomical data and illustrations (e.g., Babcock and Feldmann, 1986), the photographic illustrations here presented (Plate I, Figures 1, 2, 4, 5, 8–10; Figure 3) show that P. ediacara exhibits a complex suite of gross anatomical features that is shared only with Devonian-Triassic conulariids that have been placed in the genus Paraconularia. To be sure, there has been some divergence of opinion regarding the diagnostic characters of Paraconularia (e.g., Babcock and Feldmann, 1986; Van Iten et al., 2008; Min et al., 2021), and there is also some uncertainty surrounding the diagnostic characters and taxonomic composition of Conulariida itself (e.g., Hughes et al., 2000). Be that as it may, in recent years it has generally been assumed (e.g., Van Iten et al., 2006b; 2014b) that Paraconularia and similar genera such as Conularia are members of a single, monophyletic group that excludes medusozoan taxa lacking a finely lamellar, steeply pyramidal, organo-phosphatic or (possibly) organic periderm with (usually) sulcate corners bounding four faces bearing regularly arrayed small nodes and/or transverse ribs. Among previously known genera from the terminal Ediacaran and basal Cambrian, only Corumbella (terminal Ediacaran) and Carinachites (basal Cambrian, ca. 535 Ma, Han et al., 2017) are similar to P. ediacara and many other conulariids in having an originally quadrate skeleton showing regular corrugation of the gently tapered sides or faces. Absent in these two genera, however, are the fine details and pattern of arrangement of the transverse ribs of Paraconularia. As discussed in part by Van Iten et al. (2016), the faces of Paraconularia, including the type species, are crossed by adaperturally arcuate or angulated, generally node-bearing transverse ribs which terminate within the corner sulcus. There, the end portions of the transverse ribs trend obliquely adapertureward, with the end portions of the transverse ribs from one face alternating with and partially overlapping those from the adjoining face. Adapertural bending of the transverse ribs on the shoulders of the corner sulcus can be subtle, and it has been reported (Min et al., 2021) that in some species such bending is absent. Along the facial midline, the transverse ribs may be interrupted, with the ends of the ribs on one half face arranged in alternation with those on the other half face, or they may be continuous, as in both P. ediacara and the type species (see de Koninck, 1883, pl. LIV, Figure 9). As in most conulariids having trochoidal (longitudinally) transverse ribs, the periderm thickens from the center line of the interspaces to the axial plane of the fold-like transverse ribs, in such a way as to reduce the relief of the transverse ribs on the inner surface of the periderm (see for example Ford et al., 2016, Figure 3E). In many of the species bearing nodes, including P. ediacara, the adapertural half of each node is developed into a short, spine-like ridge (adapertural spine; Babcock and Feldmann, 1986) that extends part way across the interspaces. Finally, the pattern of arrangement of the nodes on the faces is such that the nodes of every other transverse rib are collinear, forming longitudinal series that are approximately parallel to the facial midline or nearest corner.

In addition to being far older geologically than other described species in Paraconularia, P. ediacara is also distinguishable from them morphologically, being characterized by the following unique set of gross anatomical features: (1) transverse ribs in the apertural/upper middle region low bell-curve-shaped (angulated circular curve geometry; Babcock and Feldmann, 1986), continuous across the facial midline, numbering from 10 to 11 per 10 mm; and (2) rib nodes moderately large, clearly separated from each other and developed adaperturally into a short, spine-like interspace ridge. It is perhaps the second item that is most important, though, as in nearly all other known species the transverse ribs are either smooth (nodes are absent) or the nodes are very small and mutually contiguous or nearly so (Figure 3; see also illustrations in Babcock and Feldmann, 1986). Further grounds for assigning the Tamengo Formation Paraconularia to a new species are provided by the fact that the difference in absolute age between it and Devonian Paraconularia is roughly 130 million years, which is about an order of magnitude greater than the estimated life spans of the longest-lived invertebrate/cnidarian species (Valentine, 1970; Raup, 1978, 1991; Sepkoski, 1992; Lawton and May, 1995).

Preservation and taphonomy: The holotype and only known specimen of P. ediacara n. sp. is a fragment of an elongate pyramidal periderm, the faces of which lie parallel to bedding and probably measured at least 100 mm long when complete. It consists of a combination of skeletal material, possibly altered, and external molds of the same. The periderm does not react with dilute HCl and therefore is not calcareous, as expected given that the periderm of Phanerozoic conulariids is organo-phosphatic (Ford et al., 2016). It should be noted here that Van Iten et al. (2016) incorrectly reported the length of the fragment as 26 mm, which is close to its maximum width. The incompleteness of the fossil, which lacks the apical third and probably part of the middle region, is not an artifact of collecting, as the margins of the specimen are fully bordered by silty shale, and the specimen itself was revealed by splitting the host rock slab. The present distorted state of the specimen, with the two exposed faces and corner sulcus nearly touching the relatively smooth, mostly hidden facial pair/corner beneath them (Plate I, Figure 1), is typical of thin-walled conulariids preserved parallel to bedding in shale (see for example Babcock and Feldmann, 1986, figures. 18.5 and 26.4). The smoothness of the hidden faces, the broken edges of which are nevertheless sharp, also is typical of conulariids and reflects inward thickening and consequent diminution of the relief of the transverse ribs on the inner surface of the periderm. Owing possibly to having been restricted to the exteriormost levels in the periderm, which can undergo exfoliation during splitting of the host rock, the short interspace ridges are not evident at all places on the fossil and are easiest to discern on external molds. Finally, the presence of imbricated fragments of Corumbella in the same silty shale bed that yielded P. ediacara (Rodrigues et al., 2003; Amorim et al., 2020) suggests that the specimen was transported and fragmented prior to final burial.

The high magnification SEM images (Plate I, Figures 9, 10) reveal that the periderm of P. ediacara is composed of extremely thin (< 5 μm), mutually parallel microlamellae. This is the basic microstructure of Phanerozoic conulariids preserving the periderm, which is a bi-composite material consisting of apatitic and organic microlamellae arranged in alternation (Ford et al., 2016). Color images (Plate I, Figures 1, 2, 4, 5) of P. ediacara show that it is variegated, with irregularly bounded areas of pale tan-brown periderm occurring alongside patches of darker-colored periderm. Similar color variation is exhibited by the host silty shale and may reflect post-burial diagenesis and/or chemical weathering. In other words, the original periderm probably was uniform in color, as is the periderm of Phanerozoic conulariids, which however may bear pigmented lines or bands corresponding to internal carinae at the corners and/or facial midlines (Van Iten, 1992). Lastly, and as reported previously by Van Iten et al. (2016), Raman analysis of portions of the periderm failed to yield any lines diagnostic of phosphorous, one of the elemental constituents of apatite; instead, it appears that the periderm is now, and may originally have been, entirely or predominantly organic in composition. This hypothesis can potentially be tested by imaging broken edges of the periderm at very high magnifications. If in fact originally apatitic microlamellae have been lost secondarily, then there might still be an extremely narrow gap between the preserved, organic microlamellae. Otherwise, all microlamellae should be in direct contact with each other. Our photomicrographs (Plate I, Figures 9, 10) show no clear evidence of microscopic gaps, and thus, at this point, we are inclined to think that originally apatitic microlamellae were not present.

The first known Ediacaran conulariid, Paraconularia ediacara n. sp., is diagnosed and described based on an incomplete but otherwise well-preserved specimen from the terminal Ediacaran Tamengo Formation of southern Brazil. The discovery of this body fossil has important implications for studies of the origins of the major groups of animal phyla and the early evolution of marine ecosystems. It is hoped that further collecting at the Ladário localities near Corumbá (Mato Grosso do Sul State) will yield additional material of this conulariid, which provides further support for the hypothesis of a relatively deep Neoproterozoic origin for phylum Cnidaria and therefore, possibly, of predation as well.