Palaeobiology and taphonomy of the rangeomorph Culmofrons plumosa

The deep marine Ediacaran fossil record of Avalonia is dominated by the Rangeomorpha, a clade characterized by up to four orders of fractal‐like branching. Despite their abundance, morphological diversity and the recent increase in Ediacaran studies, aspects of their palaeobiology, palaeoecology and phylogenetic position in the tree of life are still hotly debated. The clade has traditionally been interpreted as consisting of organisms that lived erect in the water column and tethered to the seafloor, based on the intuitive interpretation of their frondose body plan. However, recent work has challenged this view and instead proposes a reclining mode of life for several rangeomorphs, possibly in symbiosis with chemoautotrophic bacteria. Here, we offer a detailed description of exceptionally preserved specimens of Culmofrons plumosa from the Discovery UNESCO Global Geopark in Newfoundland, Canada. We suggest that Culmofrons plumosa should be reinterpreted as a reclining organism based on taphonomic and morphological evidence. Additionally, reproductive modes and a growth model of the species are here inferred, and they appear to be most consistent with a reclining mode of life, offering a novel palaeobiological reconstruction of the species.

Abstract: The deep marine Ediacaran fossil record of Avalonia is dominated by the Rangeomorpha, a clade characterized by up to four orders of fractal-like branching. Despite their abundance, morphological diversity and the recent increase in Ediacaran studies, aspects of their palaeobiology, palaeoecology and phylogenetic position in the tree of life are still hotly debated. The clade has traditionally been interpreted as consisting of organisms that lived erect in the water column and tethered to the seafloor, based on the intuitive interpretation of their frondose body plan. However, recent work has challenged this view and instead proposes a reclining mode of life for several rangeomorphs, possibly in symbiosis with chemoautotrophic bacteria. Here, we offer a detailed description of exceptionally preserved specimens of Culmofrons plumosa from the Discovery UNESCO Global Geopark in Newfoundland, Canada. We suggest that Culmofrons plumosa should be reinterpreted as a reclining organism based on taphonomic and morphological evidence. Additionally, reproductive modes and a growth model of the species are here inferred, and they appear to be most consistent with a reclining mode of life, offering a novel palaeobiological reconstruction of the species. a basal disc may be present (Brasier et al. 2012;Laflamme et al. 2012;Hawco et al. 2020). The clade is named after the genus Rangea, first described by G€ urich (1929, 1933) in Namibia. Rangeomorphs were later discovered in Charnwood Forest (UK) (Ford 1958) and later reported from several localities worldwide, including the famous Mistaken Point Ecological Reserve (MPER), in the Avalon Peninsula in Newfoundland (CA) (Misra 1969) and the Catalina Dome in the Discovery UNESCO Global Geopark, in the Bonavista Peninsula (Newfoundland, CA). The clade dominates the so-called 'Avalon biota' of the Ediacaran (Waggoner 2003;Boddy et al. 2022). The Rangeomorpha have been divided into: the Rangida, with double-sided firstorder branches; and the Charnida, with single-sided firstorder branches (Narbonne et al. 2009). Traditionally the Rangeomorpha have been interpreted as organisms living erect in the water column, filter-feeding or obtaining dissolved organic carbon (DOC) by osmotrophy (Laflamme et al. 2009), with the notable exceptions of the reclining epifaunal taxa Fractofusus misrai and F. andersoni (Gehling & Narbonne 2007).
This study focuses on exceptionally preserved specimens of the rangeomorph species Culmofrons plumosa from the MUN Surface, Catalina Dome (see Liu et al. 2016a; Fig. 1), which has generally been considered to belong to the Charnida (Laflamme et al. 2012). The monospecific genus Culmofrons is known only from the Ediacaran of Newfoundland and consists of a basal disc and stem attached to an oval frond with five or more first-order branches and a zig-zagged midline (Laflamme et al. 2012; Fig. 2A). First-order branches of Culmofrons are single-sided (sensu Narbonne et al. 2009), and are composed of displayed, sub-parallel second-order branches that in turn have displayed third and fourthorder branches (Laflamme et al. 2012). The type material comes from the Lower Mistaken Point Surface (LMP) at the MPER but is not as well preserved as specimens from the MUN Surface (Liu et al. 2016a).
In this work, we employ developmental and taxonomic statistical analyses, integrated with a new taphonomic model for the MUN Surface, to describe the palaeobiology, palaeoecology and ontogeny of Culmofrons plumosa. Newly discovered impressions beneath exceptionally preserved specimens are also investigated as potential reproductive structures.

GEOLOGICAL SETTING
The

Dataset compilation and statistical analyses
Twelve specimens of Culmofrons from the Bonavista and Avalon peninsulas (four figured herein), the holotype of Charnia masoni from the Charnwood Forest (UK) and the holotype of Beothukis mistakensis from the MPER (the only complete specimen of Beothukis mistakensis known; McIlroy et al. 2020; (Fig. 2C) have been studied morphometrically. As these taxa are protected and preserved in situ; silicone moulds of the fossils have been used to produce Jesmonite replicas for photography under controlled lighting. Both the silicone moulds and the Jesmonite casts of the four figured specimens are accessioned at The Rooms Corporation of Newfoundland and Labrador (NFM) (St. John's, NL), under the accession numbers NFM F-3972-3975. Morphometric traits and non-equidistant semi-landmarks outlining the frond outer perimeters were digitized from high quality pictures using imageJ and analysed in R (RStudio v1.2.5019; RStudio Team 2020). Data are available at Dryad Digital Repository (Pasinetti & McIlroy 2023).
Relationships between continuous variables were initially explored with regression analyses. A multivariate analysis of principal components (PCA) was run on scaled selected variables and the output was plotted along the major components, using the R packages factoMine v2.4 and FactoMineExtra v1.0.7 (Lê et al. 2008). Equidistant semi-landmark coordinates were computed using the package Stereomorph v1.6.4 (Olsen & Westneat 2015;Olsen 2017) and Procrustes analyses were run to obtain generalized coordinates. Principal components analyses were run on the coordinate dataset to characterize Culmofrons morphospace and were re-plotted in a backtransform morphospace to visually represent shape variations within the taxon (cf. method of Olsen 2017, fig. 1).

Retrodeformation
Retrodeformation is typically applied to Ediacaran fossils from the Avalon Assemblage. This technique consists of estimating the degree of metamorphic distortion of the fossiliferous surfaces and subsequently removing the distortion from digital images of the fossils. The degree of distortion is estimated based on direction of cleavage as well as the use of strain ellipses: holdfasts and discs (such as Aspidella) are typically assumed to have been originally circular in life, and their observable eccentricity is interpreted as proof of distortion. However, no Aspidella are preserved at the LMP locality and on the MUN Surface they are rare and show minimal distortion. Furthermore, there is no direct evidence for Culmofrons basal discs having been perfectly circular. As the assumptions necessary for linear models derived from different morphometric traits were largely satisfied, therefore no retrodeformation has been applied to the specimens in this study.

Morphological description of Culmofrons plumosa
The type material of Culmofrons plumosa is located at LMP locality in the MPER in the southern Avalon Peninsula of Newfoundland (Laflamme et al. 2012). Subsequent discoveries from the Bonavista Peninsula outcrops (here described, some illustrated in Fig. 4), while initially compared to Beothukis (Liu et al. 2016a), were later assigned to Culmofrons based on morphometric analysis (Hawco et al. 2020;McIlroy et al. 2020).
Culmofrons plumosa typically consists of a sub-elliptical frond that is basally tapered towards a parallel sided, sometimes curved stem that usually ends in a globose structure (sometimes referred as a 'holdfast' or 'basal disc'). Our descriptions follow the descriptive terminology proposed by Dunn et al. (2021): the terms first-order, second-order, third-order. . . refer to the hierarchical position of the branch in the fractal organization, while the adjectives primary, secondary, tertiary. . . refer to the ontogenetical order of branch formation. The frond is typically preserved as a negative impression here, we describe the branching patterns recorded as impression on the palaeo-seafloor, but the three-dimensional body plan of the organism can only be inferred. In the most complete specimens (Fig. 4B, C) at least nine first-order branches can be recognized, preserved in negative epirelief, with little to no evidence of branch overlap. The first-order branches are arranged with a glide plane symmetry, with branches alternating between the right ('d', dextral) and left sides ('s', sinistral) of a zig-zagged midline (Fig. 5). These first-order branches typically share their proximal margins with their respective precedent first-order branch, forming the midline. First-order branches d1 (the basalmost first-order branch) and s1 are located at the right and left margins of the frond, respectively. First-order branches d2 and s2 appear to originate respectively from the basal portion of the first-order branches s1 and d2 (cf. Figs 3B, 5A). No clear separation is visible between the impressions of the basal portion of s1 and d1 and the branches that originate from them, suggesting that successive branches were originating in a sympodial fashion (cf. Dunn et al. 2018). Alternatively, it is possible that firstorder branches arise from a central stalk, which could  have been above the preservational plane of the fossils. As s2 inserts immediately after the first second-order branch of d2 (d2.1) and shares with it a margin for the length of the first 4 s-order branches (d2.1-4), s2 and d2 show an apparent bilateral symmetry (actually glide-plane symmetry; see Fig. 5A), while s1 assumes a more distal position and might lose its connection with d2 with maturity. The successive first-order branches (d3, s3) insert at the proximal margin of the preceding and opposite first-order branch in proximity of, respectively, s2.7 and d3.5, wedging between the two precedent first-order branches and separating them. Branches increase in length towards the base of the frond with the exception of the two basal-most branches (d1 and s1), which are typically slightly shorter than the neighbouring branches. Second-order branches originate from the first-order branches with increasingly more acute angles in the basal-apical direction (cf. branch s2 in Fig. 5A). Second-order branches typically increase in size towards the midpoint of their first-order branch (cf. branches s2.1-s2.10 in Fig. 5A). Second-order branches typically have consistent sigmoidal shapes (e.g. Figs 4A-C, 5B) throughout a first-order branch, with little to no evidence of overlapping. The second-order elements of each first-order branch close to the margin are commonly sub-triangular (e.g. Fig. 5A, s1.7, s2.10). Most specimens do not preserve second-order F I G . 5 . A, interpretative drawing of a complete Culmofrons plumosa specimen. First-and second-order branches are labelled; each first-order branch has a different colour. B, detail of second-order branch s2.7, showing third-and fourth-order branches. Scale bars represent: 5 cm (A); 1 cm (B).
branching associated with the basal-most first-order branch on the right-hand side of the fossil (d1). Secondorder branches of the second first-order right branch (d2) may present secondary growth, extending behind the outer margin of the organism and growing above (and leaving an impression on the upper surface of) the most basal first-order branch on the right-hand side (d1) (impressions si in Fig. 4B).
The sigmoidal shape of the second-order branches is dictated by the arrangement of the third-order branches, which are typically arranged in alternating series on the left and right side of the second-order branches (cf. Fig. 5B). Proximally the left series is predominant, with the right series presenting increasingly bigger thirdorder branches distally, resulting in a glide-plane symmetry along the second-order branches axes in the central and distal portion of the second-order branches. The subtriangular apical-most second-order branches present only a radiating left series of third-order branches, with the biggest third-order branches at the centre of the series (e.g. Fig. 5A, s1.7, s2.10).
Fourth-order branching can be observed on both sides of the third-order branches, presenting the typical alternation of left and right series in a glide plane symmetry (cf. Figs 4, 5B).

Atypical rangeomorph structures
At least six of the best preserved Culmofrons (three of which figured herein) possess hitherto undescribed atypical rangeomorph structures below the tips of secondary branches located on first-order branches s2 and s3 (Figs 4,6). The impressions are all morphologically similar and share a homologous position within the frond (cf. Fig. 4A-C), they should thus be interpreted as functional biological structures rather than taphonomic artefacts. These atypical impressions resemble third-order branches that are rotated towards the tip of the frond and are clustered in a bundle-like arrangement (Fig. 6C).
Two third-order branches are arranged symmetrically at the centre of the bundle, originating from a common central position (Fig. 6C, blue and purple), with supplementary branches overgrowing the central branches on their distal margins (Fig. 6C, green). The structures are separated from the rest of the secondary branch via a constriction of the epithelium at the base of the bundle. Fourth-order branches are recognizable and appear to be rotated towards the centre of the bundles (Fig. 6C).
In NFM-F-3973, a third atypical structure can be recognized: impression aiii is sub-elliptical and underlies the third-order branches of the second-order branch d2.4 (Fig. 4A). This impression has no recognizable rangeomorph architecture and does not appear to have a developmental relationship with the secondary branch that hosts it, as it is in a central position in the secondorder branch d2.4, cutting across third-order branches.

Morphometric analysis
Linear regression. Explorative linear regressions were performed between selected variables: total length against maximum frond lengths (Fig. 7A) and total lengths against maximum frond widths (Fig. 7B). Additionally, maximum stem lengths (Fig. 7C) were tested against the total length of the specimens, then maximum frond lengths and frond widths were tested against each other (  (Fig. 7). Assumptions of linear relationship, independence, homoscedasticity and normality were satisfied for each of the linear models applied to continuous variables. However, introducing Beothukis to the dataset creates an outlier with high leverage: the linear regressions figured were therefore only applied to Culmofrons specimens. Higher R 2 values were recorded in all cases when the regressions did not include Beothukis mistakensis. Positive relationships can be observed in each of the computed regressions, suggesting an allometric growth model (all p-values << 0.05). Lower positive relationships and lower R 2 values are obtained when comparing widths of the specimens with total lengths (Fig. 7B), suggesting either faster growth rates in a basal-apical direction (in the stem and the frond length; Fig. 7A, C) than laterally (frond width; Fig. 7B, D) or a higher variability in frond length compared to frond width.
PCA. Multidimensional analysis of principal components allows for reduction of the dimensionality of a dataset comprising several continuous variables by linearly combine them into a new set of variables, called principal components (PCs).
PCA was compiled using the seven following continuous variables, chosen using the Kaiser-Meyer-Olkin test for sampling adequacy (Kaiser 1970): specimen total length, frond length, frond width, stem length, stem width, discs diameter, and length of the first left and first right first-order branches. For this study, only continuous variables consistently measurable in all the studied specimens were used, as consideration of categorical variables, such as descriptors typically used to describe rangeomorphs (e.g. furled/unfurled, rotated/unrotated; Brasier & Antcliffe 2009; Brasier et al. 2012), involves more subjective biological and taphonomic inferences, and might therefore be independent of taxonomy. Characters relating to branch organization (e.g. number of second-order branches) are not uniformly preserved across all the specimens, particularly second order and higher order branches and therefore only well-preserved specimens (12 Culmofrons plumosa specimens and the holotypes of Beothukis mistakensis and Charnia masoni) were included in the analyses. Continuous variables were scaled to have a mean of 0 and standard deviation of 1.
In both graphs, showing the ordination of the specimens in Dim-1 and Dim-2 (Fig. 9A) and Dim-2 and Dim-3 (Fig. 9B), Culmofrons plumosa specimen from the MUN Surface appear to be grouped together while the LMP and the Back Cove specimens lie at the margins.
The holotypes of Beothukis mistakensis and Charnia masoni consistently plot far from Culmofrons specimens Frond length (cm) Frond width (cm) F I G . 7 . Linear regressions between variable pairs. A, total length of the specimens and length of the fronds (y = 0.52589x + 0.79318, R 2 = 0.9059; p = 1.887 9 10 À6 ). B, total length of the specimens and width of the fronds (y = 0.20937x + 0.86703; p = 0.01745). C, total length of the specimens and length of the stems (y = 0.46796x À 0.68655, R 2 = 0.8826; p = 5.768 9 10 À6 ). D, length of the fronds and width of the fronds (y = 0.4228x + 0.2717, R 2 = 0.5564; p = 0.005341). A-D, morphometric data for the holotype of Beothukis (not included in the regression) is indicated with a blue data point; casts of MUN1-3 accessioned as NFM F-3973-5 respectively.
Backtransform morphospace analyses. To overcome the limitations of traditional PCA, we propose the use of generalized Procrustes analysis (GPA) to create a backtransform morphospace graph within which different specimen shapes can be plotted. A backtransform morphospace graph allows visualization of theoretical shape variability within a group of biological entities. In this study, Procrustes coordinates were obtained performing a GPA on a dataset of Culmofrons (traced from digital images) and transformed in a series of equidistant semi-landmarks. This approach (first used in Ediacaran taxonomy by Laflamme et al. (2007) to characterize Charnia species from Newfoundland) can be used to perform an ordination of the specimens based on similarity of their shapes scaled to the same centroid size, plotting them against a backtransform graph and disregarding variance due to specimen size, orientation and position. When GPA is performed on the Culmofrons dataset it shows that two components explain 94% of the observed variation in Culmofrons shape (PC1: 87%, PC2: 7%, Fig. 10). The biggest variation can be observed between the MUN Surface specimens, while the holotype and other material from the LMP Surface plot within the morphospace occupied by the MUN Surface specimens (Fig. 10).

Taphonomy of the Ediacaran biota of Avalonia
Ediacaran fossils of the Avalonian assemblage have been traditionally interpreted as the result of mouldic preservation of body impressions of the organisms between the seafloor and the smothering sediments ('death mask' model), in what is known as 'Conception-style' preservation (Gehling 1999;Narbonne 2005). This preservation style is the norm in the siliciclastic successions of Avalonia (cf. Liu 2016). To explain the differential preservations as positive (typically stems and basal discs) and negative (fronds) epireliefs often observed in a single specimen, Gehling (1999) initially proposed that some tissues remained intact until early lithification of the overlying ash cast them in positive epirelief. Delicate tissues, prone to faster decomposition, would only leave an impression by smothering the underlying microbial mat (producing negative epireliefs), as the three-dimensional form would decay before the lithification of the ash (Gehling 1999). Sturdier elements, such as stems, would take more time to decay, allowing the sediments above them to lithify and preserving their external mould. The poorly lithified sediments and the microbial mat underlying the stems of the organisms would have been pushed in the overlying mould after the stem decayed, preserving the mould as a positive epirelief. Authigenic mineralization of a microbial matgrounds growing above dead organisms may have played a crucial role in the death mask preservation of the fossil ( The mineralized veneer is immediately overlain by a normally-graded, 6 mm thick, fine-tuff ('4' in Fig. 11A) followed by a succession of thickly bedded coarse-grained cross-bedded sandstones.
We interpret that the ferruginous veneer represents the redox boundary separating oxic and anoxic conditions in the seafloor at the time of burial. The position of the redox boundary is controlled by microbial activity and F I G . 1 0 . Backtransform projection of the measured morphospace of Culmofrons plumosa, specimens are ordinated according to PC1 and PC2, obtained with a GPA. Casts of mun1-3 accessioned as NFM F-3973-5 respectively. could be found within a microbial matground or at the interface between the matground and the water column. If it is accepted that the ferruginous veneer reflects the original presence and position of microbial mats/death masks ('2' in Fig. 11B), this has potential implications for the taphonomy of Culmofrons on the MUN Surface.
Akin to the Gehling death mask model (Gehling 1999), it is likely that the frondose parts of the organisms decayed early after being buried by the ash (Fig. 11C), leaving an impression on the microbial matground, being quickly cast by the lithifying tuff/tuffite (see Matthews et al. 2020). The fine detail in the frondose portions of the organisms suggests that the organisms spent a considerable amount of time in contact with the microbial matground, allowing a deep impression of the fronds to form in the matground (cf. Fractofusus, Beothukis and Charnia). It is unlikely that the veneer was precipitated on the top of the Culmofrons fronds: if that was the case, we would have to assume that there was a period of time during which the buried frond decomposed, leaving an impression on the underlying silt, which would have been later transferred to the veneer. This would imply a collapse of the veneer into the impressions left on the siltstone and a subsequent loss of resolution. Moreover, we hypothesize that Culmofrons, a multicellular opisthokont, probably of metazoan grade, would not have been able to survive entirely below the redox layer, but was more likely to have lived at the interface, with the upper portion of the frond exposed to oxic waters.
Since the ferruginous veneer covers positive features of fossils, such as Culmofrons stems on the MUN surface (cf. Fig. 11B), and those features are often preserved in great detail, without any evidence for mat tearing or displacement, it is possible that it reflects matground growth over the tissue of the stem during the life of the organism, rather than having been displaced after the decay of the stem as the Gehling (1999) model suggests. Notably, the arboreomorph Charniodiscus procerus is also inferred to have lived with the stem buried underneath the matground, sometimes even with Fractofusus specimens growing on top of the stem (P erez-Pinedo et al. 2022).
The resulting fossil (Fig. 11D), a combination of negative impressions and positive epireliefs, is likely to be the result of the preservation of an organism living partially buried under the matground (stem portion) and partially above it, smothering the matground and exchanging oxygen and nutrients with the water column above and porewater system below (frondose portion). This is consistent with the presence of abundant preserved filaments on the surface (Liu & Dunn 2020), which are preserved as positive epireliefs on the veneer, but are interrupted where fossil impressions are present. It is therefore unlikely that the fronds fell above the filaments, which would have been otherwise preserved in positive epirelief under Gehling's model, but rather pre-existing fronds would have prevented the filaments from being in contact with the matground, precluding their preservation.

Culmofrons systematics
PCA consistently groups together Culmofrons specimens from the MUN Surface, leaving specimens from LMP and Back Cove consistently at the margins of the ordination (Fig. 9). However, the result of the GPA (shape analysis, Fig. 10) suggests that all of the Culmofrons specimens studied occupy a very similar morphospace and have a similar broad frond profile. We suggest that all the F I G . 1 1 . A, thin section of the MUN surface. B, Culmofrons plumosa in life, with the frond (5) reclining of top of the microbial matground and the stem (6) underneath the matground. C, Cu. plumosa buried underneath ash that would eventually turn into tuff (4). D, Cu. plumosa preserved as a negative impression (-) and positive epirelief (+) on a mineralized veneer (3). Numbered features: 1, siltstone; 2, hemipelagite; 3, mineralized veneer; 4, normally graded tuff; 5, frond; 6, stem. Scale bar in A represents 1 cm. studied specimens could therefore be classified within the same species and the observed variation can be partially explained by ontogeny. As the holotype plots well within the species morphospace yet presents a much greater size than the rest of the specimens, we interpret it to be a super-mature specimen. We hypothesize that the MUN Surface population represented a group of organisms of a similar age class, due to their low size variability, as shown by the PCA ordinations (Fig. 9).
We note that the absence of proper outgroups and the small size of the database preclude accurate multidimensional analyses, and result in different clustering possibilities for the specimens assigned to Culmofrons. Bigger datasets of well-preserved material will be necessary to properly address systematics within the Charnida.
PCA shows that Beothukis mistakensis and Culmofrons plumosa consistently plot away from each other. Including the holotype of Charnia masoni in the PCA ordinations results in Ch. masoni plotting even further away from both Cu. plumosa and B. mistakensis. It is also important to note other major differences between the two Newfoundland species and Charnia, such as presence of rotated second-order branches and absence of fourthorder branches in the latter, are not recorded in the ordination.
Even though it is not possible to draw taxonomic conclusions based on a small number of specimens, our results suggest that B. mistakensis and Cu. plumosa might represent, together with other yet undescribed stemmed rangeomorphs from the Bonavista Peninsula, a monophyletic group within Charnida.
In our view, this result suggests that the traditional classification of Cu. plumosa (Laflamme et al. 2012) andB. mistakensis (Narbonne et al. 2009) as belonging the rangeomorph clade Charnida might need review and the possibility of erecting a new rangeomorph clade should be investigated, whilst highlighting the necessity of a taxonomic revision of the major rangeomorph groups.
Contributions to the PCs from variables related to the stems and discs (which are absent in Charnia and Beothukis) (Fig. 8) are important but limited. Excluding the variables related to stem and disc (stem length, stem diameter and disc diameter) (not figured) still results in Charnia and Beothukis plotting outside the Culmofrons space, which suggests that the three taxa differ substantially in frond Bauplan.

Palaeobiology: reclining lifestyle
The Rangeomorpha and the coeval (possibly related) clade Arboreomorpha have been historically interpreted as living erect in the water column and obtaining organic carbon via filter-feeding on particulate organic carbon (POC) or osmotrophically absorbing DOC (Narbonne 2004;Laflamme & Narbonne 2008;Laflamme et al. 2009). This notion had been implied in the reconstruction of the Ediacaran ecosystems and is a precept that underpins ecological and tiering models proposed for the biota (Clapham & Narbonne 2002;Darroch et al. 2013;Mitchell & Kenchington 2018). However, based on a lack of direct evidence for rangeomorphs having lived erect in the water column, it is necessary to assume the null hypothesis that they lived in the orientation that we see them preserved for at least part of their life cycle (i.e. reclining on the seafloor; McIlroy et al. 2021). It has also been suggested (Dufour & McIlroy 2017;McIlroy et al. 2021) that ectoor endo-symbiotic relationships with chemosynthetic bacteria would have provided the rangeomorphs with a more reliable source of organic carbon than filter-feeding or osmotrophy, questioning whether osmotrophy is a realistic exclusive feeding strategy for large organisms living in seawater ( In particular, Charniodiscus procerus, an Arboreomorph from MPER which has both positive (stem and holdfast) and negative (frond) epirelief preservation, has been shown to have lived recumbent on the seafloor, with the stem covered by the microbial matground and the frond exposed to the water column (P erez-Pinedo et al. 2022).
Here, we also propose a reclining lifestyle for the rangeomorph Culmofrons plumosa. This can be inferred from the taphonomy of the MUN Surface: as the mineralized veneer is uninterrupted in preserving negative impressions as well as coating positive epirelief features, a certain amount of time would have been required for the microbial growth to colonize the top of positively preserved structures. Laflamme et al. (2011) suggested that organisms can be preserved as positive epireliefs if they were already felled on the seafloor before burial if the overlying layer lithified before the decomposition of the stem. This is also the first step towards the formation of diffuse ivesheadiomorphs (Liu et al. 2011). The classic death mask model (Gehling 1999) does not fully account for preservation of taxa by a single mineralized veneer that both drapes the stem and underlies the frond as it does in Culmofrons from the MUN surface.
Structures observed in positive epirelief typically include the most basal dextral first-order branch of each specimen, which usually does not show evidence of second or higher order branching (Fig. 5). Since evidence of well-preserved secondary growth from more apical firstorder branches can be observed as a negative impression atop the positively preserved basal dextral branches (Fig. 4B, si), it can be inferred that the organisms were still alive when the positively preserved structures were covered by microbial growth (preserves as the ferruginous veneer) and before secondary growth occurred.
The specimens described herein are considered to be thin-bodied reclining organisms with branches at a level slightly below the seafloor, or possibly even living partially buried in the sediment with furled tips of the branches partially extending into the water column. It is possible that first formed parts of the organism, such as basal discs and stems may have been covered by a microbial mat for much of lifespan of the Culmofrons organism.

Palaeobiology: development and growth
A well supported rangeomorph growth model for Charnia masoni, the type species of the Charnida, has been recently proposed (Dunn et al. 2019(Dunn et al. , 2021. That research suggests that Charnia grew by apical addition of firstorder sigmoid-shaped branches that are arranged with a glide-plane symmetry, expanding by inflation after the specimen had reached a certain number of branches (see Dunn et al. 2021, fig. 4). The frond of Charnia can thus be interpreted as a series of first-order branches that successively originated from the tip of an axial branch. First-order branches in Charnia typically have the same number of second-order branches, suggesting that they originated early in the growth of new first-order branches and then became larger through ontogeny by a process of inflationary growth (Dunn et al. 2021).
While homologies and phylogenetic relationships between Charnia and Culmofrons are yet unclear, a similar growth program can be inferred for Culmofrons. Firstorder branches in Culmofrons all have similar numbers of second-order branches (10-12) (Fig. 12B) but, unlike Charnia (Dunn et al. 2021), these second-order branches reach a maximum size near the midpoint of the firstorder branches (Fig. 5). This suggests that second-order branches were added early during the ontogeny and consequently inflationary growth occurred progressively. However, the most basal (i.e. oldest) first-order branches on both sides of the axis (d1 and s1) are usually slightly smaller than the younger first-order branches (Fig. 12A), suggesting an allometric development with increased inflationary growth of the second-order branches at the midpoint along the length of the first-order branches, perhaps once the specimens reach maturity. The most basal first-order branches are consistently on the right side of the specimens (in Ch. masoni we note that it is most commonly the left side); they do not demonstrate higher order rangeomorph branching and they can be overlain by secondary growth structures originating from the adjacent (more apical) branch (Fig. 4B). It is therefore possible that the oldest first-order branches of Culmofrons were eventually subject to die-off during the life of the organisms and offered a substrate for secondary growth, with potential reabsorption of the disused structures. Alternatively, the basal-most non-rangeomorph branches might represent a support structure or the remainder of a generative region involved in the organism development. The consistency of the first branch being on the dextral side of Culmofrons in all of the analysed specimens (and the sinistral side of Charnia) supports a reclining mode of life for both taxa (if they were erect taxa that fell to the substrate upon death/burial it seems unlikely that they would always fall the same way up; see McIlroy et al. 2021). Dunn et al. (2021) found evidence of interconnection between each first-order branch and the branch in an opposite and immediately more basal position. The two most basal first-order branches in Culmofrons are separated by the stem and are thus separate from one another (at least at the surface of the organism-they may have been connected within the stem), however, all successive first-order branches appear to be connected at their basal regions with the immediately opposing branch (e.g. Fig. 5, branches d1 and s1). This suggests that, like Charnia, first-order branches in Culmofrons originate in succession from a basal and proximal portion of their respectively opposite and immediately more basal first-order branch. This is coincident with the first and second second-order branches, separating and moving distally after the fifth or sixth second-order branches as younger first-order branches develop, inflate and occupy the space between them (e.g. Fig. 6, d3 inserting between s2 and d2 in proximity of s2.6). First-order branches are therefore likely to have originated from a specific 'generative' area at the distal apex of the frond, located between the two youngest first-order branches of each specimen.
It is also possible that second-order branches within each first-order branch originated from the apical portion of the respective first-order branches, but due to the preservation of rangeomorph branching as negative impressions, it is not possible to determine three-dimensional morphology to assess interconnectivity between adjacent second-order branches. Second-order branches develop allometrically, with the basal-most second-order branches being smallest, and inflationary growth being most developed in the medial positions along the first-order branch. Additionally, second-order branches may be triangular rather than sigmoidal, particularly towards the frond margin where the second-order branch morphology appears to be modified to fill any space between adjacent firstorder branches.
We suggest here that the position of Culmofrons and Beothukis within Charnida should be reviewed in future works, as it is difficult to assess branching homologies between the two genera and the type species of the family Charnia. Additionally, secondary growth of the secondorder branches can be recognized in some Culmofrons specimens (Fig. 4B), suggesting that Culmofrons was more F I G . 1 2 . A, lengths (cm) of first order branches according to their ontogenetical position (d1: oldest -s3: youngest) for each specimen. B, number of second-order branches for each first-order branch according to their ontogenetical position (d1: oldest -s3: youngest) for each specimen. Note that, on both plots, some values are missing as the branches are not preserved in the fossils; d1 typically does not preserve second-order branches. Casts of MUN2-3 accessioned as NFM F-3974-5 respectively. morphologically variable (similarly to the Rangida Bradgatia linfordensis) than even the very large super-mature Charnia masoni (grandis type) in which morphology is strongly conserved throughout growth with no evidence of secondary growth (Dunn et al. 2021).
Palaeobiology: reproduction in Rangeomorpha and reproductive structures in Cu. plumosa The reproductive strategies of the Rangeomorpha are not entirely understood and a consensus has not been reached on the prevalent reproductive strategies adopted within the clade, in part due to the scarcity of fossil evidence of the earliest life stages of the group (Liu et al. 2013). It has been suggested that the Rangeomorpha reproduced seasonally and sexually, based on the size-frequency distributions of specimens of the reclining rangeomorph Fractofusus misrai on the D and E surfaces at MPER (Darroch et al. 2013). Based on spatial analyses of the D, E and LMP surfaces at MPER, Mitchell & Kenchington 2018 suggested that efficiency of dispersal of propagules was the main driver in the evolution of Ediacaran ecosystems of stemmed taxa, rather than competition for resources in a tiering model, as previously proposed by Clapham & Narbonne (2002). It has also been hypothesized, based on analysis of spatial distributions, that at least one taxon (Fractofusus andersoni) reproduced asexually via stolons, resulting in aggregation of smaller specimens around the parent organism, which may have had a secondary dispersal stage (Mitchell et al. 2015). Liu & Dunn (2020) observed filaments interconnecting different specimens, suggesting that those structures could have been involved in stoloniferous reproduction. However, filaments can sometimes connect specimens of seemingly similar age classes or specimens belonging to different species (Liu & Dunn 2020), which weakens this hypothesis. The resulting corpus of literature proposes several different possible reproductive strategies, both sexual and asexual, many of which are based on mathematical models, while direct fossil evidence for reproductive strategies remain scarce.
Aberrant structures in the multifoliate rangeomorph Hylaecullulus fordi from Charnwood Forest (UK) are interpreted as over-compensatory damage response (Kenchington et al. 2018). Those structures are preserved on the same plane as the rest of the organism, in continuity with the normal branches and displacing neighbouring branches, suggesting contemporaneity. The structures in H. fordi appear to be reverting to lower order branches, demonstrating the truly modular nature of the Rangeomorpha and the ability of individual functional modules to grow independently. In contrast, at least six Culmofrons specimens from the MUN Surface, three of which  are figured herein (Fig. 4), show systematically distributed anomalous structures that occupy homologous positions in well-preserved specimens. These structures are continuous with normal branches, do not extend beyond the associated second-order branches (Fig. 6) and they are inserted within (beneath) secondorder branches without displacing the neighbouring third-order branches. It is thus unlikely that the observed structures in Culmofrons represent damage repair or overcompensatory secondary growth: the strong morphological resemblance and positioning of bi bi and bii, and ci and cii with their counterparts aii and aii suggest a functional interpretation rather than an accidental occurrence (Fig. 4). The exception to this is impression aiii, which is not terminally placed (Fig. 4A) and does not have rangeomorph branching, meaning that it might either: (1) not be part of the associated Culmofrons and could be an unrelated taxon; (2) represent damage repair; or (3) represent a taphonomic artefact.
We suggest that the unusual rangeomorph impressions observed in Culmofrons could represent bundles of thirdorder branches in the process of separating from the organism, in a process akin to asexual reproduction.
Comparison with placozoan reproductive structures. Placozoans have a simple body plan consisting of two epithelia bounded by mesenchyme (some with symbionts; Gruber-Vodicka et al. 2019) and have two reproductive strategies: binary fission and budding. Placozoan binary fission involves creating a division origin followed by separation into two sister organisms (Pearse 1989; Zuccolotto-Arellano & Cuervo-Gonz alez 2020). This differs from Culmofrons, in which the structures appear at higher orders of branching; this may represent a true reproductive process, resulting in the production of daughter organisms. Placozoa can also reproduce by budding, resulting in motile larvae with two undifferentiated cell layers (Thiemann & Ruthmann 1990).
Unlike placozoan modes of reproduction, Culmofrons appears to produce complex structures, possessing preestablished rangeomorph architecture differentiated from a single functional unit (second-order branch) rather than by binary fission. The complexity of the structures observed in Culmofrons may provide evidence for a life cycle with several morphologically distinct phases (Brasier & Antcliffe 2004). The generation of a fundamental reproductive module is an effective reproductive strategy for rapid establishment of new growth and reduced genetic requirements in that each reproductive unit, even though composed of third-order branches (and therefore functionally a portion of a second-order branch) has the potential to generate an entire frond. The potential ability of bundles of third-order branches to revert into firstorder branches and establish a new organism is consistent with the growth model proposed for Cu. plumosa above. Employment of modular reproductive structures has a further advantage in that each element in a rangeomorph organism could potentially become a new individual, allowing for efficient dispersal. Hydrozoans can reproduce asexually by developing juvenile organisms as evaginations of the endoderm and the ectoderm (Technau & Steele 2012). Hydrozoan buds develop from the side of the body of the mother organism and get displaced basally as they mature and start to separate (Bode et al. 1973;Shostak 2018). Separation of the modular structures (impressions ai-aiii, bi-bii, ci-cii) might have evolved to overcome the anatomical impediment to metazoan-like budding imposed by the inferred thick epithelium of the Rangeomorpha (Butterfield 2020). Weak constricted connections between the third-order branches of the observed structures and associated secondary branches might be a convenient way to separate the mature reproductive modules.
The morphology of Hylaecullulus fordi has led to the proposal of two means to evolve modularity, either by: (1) greater integration (cf. octocorallian coloniality); or (2) by relaxation of integration (cf. the green algae Caulerpa prolifera) (Kenchington et al. 2018) to generate new stems. Our discovery of abnormal structures in Culmofrons, if their reproductive function was confirmed, would suggest that rangeomorphs might have had true modularity and the ability to separate modules as an asexual reproductive strategy (see Brasier & Antcliffe 2004). This would suggest that multifoliate rangeomorphs should not be interpreted as colonial organisms, and that their modularity was probably achieved by relaxation of integration via the organization in higher order branches of the fundamental functional unit, the second-order branches.
The employment of secondary asexual reproductive strategies such as budding and binary fission have been documented as a response to starvation periods in the Cnidaria (Technau & Steele 2012). It is possible that the seafloor where Culmofrons reclined was subject to episodic smothering by thin layers of sediment. In such a setting the generation of upward growing budding structures into the overlying defaunated seafloor would have provided a viable reproductive strategy for repopulation of slowly aggrading seafloors (Kenchington & Wilby 2014). Further identification of reproductive structures in wellpreserved specimens of Culmofrons plumosa and other rangeomorphs will be necessary to positively interpret structures such as the impressions ai, aii, aii, bi, bii, ci and cii as reproductive modules.

CONCLUSION
Recently discovered specimens of Culmofrons plumosa from the Bonavista Peninsula reveal exquisite details of rangeomorph branching, revealing hitherto undocumented structures and providing insights into rangeomorph palaeobiology. Taphonomic observations on the Catalina Dome material suggest a reclining lifestyle for Cu. plumosa, possibly hosting symbionts as has recently been described for Beothukis mistakensis (McIlroy et al. 2020(McIlroy et al. , 2021. A morphological description of the new material and morphometric analyses support the validity of the genus Culmofrons and its differentiation from the genera Beothukis and Charnia. The species Culmofrons plumosa reveals a high morphometric variability, but a rather conserved morphology and a deterministic growth plan. Three specimens present a total of seven abnormal structures, six of which have been recognized as potential reproductive structures. These six impressions resemble bundled tertiary order branches and can thus be differentiated from over-compensatory growth or secondary growth as is observed in Hylaecullulus, Beothukis (Kenchington et al. 2018;McIlroy et al. 2020) and some Culmofrons (Fig. 2B).
Based on the similarity of the structures with the secondary growth tips observed in Bradgatia (Brasier & Antcliffe 2009) and on the growth models proposed for Charnia (Dunn et al. 2018) and Beothukis (McIlroy et al. 2020) we suggest that the structures observed in Culmofrons might represent modular reproductive structures, possibly as an adaptation to sediment smothering events or nutrient crises. Comparison of the newly described impressions within Culmofrons with the reproductive strategies of extant basal metazoans suggests that the Rangeomorpha were non-colonial and did not have a metazoan Bauplan, and that rangeomorph branching might represent true modularity. We also infer a growth model for Cu. plumosa and we suggest that, similar to Ch. masoni, the species grew by early addition of firstorder, primary branches and later inflation.
Acknowledgements. Fossil surfaces in the Bonavista Peninsula are protected under Reg. 67/11, of the Historic Resources Act 2011 and their access is only allowed under permission from the Government of Newfoundland and Labrador. Fieldwork near Port Union was conducted under permit from the Government of Newfoundland and Labrador. Thanks are due to: Rod S Taylor, for his great contributions in the field, and in the conception and editing of this manuscript; Edith and Neville Samson for their kind support; J Matthews and A Liu for discovery of the MUN Surface. P Thurlow, H Fitzgerald, C McKean, JM Neville, D P erez-Pinedo, C Locatelli and B Smith for field assistance; and Natural Sciences and Engineering Research Council of Canada (NSERC) for Discovery Grant (individual) and DAG funding to DM. A Liu and an anonymous referee commented on an earlier draft of this manuscript.