A crown-group cnidarian from the Ediacaran of Charnwood Forest, UK

Cnidarians are a disparate and ancient phylum, encompassing corals and jellyfish, and occupy both the pelagic and benthic realms. They have a rich fossil record from the Phanerozoic eon lending insight into the early history of the group but, although cnidarians diverged from other animals in the Precambrian period, their record from the Ediacaran period (635–542 million years ago) is controversial. Here, we describe a new fossil cnidarian—Auroralumina attenboroughii gen. et sp. nov.—from the Ediacaran of Charnwood Forest (557–562 million years ago) that shows two bifurcating polyps enclosed in a rigid, polyhedral, organic skeleton with evidence of simple, densely packed tentacles. Auroralumina displays a suite of characters allying it to early medusozoans but shows others more typical of Anthozoa. Phylogenetic analyses recover Auroralumina as a stem-group medusozoan and, therefore, the oldest crown-group cnidarian. Auroralumina demonstrates both the establishment of the crown group of an animal phylum and the fixation of its body plan tens of millions of years before the Cambrian diversification of animal life.

with most other fossils in the assemblage but the rims are notably sharper and higher and the goblet surfaces are smooth (Fig. 1). The central ridges show the greatest relief of any fossilized remains on the surface (Fig. 1). The absence of evidence for deformation, the sharper definition and the higher relief of the fossil relative to other co-occurring taxa (for example, rangeomorphs) all imply that the goblets were constructed of stiffer material. As these are negative epirelief impressions, the relief of structures is in the opposite sense-so in life, the raised structure would have been a trough, separating distinct faces of the goblet as a sulcus. There is no evidence for the former presence of biominerals (for example, brittle fracture or dissolution features), leading us to conclude that the goblets were originally organic-walled. There is no original carbonaceous material remaining in any Ediacaran Charnwood Forest locality. The preservation of two faces separated by a deep sulcus is common in fossil cnidarians, such as conulariids (Fig. 3e) and is a consequence of the compression of a three-dimensional organism onto a two-dimensional surface during burial (Fig. 3e).
The different bisections of the two goblets imply that they are preserved in different orientations. The occurrence of two symmetrical faces in one goblet (left-hand), and of a similarly sized face and partial face in the other (right-hand), is consistent with each goblet presenting a different view of an originally tetraradially symmetrical structure, much as in fossil conulariids and Carinachites 3,4 (Fig. 3e). However, it could also represent a biradial structure as with hexangulaconulariids 5 or-if the margins of the left-hand goblet do not represent the margins of the periderm faces-triradial symmetry 6 . The preservation cannot be reconciled with pentaradial symmetry, which would require unequal face widths, a condition not currently known in cnidarians with those symmetry states (for example, refs. 7,8 ). We view tetraradial symmetry as most plausible because we consider that the margins of the left-hand goblet probably reflect the margins of faces and note that the maximum width of the largest face in the right-hand goblet is the same as the maximum width of the equi-sized faces in the left-hand goblet. However, we acknowledge uncertainty that might be resolved by discovery of showing the differentiated stalk and cup of each goblet, well-defined corner sulci (now ridges) and texturally distinct tentacles. the proximal portions of both goblets, including their mutual branching point, are concealed beneath a thin cover of sediment but are nonetheless discernible as topographically and texturally distinct tracts (dashed grey line); see Fig. 2 for more information. RtI file available 76 . additional specimens. The basal-most part of the specimen, past our inferred bifurcation point, does not align with the orientation of either of the two goblets individually, which supports our interpretation of the goblets as bifurfcating rather than two separate individuals. The obscured and indistinct nature of the most basal point means that we cannot say how much of the original organism is missing-the specimen we have may have been much larger in life, with additional goblets that are absent from our preserved view of the organism.
Unlike the goblets, the crowns are preserved in positive epirelief, recording the upper surface of the organism. They have poorly defined margins and faint wrinkling, recording the combined impressions of multiple overlapping projections (Fig. 3f) as is seen in, for example, multifoliate rangeomorphs 9 where multiple branches overlap. The specimen is preserved in lateral view-as with all other fossils on the surface-so it is not possible to see the arrangement of this crown axially, on the interior of the goblets. The projections in the crown bear greatest similarity to tentacles of living animals, but are preserved as a compound impression rather than as individual tentacles. They lack external ornament and do not appear to taper.
The combined taphonomic expression of the fossil suggests stark differences in tissue toughness between the two parts, implying that these were originally constructed of different materials: one more rigid than rangeomorph fronds and able to deform the underlying sediment surface (the goblets) and the other sufficiently less resilient than rangeomorph fronds to have had its volume cast by sediment ingressed from below (the crown) 10,11 . We therefore interpret Auroralumina as a polypoid cnidarian, with a smooth, resistant, organic-walled periderm encasing a soft polyp that bears unbranched tentacles (Fig. 4a). The combination of a polyhedral organic-walled exoskeleton and corner sulci with associated softer tissues emerging from the aperture is compatible with interpretation of this structure as a cnidarian periderm to the exclusion of other potential structures. The body of the polyp would have been inside the cup in life and so only the protruding tentacles are preserved in this lateral view.

Phylogenetic position and morphospace occupation
Our phylogenetic analysis recovers a topology that agrees with most modern molecular studies (for example, ref. 12 ) and places Auroralumina in the medusozoan stem group ( Fig. 4b and Extended Data Fig. 1). We recover olivooids, Pseudooides and conulariids as stem-group medusozoans, which differs from recent analyses that have resolved them as crown-group scyphozoans (for example, ref. 13 ). Together, these data reconstruct the medusozoan ancestor as being broadly scyphozoan-like, with a polyp-encasing periderm (Fig. 4b). Our results are stable when ctenophores are constrained as the sister to all other animals (Extended Data Fig. 2a) and when the specific inter-relationships of the Cnidaria are fixed to match recent molecular phylogenies (Extended Data Fig. 2b).
We investigated morphospace occupation of tubular fossils (those with an external tubular skeleton within which an animal resided) across the Ediacaran-Cambrian transition as a mechanism to determine whether Auroralumina is significantly different from other Ediacaran tubular fossils and whether it is more similar to those fossils found in the early Cambrian period. As disparity analyses are phylogenetically independent, we incorporated a large suite of Ediacaran tubular taxa including those that are controversial and may or may not represent ancient cnidarians. The disparity matrix used in our analyses was based on characters published previously (refs. 14,15 and other publications; see Supplementary Data 2 for a full list) which compared various phenotypic features of tubular, exoskeletal fossils across the Ediacaran and early Cambrian periods.
Calculating the non-metric multidimensional scaling (NMDS) with four axes produced a fair fit (stress value <0.1), so was used as the basis for further analysis. Inclusion of Auroralumina in the Ediacaran tube morphospace increased all aspects of disparity measured here (Fig. 5).
Auroralumina has a major impact on the extent of Ediacaran tube morphospace and brings the Ediacaran tube morphospace closer in position and size to that of the Cambrian. The variance and extent of tubular morphospace occupation is comparatively low in the Ediacaran, indicating that tubular anatomies were not highly distinct, despite an increase in the abundance of tube-forming group(s) at this time 14 . Auroralumina's location in the morphospace confirms that its anatomy is distinct from all other known Ediacaran tubular fossils and it is nested within Cambrian cnidarians, between presumed anthozoan and medusozoan taxa. Overall, morphospace variance increases into the Cambrian for all metrics we analysed, as tubular body fossils become more distinct and disparate and the distinctive Ediacaran nested tube-in-tube morphology 14 declines. Analysis of variance of disparity by group shows that the morphospace occupied by Ediacaran tubular taxa without Auroralumina is significantly different to the Cambrian morphospace (R 2 Pr(>F) < 0.001) but, when Auroralumina is added, the Ediacaran and Cambrian morphospaces are not stastistically distinguishable (R 2 Pr(>F) = 0.586), while the Ediacaran without Auroralumina is significantly different from Ediacaran with Auroralumina (R 2 Pr(>F) = 0.045). This further supports the greater similarity of Auroralumina to Cambrian rather than other Ediacaran taxa.

Discussion
Molecular clocks recover a Precambrian divergence between the cnidarian lineages, with subsequent radiations through the early Palaeozoic 16,17 and there have been many claims of Ediacaran fossils with cnidarian affinity, perhaps most notably Haootia, Corumbella and Cloudina.
Haootia quadriformis is a polypoid organism from the Fermeuse Formation of the Bonavisata Peninsula, Newfoundland, that has been described as a total-group cnidarian 18 but has been broadly discussed as a stem-group medusozoan or as staurozoan-like, primarily based on the presence of interpreted tetradial symmetry and an open calyx. A stem-group medusozoan hypothesis for Haootia is contingent on the placement of Staurozoa as sister to all other medusozoans and their anatomy consequently being plesiomorphic for the group as a whole. However, this topology has not been recovered by recent phylogenies of either molecular or morphological data 12,19 nor is it recovered here. This means that it is likely that their medusan anatomy-the major life stage-is derived and does not represent the plesiomorphic condition for Medusozoa. While a staurozoan affinity has never been formally proposed, clear similarities in the reconstructed muscle anatomy of Haootia and stauromedusans have invited comparison. However, the arrangement of muscle fibres is contested 20,21 and the currently described muscle arrangement is not compatible with the feeding mode of known staurozoans or any extant cnidarian. Because of confusion over the life history stage of Haootia, we elected not to include it in our matrix.
There are also a number of latest Ediacaran skeletal fossils that are potential early cnidarians. Corumbella wernerii was likened to the conulariids on the basis of an externally annulated skeleton and tetraradial symmetry 22 . However, more recent work on three-dimensional specimens from the Tamengo Formation, Brazil 23 , has challenged this, arguing that Corumbella is circular in outline and lacks a pyramidal shape, carinae, straight facets or corner grooves, all of which are incompatible with a conulariid affinity. Previous authors 23 go on to suggest that the surface ornamentation of Corumbella is much more like another Ediacaran genus, Sinotubulites. However, the annulated, elongate, tapering tube (with approximately circular cross-section) of Corumbella is compatible with a cnidarian affinity and does bear notable similarity to extant coronate scyphozoan dwelling tubes 24 .
The late Ediacaran genus Cloudina is one of a number of Ediacaran tubular taxa that possess a distinctive funnel-in-funnel morphology. The affinities of Cloudina and similar taxa are controversial, with some authors arguing for an annelid affinity while others compare them with non-bilaterians, chiefly cnidarians. Proponents of an annelid affinity for Cloudina have argued that the putative presence of direct development excludes a placement in Cnidaria 25 ; however, there are several Cambrian, skeletonizing fossil cnidarian taxa known to undergo direct development (see below). Furthermore, the annelids with which Cloudina has been closely compared (Serpulidae 26 and Siboglinidae 25 ) both go through indirect development via a trochophore larva 27,28 , a feature common to many marine annelids and their close relatives. The tube microstructures in Cloudina that are comparable with those of annelids have evolved many times (for example, in Alvinellidae and Siboglinidae 25 ), while the granular tube microstructure of Cloudina is found in living cnidarians but is absent in calcareous tube-forming annelids 29 , along with polytomous branching 30 , a lack of attachment structures and a closed tube base (except in individuals that have undergone damage) 29 . Further evidence for a total-group bilaterian affinity was provided by the discovery of fossilized soft tissues, interpreted as a through gut 31 . The proposed gut morphology was used as evidence against a cnidarian affinity due to the absence of features characteristic of anthozoans, such as an actinopharynx, and longitudinal septa are also absent from the skeleton. However, these features are not present in medusozoan polyps 32 with many medusozoans having a gut gross morphology that is broadly comparable with that observed in the soft tissues of cloudinomomorphs. Furthermore, there are a variety of annelid-mimicking bilaterian groups known from the Palaeozoic era, although these mostly first appear from the Ordovician period onwards 33 . While recent discoveries have provided critical insights into the tube ultrastructure, growth and  soft-tissue structures of cloudiniids, placing Cloudina in the total group of any animal phylum may be premature and we chose not to consider it in our phylogenetic analysis. Embryonic and post-embryonic fossils from the earliest Cambrian, alongside small shelly fossil remains, provide the most compelling evidence for a diverse cnidarian fauna by this time and are the most character-rich of any proposed cnidarian fossils so far covered here. These fossils are likely to represent at least a grade of organization and are sometimes considered a clade, sister to the coronate Scyphozoa. Olivooides 8 and Quadrapyrgites 34 possess angular, ornamented periderm and an aperture constructed of subtriangular lappets. Intriguingly, stacked pentameral ephyrae have been found in association with Olivooides, suggesting that it underwent polydisc strobilation and probably produced medusae. Similarly, hexanguloconulariids are direct developing polypoid organisms 13 that possess a tubular periderm with angular faces, different levels of external ornamentation and peridermal teeth/ ridges but they lack the apertural lobes of olivooids. Carinachites 3 possesses an Olivooides-like apical aperture but, where the periderm of olivooids and hexangulaconulariids is marked by angular faces, Carinachites is marked by deep-corner sulci, much like the conulariids. In addition, conulariids themselves may possess apertural lobes and peridermal teeth/ridges 4,35 . There is some evidence to suggest that conulariids may have undergone strobilation 36 but this is contested 37 . Unlike olivooids, carincachitids and hexangulaconulariids, conulariids possess majority tetraradial symmetry. Conulariids are the longest-ranging of these groups, extending from the latest Ediacaran period (with the occurrence of Paraconularia in the Tamengo Formation) to the Late Triassic period (for example, ref. 4 ). Much has been made of the similarities not just between olivooids, hexangulaconulariids, carinachitids and conulariids but between these fossil groups and extant coronate scyphozoans (for example, ref. 38 ). There are, however, several differences between these Palaeozoic fossil groups and living scyphozoan polyps. The fossil groups can exhibit deep-corner septa in their periderm, or at least the periderm is constructed of smooth faces and the ornamentation across these polyps is regular. This is unlike scyphozoan polyps, which exhibit a smooth, conical periderm, often with irregular ornamentation 24,39 . Furthermore, scyphozoan ephyrae present a number of characters, including velar and rhopalial lappets 40 , which are lacking in the ephyrae of Olivooides. These characters render the morphology of the Olivooides ephyrae unique among medusozoans. Additionally, the interpreted direct development in many of these fossils is unlike living scyphozoans 8,13 .
Auroralumina presents a suite of characters in common with these early Cambrian forms: deep-corner sulci and a polyhedral, probably tetraradial, tubicolous periderm. However, unlike these other groups, the periderm is not ornamented and is not tapering: it is split into two distinct regions, the stalk and the cup. However, the fossil is incomplete, raising the possibility that Auroralumina may only have been annulated over the most proximal part of its body. This is a condition observed in some living hydrozoans 41 , where annulation is often less prevalent 42,43 and where tubicolous periderm can additionally form a stalk and cup 32 . However, a non-annulated periderm also warrants comparison with the skeleton of certain anthozoan cnidarians including cerianthids, anemones and certain octocorals. Cerianthids produce organic tubes with adhered sediment grains. These tubes are circular in outline and can show either irregular concentric annulations or longitudinal striations. This is not compatible with the anatomy of Auroralumina which is polyhedral in outline, has no evidence of adhered sediment grains (evidenced by the smooth nature of the periderm) and is additionally   fundamentally inconsistent with burrowing. Furthermore, the presence of corner sulci is distinct from the longitudinal striations seen in cerianthids 44 . Some anthozoans, including the anemone   an irregular outline 47 and is therefore not comparable to the condition in Auroralumina. The octocoral Cornularia has been described as showing periderm and was recently recovered as the sister to all other octocorals 46 but this condition is unique amongst octocorals, requiring it to be lost in all other octocorals and hexacorals to be plesiomorphic for anthozoans; this is a scenario we consider unlikely. Rather, members of the polyphyletic grouping Stolonifera, to which Cornularia belongs, show differing levels of thickening surrounding the anthostele-the basal part of the polyp which is connected to other polyps in the colony-and around the stolons 48 . It is possible that the periderm of Cornularia represents a derived condition within the group, derived from the chitinous covering of the stolons which, as the plesiomorphic condition of crown-group anthozoans, medusozoans or cnidarians, is not inferred to have been colonial 49 .
On the basis of these data, we consider it most likely that the periderm of some anthozoans is the result of convergence, rather than shared ancestry with medusozoan periderm, and we argue that Auroralumina exhibits a periderm that is homologous with those of medusozoan cnidarians given the additional shared characters. Living members of Staurozoa, Cubozoa and Scyphozoa, which all produce medusae via transformation of their polyp apex, possess polyps that are an order of magnitude smaller than those of Auroralumina ( Fig. 4b and Extended Data Fig. 3). Hydrozoan polyps show far greater variation but do not all produce medusae; those that do produce medusae use a strategy of budding laterally via an entocodon, derived as per our ancestral state reconstructions (Fig. 4b).
The small polyp size in living scyphozoans, staurozoans and cubozoans ( Fig. 4b and Extended Data Fig. 3) could be the result of close phylogenetic relationship but it may also be a feature of their medusa-formation strategy. Living hydrozoans are not so constrained (though they are still smaller than Auroralumina), despite close phylogenetic relationship, and do not use a strategy that is so obviously contingent on polyp apex size. It is, at present, impossible to say when the medusa stage evolved along the medusozoan stem and it is not known whether there is a maximum polyp size after which medusa formation is impossible. However, the large size of Auroralumina in comparison to living medusozoans and our reconstruction of apical/ oral medusa formation as plesiomorphic for the group, does raise the possibility that Auroralumina was not able to produce medusae. Refinement of the phylogenetic inter-relationships of stem-group medusozoans may shed further light on the position of Auroralumina in the stem and on the timing of the evolution of medusa formation, which is currently constrained as minimally earliest Cambrian 50 .
Auroralumina displays a distinctive combination of characters that is not present in other fossil taxa and which helps to better resolve the phylogenetic affinities of several extinct medusozoans, shedding light on the early evolution of a number of key cnidarian traits. We infer the presence of a tubicolous periderm in the ancestor of both the total-group and crown-group Medusozoa, implying its independent loss/reduction in living staurozoans, cubozoans and some hydrozoans (Fig. 4b) and, together, our data suggest a scyphozoan-like ancestor for the crown group of Medusozoa. Additionally, Auroralumina has the greatest polyp width of any medusozoan we are aware of ( Fig. 4b and Extended Data Fig. 3) and is at least an order of magnitude larger than living staurozoans, cubozoans and scyphozoans. There is substantial variability in the size of hydrozoan polyps, with some solitary polyps (for example, Branchiocerianthus imperator 51 ) reaching metre-scale sizes but these are outliers from the largely colonial and miniature group. Conulariid and Corumbella polyp widths are also larger than any living member of these groups, suggesting that small polyp size is a feature of the crown group of the medusozoan lineages. Auroralumina possessed a polyhedral, probably tetraradial periderm with deep-corner sulci, allying it with other fossil medusozoans. A tubicolous, sulcate periderm clearly differentiates early medusozoans from early anthozoans, which we infer to have been naked, anemone-like, polyps (sensu 49 ). A naked anemone-like polyp would have a reduced preservational potential under most settings as compared to a peridermal polyp 52 , perhaps going some way to explain the preponderance of stem-group medusozoans in the fossil record to the exclusion of clear stem-group anthozoans.
Auroralumina confirms the presence of crown-group cnidarians coeval with the oldest assemblage of the Ediacaran macrobiota and is the most ancient fossil that can be reliably ascribed to the crown group of any living animal phylum. Living cnidarians (other than derived, parasitic groups) use their tentacles to catch food and the presence of a dense tentacular crown would support a similar life-habit for Auroralumina. Auroralumina may have fed on diversifying phytoplankton 53 or protists 54 but, additionally, may have consumed an emerging zooplankton. The presence of a stem-group medusozoan necessitates the presence of other cnidarians, as well as other early-diverging animal lineages known to have planktonic phases in their life cycle (poriferans and ctenophores). Additionally, the cosmopolitan distribution of rangeomorph taxa has been used to infer the presence of a water-borne planktonic propagular stage 55 on which Auroralumina may have fed.
Auroralumina is a thecate cnidarian with a polyhedrally symmetrical periderm and extends the fossil record of crown-group cnidarians by ~25 million years, deep into the Ediacaran period. Animal body plans are widely assumed to have become fixed during the Cambrian Explosion but Auroralumina demonstrates that at least one crown-group cnidarian body plan had already been established tens of millions of years previously.

Methods
The specimen was imaged at the British Geological Survey using RTI 56 . Images were derived either from the RTI or from photography using low-angled light.
We analysed our morphological data matrix in MrBayes 3.2.7, under the Mki with gamma model, with correction for use of only informative characters. Our matrix was based on a previously published dataset 13,19,57 which was expanded to include additional taxa and fossil organisms. The R package Claddis was used to perform safe taxonomic reduction and prune uninformative taxa 58 . Analyses were set to run for 20 million generations, with a requested stop value meaning that the analyses stopped when the deviation of split frequencies dropped below 0.01. Convergence was assessed by checking that the effective sample size was >200 and that the potential scale-reduction factor was ~1. Our topology recovers a monophyletic Anthozoa and reciprocal monophyly of the two scyphozoan clades in a polytomy with Cubozoa and Scyphozoa (which are monophyletic). This tree differs in some ways from recent molecular phylogenies. Therefore, we constrained the lineages in our tree to conform to molecular results 12,49 but allowed all fossils to wander. Additionally, we constrained ctenophores as sister-group to all other animals 59 to test the sensitivity of our results to recovering a monophyletic Coelenterata. Finally, we performed ancestral state reconstructions incorporating phylogenetic uncertainty on characters of interest and these were also assessed for convergence using the above parameters. See Supplementary Information for full description.
All disparity analyses were performed in R. The distance matrix used in the disparity analyses was computed using Gower's similarity metric, as this allows for handling of nominal, ordinal, multistate and (a)symmetric binary data 60 . The multidimensional space was then constructed using NMDS, through the vegan package 61 . NMDS is a non-eigenvector-based multivariate method that does not assume linear relationships between the variables and allows for large amounts of missing data. The number of axes used in the calculation of the multidimensional space was determined through visual assessment of a stress scree plot and calculation of the stress values for set numbers of axes. Stress values of <0.1 were taken as a good fit and <0.05 as an excellent fit, following ref. 62 . The data were then analysed using the dispRity package 63 . This package allows for useful visualization of the morphospace, as well as allowing the user to define the metrics by which to analyse the data. It also includes functionality for analysing subsets of the data within the morphospace and for non-parametric analysis of variance (NPMANOVA). Several metrics were used herein to gain a complete picture of how disparity changes through time (following ref. 64 ). Each metric was run on a bootstrapped matrix output for each data subset. For details on disparity matrix construction, see Supplementary Information.
We expanded that dataset to include Auroralumina and a number of additional Cambrian genera with tubular skeletons that are inferred to represent ancient cnidarians: Byronia 65 , Sphenothallus 66 , Olivooides 8 , Quadrapyrgites 34 , Pseudooides 13 , Arthrochites 67 , Carinachites 3 , Septaconularia 67 , Anabarites 68 , Conulariella 4,69 , Paraconularia 38 and Baccaconularia 4,37 . We also separated the genus Cloudina and scored two individual species-Cloudina carinata 70 and Cloudina hartmannae 71but did not include any additional species because distinguishing characters were too specific for the purposes of this study (assessed in ref. 72 ). We also separated the genus Sinotubulites into four individual species-S. triangularis, S. pentacarinalis and S. hexagonus from ref. 7 , alongside S. baimatuoensis 7 . Additionally, we rescored the symmetry state of Corumbella from ref. 14 after ref. 23 , which reports new data suggesting a circular cross-section and not a polyhedral cross-section. Finally, we added taxa forming the CLT clade of Park et al. 15 and additional corralomorph taxa (see Supplementary Information).
Our list of included Cambrian tubular fossils is by no means exhaustive but we have included major early Cambrian tubular fossils that are of proposed cnidarian affinity. Our list of included Ediacaran tubular fossils after ref. 14 , does include all described forms to at least genus level.
The sum of variance metric calculates the sum of variance of each computed axis of each matrix subset-higher variance indicates less even occupation of the matrix 64,73 . Adding Auroralumina increases both the mean and interquartile range (IQR) of the sum of variances of each axis occupied by the morphospace but these are still lower than that of the Cambrian tube morphospace. There is overlap between the IQRs of the Ediacaran and the Ediacaran including Auroralumina and between the IQRs of the Ediacaran including Auroralumina and the Cambrian. Adding Auroralumina means that the Ediacaran morphospace is occupied less evenly but not as unevenly as the Cambrian morphospace. This is also clear from the NMDS plot-all Ediacaran tubes excluding Auroralumina plot in the same region of space, while Auroralumina occupies a different area on the NMDS axis 2. In the Cambrian morphospace, each group occupies a distinct region of the space, with Cothonion and Sphenothallus close to Auroralumina in this elevation, with most Cambrian groups showing high positive values of NMDS axis 2 but with members occupying the full span of NMDS axis 1 (Fig. 5d).
The sum of ranges calculates the ranges of each axis occupied by the matrix subset and indicates the extent of the morphospace occupied by the subset 64,73 . Adding Auroralumina increases the mean sum of ranges of the Ediacaran tube morphospace to above that of the Cambrian tube morphospace (though IQRs of the Cambrian and Ediacaran including Auroralumina overlap). Again, this is also apparent from the NMDS plot, given that the position of Auroralumina lies far from the main occupation of Ediacaran tube morphospace (Fig. 5).
The median of centroids calculates the median of the distances between each row in the matrix (taxa) and the centroid of the matrix or subset-this indicates how tightly species are clustered around the centroid 74 . Adding Auroralumina slightly increases the mean median of the Ediacaran tube morphospace centroids, meaning that species are slightly less clustered around the centroid. This makes sense again given the position of Auroralumina far from the main group. In Cambrian morphospace, individuals are more clustered around the centroid-this is evident again from the NMDS plot as Cambrian taxa occur across the morphospace. The wider IQR of Cambrian morphospace is because the corners of the morphospace are in places occupied by only one or two taxa (for example, removal of Feiyanella, Cambroctoconus orientalis or Quadrapyrgites would all serve to contract the morphospace and shift the position of the centroid; Fig. 5).
NPMANOVA analyses can test the null hypothesis that the centroids and dispersions of groups are equal 75 . When conducted on these three subsets, the null hypothesis was statistically rejected for Cambrian versus Ediacaran (P = 0.45) and for Cambrian versus Ediacaran including Auroralumina versus Ediacaran subsets (P = 0.001). The null hypothesis was not rejected for Cambrian versus Ediacaran including Auroralumina (P = 0.75).
Reporting summary. Further information on research design is available in the Nature Research Reporting Summary linked to this article.

Data availability
All data analysed in this paper are available as part of the article, Extended Data Figs. 1-3 or Supplementary Data 1, 2 and 3. An RTI of the fossil is available at https://doi.org/10.5285/4c2f9f34-184d-43db-97e0-eecb13918375 (ref. 76 ). This published work and the nomenclatural acts it contains have been registered in ZooBank, the proposed online registration system for the International Code of Zoological Nomenclature. The ZooBank LSIDs (Life Science Identifiers) can be resolved and the associated information viewed through any standard web browser by appending the LSID to the prefix 'http://zoobank.org/' . The LSIDs for this publication are: urn:lsid:zoobank. org:act:869AF2A2-FB6B-44DF-88A1-D266B3D101F4

Code availability
The phylogenetic dataset, commands and constraints needed to run all analyses are included as Supplementary Data 1, 2 and 3 or part of the Supplementary Information. The data required to run the tube morphospace disparity analyses and the polyp size data are included in tabular form as Supplementary Data 1 and 2. The R code required to run the tube morphospace disparity analyses is included as a Supplementary Data 3.

March 2021
Corresponding author(s): Frances Dunn Last updated by author(s): Oct 5, 2022 Reporting Summary Nature Portfolio wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in reporting. For further information on Nature Portfolio policies, see our Editorial Policies and the Editorial Policy Checklist.

Statistics
For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.

n/a Confirmed
The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly The statistical test(s) used AND whether they are one-or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section.
A description of all covariates tested A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals) For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted
For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors and reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Portfolio guidelines for submitting code & software for further information.

Data
Policy information about availability of data All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: -Accession codes, unique identifiers, or web links for publicly available datasets -A description of any restrictions on data availability -For clinical datasets or third party data, please ensure that the statement adheres to our policy Provide your data availability statement here.
2 nature portfolio | reporting summary

March 2021
Field-specific reporting Please select the one below that is the best fit for your research. If you are not sure, read the appropriate sections before making your selection.

Life sciences Behavioural & social sciences Ecological, evolutionary & environmental sciences
For a reference copy of the document with all sections, see nature.com/documents/nr-reporting-summary-flat.pdf

Ecological, evolutionary & environmental sciences study design
All studies must disclose on these points even when the disclosure is negative.

Study description
The material described is a fossil specimen from the Ediacaran of the UK, which have been imaged using low-angle lighting (RTI available also) which has been included in morphological disparity and morphological phylogenetic analyses.

Research sample
The fossil material is described from the Ediacaran of the UK (Charnwood Forest, ~560 million years old). The morphological phylogenetic matrix was based on a previously published matrix (Zhao et al. 2019) and the disparity matrix was based on a previously compiled dataset (Selly et al. 2020

Data exclusions
Five taxa -Agalma, Keratoisidinae, Nephthyigorgia, Scleronepthya and Craseoa -were excluded from morphological phylogenetic analyses as they contained redundant information, following safe taxonomic reduction. Two further fossil taxa -Namacalathus, Eolympia -were excluded as they had previously been determined to have an extremely high (~95%) proportion of missing data (Namacalathus), the status of the genus was uncertain (Eolympia) (both following Zhao et al. 2019).
Reproducibility NA Randomization Taxa were allocated into groups for disparity analyses and in plotting polyp size following previously described taxonomic groupings. Our fossil specimen was determined to belong to its own, new species because it displays a character combination not observed in other taxa included in our phylogenetic matrix or from the Ediacaran fossil record in general.

Blinding
Blinding of specimens is not relevant to morphological study of fossil specimens or phylogenetic analysis.
Did the study involve field work?

Yes No
Field work, collection and transport

Field conditions
The specimen holotype remains in the field, but the plastotype was collected during a molding initiative of several outcrops in the Charnwood Forest area by British Geological Survey (BGS) staff. It is currently housed at the BGS Keyworth site and was the subject of study.

Location
The fossil is located in the Bed B assemblage of the Ediacaran of Charnwood Forest. The entire surface was molded and is now available at the British Geological Survey, Keyworth.