Benthic microbial structures (stromatolites and other microbialites) provide conspicuous evidence of pre-Phanerozoic life, but are difficult to understand because they rarely preserve recognizable evidence of the organisms involved. Stromatolitologists have struggled for over a century to decipher their microscopic laminae and clots, which are assumed to have been produced or influenced by in vivo and/or post-mortem biogeochemical activity, and to formalize the ‘taxonomy’ of their morphology and microstructure5,11.

The existence of metazoans by the Ediacaran period (the last period of the Neoproterozoic) is indicated by bilaterian ‘body’ and trace fossils12, and geochemical evidence (biomarkers)13 provides disputed14,15, indirect evidence for Cryogenian poriferans. The search for definitive physical evidence of pre-Cryogenian metazoans is confounded by uncertainty about what to look for, but preserved physical evidence should be small, subtle and possibly altogether unfamiliar. Given that sponges are the most basic of known animals2,3, physical evidence of Neoproterozoic sponges could be sought, but effort focused on the characteristics of mineralized sponge skeletons (siliceous or calcareous spicules)16,17,18 overlooks sponges with only proteinaceous (spongin or keratin19,20) skeletons. Early metazoan evidence might instead resemble taphonomic (preservational) products of sponge soft tissue21,22,23 rather than mineralized sponge skeletal components. Although molecular clock data suggest that sponges emerged in the early Neoproterozoic1, the oldest undisputed sponge body fossils are from the Cambrian period15.

Recent work6,7 has shown that vermiform microstructure4,5—an unusual microscopic feature in Phanerozoic reefs and stromatolites that was initially interpreted as being related to algae4 or protozoans24,25—is instead a keratose sponge body fossil comprising complexly anastomosing cement-filled microtubules enclosed in carbonate microspar. It is produced taphonomically6,10 in nonspicular keratose demosponges through post-mortem calcification of soft tissue to produce carbonate microspar (automicrite), which surrounds the tough spongin fibres of the ‘skeleton’ of the sponge. Decay of the spongin then produces a network of complexly anastomosing tubular moulds that eventually become passively filled with sparry calcite cement. Although questioned26, the association between vermiform microstructure and sponges has been confirmed in undisputed body fossils of Phanerozoic sponges7. Three-dimensional reconstruction of vermiform microstructure has shown that tubule shape and branching configuration are too consistent and complex to be abiogenic (for example, compacted peloids), do not resemble the branching style of other possible organism types (microbial or fungal)6 and are identical to the spongin meshworks of keratose sponges6. Although the existence of Proterozoic vermiform microstructure has been predicted6,10,27, published examples are rare28,29 and difficult to understand.

The calcification of decaying sponge soft tissue has been documented in modern sponges21,22, and produces sponge ‘mummies’22 as well as a range of subtle carbonate sedimentary textures (such as peloid clusters) in living and Phanerozoic fossil sponges7,9,21,22. Taphonomic sediment textures (polymuds) that may be poriferan-related have previously been identified in the reefs that are the subject of this Article23.


This petrographic study presents possible evidence of sponge body fossils in thin sections (30-μm-thick rock slices, viewed microscopically in transmitted light) from the approximately 890-million-year-old (Ma)30,31 Little Dal reefs (Stone Knife Formation32, northwestern Canada) (Fig. 1a, b). These large (about 500 m in thickness, and kilometres in diameter) microbial reefs33,34,35,36 were built mainly by variably preserved calcimicrobes that have been interpreted as filamentous cyanobacteria (photosynthesizers)33,35,36, and developed palaeotopographic relief of up to about 100 m above the surrounding subphotic, level-bottom carbonate-mud seafloor. Reef framework, which is generally not discernible in natural exposures, was documented from slabbed hand samples and thin sections35.

Fig. 1: Geographic and stratigraphic location of the study material.
figure 1

a, Location of Little Dal reefs in northwestern Canada. Scale bar, 100 km. b, Stratigraphic position of the Little Dal reefs in the Stone Knife Formation (Fm)32; depositional age is known through litho- and chemostratigraphic correlation with the 892 ± 13-Ma (Re–Os black shale)30 Boot Inlet Formation31 (Shaler Supergroup (Sg)), together with other geochronological constraints50,51. Gp, group. c, Reef growth stages34,35, simplified summary of framework morphologies35 and off-reef cycles34.

The reefs grew in five stages (Fig. 1c), each with different microbialite morphologies: anastomosing millimetre-to-centimetre-scale masses with no consistent shape (stages I–III); centimetre-scale anastomosing columns and digits (stages II, III and V); and steep sheet-like masses at a scale of decimetres to 10 m (stage IV). Stage V includes cement-rich to micritic, domical, turbinate and columnar stromatolites that generally lack calcimicrobes, with associated ooids and stromaclasts. Microbialites of stage I to IV grew predominantly in moderate-energy, illuminated palaeoenvironments33,35,36, but stage V records a shallow-water, high-energy environment. The microstructure that forms most of the reefal microbialites (especially in stages III and IV) comprises filaments that are about 10 μm in diameter, separated by 10–100-μm masses of marine cement that probably represents the calcified sheath polysaccharide of cyanobacteria33,36 that was permineralized during microbialite growth. The reef framework, consisting predominantly of this microstructure and its taphonomically degraded equivalents36, defines primary void networks (millimetres to centimetres in size) that are commonly floored with geopetal carbonate mud and lined by isopachous, fibrous marine calcite cement. The relative timing of void-filling by marine cement precipitation versus geopetal sediment accumulation is variable, attesting to the very early timing of marine cement precipitation.


Vermiform microstructure in samples from stages II, III and V of the Little Dal reefs (Fig. 1c) is identifiable only in rare thin sections, in which it forms millimetre-to-centimetre-scale masses of anastomosing tubes that are filled with calcite spar and surrounded by calcite microspar groundmass (Fig. 2a, b). The approximately 20–30-μm-wide tubules have complex, divergent branching and rejoining at a spacing of about 30–100 μm, form very irregular three-dimensional polygonal meshworks, are defined by enclosing microspar, lack walls, and are filled with clear, equant calcite crystals up to 20 μm wide (Fig. 2b). The homogeneous microspar groundmass that encloses the tubules comprises cloudy, equant, interlocking calcite crystals of about 2–8 μm wide, and differs texturally and compositionally from other fine-grained reefal carbonate in its uniform crystal size, lack of sedimentary texture, and dearth of detrital silicate impurities. Vermiform microstructure preservation is good to barely discernible.

Fig. 2: Characteristics and distribution of Little Dal vermiform microstructure in stratigraphically oriented 30-μm-thick thin sections.
figure 2

a, Well-preserved vermiform microstructure exhibits a polygonal meshwork of anastomosing, slightly curved, approximately 30-μm-diameter tubules embedded in calcite microspar (KEC25). Scale bar, 500 μm. b, Enlarged rectangle from a, showing branching tubules forming three-dimensional polygons intersected at various angles by the thin section; clear calcite crystals, about 10–20 μm in width, fill tubules in groundmass of more finely crystalline calcite (dark grey). Scale bar, 50 μm. c, Three-dimensional fragment of spongin skeleton from a modern keratosan sponge, illustrating its branching and anastomosing network of fibres (incident light). Scale bars, 100 μm (main panel), 20 μm (inset). d, Vermiform microstructure in debris that includes calcimicrobialite and other reef-derived clasts (C) flanking reef stage III (MV63). Scale bars, 1 mm (main panel), 100 μm (inset). e, Vermiform microstructure in shelter pore beneath microbialite clast, in detrital sediment occupying a reef-top depression; pore is thinly lined with marine calcite cement (MC) (indicated with an arrow), and partly filled with geopetal carbonate sediment (GCS) (KEC25; stage-III reef core). Rectangle is enlarged in a. Scale bar, 1 mm. f, Vermiform microstructure in a microbialite (M) framework void is overlain by pore-occluding marine calcite cement; circled area indicates moderately well-preserved tubule meshwork (DL32a; reef stage II; detailed characteristics depicted in Extended Data Fig. 1). Scale bar, 1 mm. g, Patches of vermiform microstructure in various relationships with micritic microbialite masses (white dashed outline) and detrital carbonate sediment (CS) (KES23; resedimented stage-II reef clast). Rectangles are enlarged in ik. Scale bar, 1 mm. h, Simplified depiction of relationships among vermiform microstructure, microbialite masses and detrital carbonate sediment in g. Scale bar, 1 mm. i, Vermiform microstructure mingled with microbial micrite within a microbialite digit (enlarged from g). Scale bar, 500 μm. j, Vermiform-microstructured mass within sediment between microbialite digits; also contains diagenetic dolomite patches (D) (enlarged from g). Scale bar, 500 μm. k, Vermiform-microstructured mass adhering to the margin of microbialite digit (enlarged from g). Scale bar, 500 μm. All images except c are in plane-polarized transmitted light. Samples from resedimented reef debris are depicted in depositional orientation based on geopetal structures. Reef locations and abbreviations (such as KEC) are described in a previous publication35. Larger versions of vermiform microstructure photomicrographs are provided in Extended Data Figs. 15.

Vermiform microstructure is present in three microfacies (i, ii and iii (the last divided into iiia and iiib subsets)), representing three palaeoenvironments (Figs. 2, 3). It is not present in calcimicrobe-dominated stromatolites or level-bottom carbonate mudstone that is distal to reefs.

Fig. 3: Palaeoenvironments occupied by the Little Dal vermiform microstructure interpreted as possible body fossils of keratose sponges.
figure 3

The organism lived (i) on poorly illuminated to non-illuminated carbonate mud surfaces in depressions on the reef surface and on debris aprons mantling reef flanks; (ii) in voids produced by the growth of the complex microbial framework of the reef; and (iii) interlayered with non-calcimicrobial microbialites (cement-rich and muddy-laminated stromatolites in high-energy reef-capping phase V; irregularly muddy-laminated to clotted microbialites in moderate-energy environments of reef stages II and III). Scale bars, 5 mm.

In microfacies i (Fig. 2d, e, g, j, Extended Data Figs. 1, 2), vermiform microstructure is intercalated with carbonate mud (with or without larger reef-derived clasts and terrigenous impurities) in (1) synsedimentary debris flanking reef stages II and III, and (2) millimetre-to-metre-scale palaeodepressions on reef growth surfaces (stages II and III). It locally encrusts sides of reef-framework clasts in sediment, extends into crevices, and occupies shelter porosity under clasts (Fig. 2e).

In microfacies ii (Fig. 2f, Extended Data Fig. 3), vermiform microstructure occupies millimetre-to-centimetre-scale framework voids of reef stages II and III. Void-filling vermiform microstructure of microfacies i and ii either underlies (Fig. 2f) or overlies (Fig. 2e) isopachous void-lining marine cement.

In rare microfacies iiia (Fig. 2g–k, Extended Data Fig. 4), vermiform microstructure encrusts non-calcimicrobial microbialite columns and mingles with irregular muddy microbialite microstructure of reef stages II and III. In rare microfacies iiib (Extended Data Fig. 5), vermiform microstructure is sub-millimetrically interlayered with stage-V non-calcimicrobial stromatolites where it locally passes laterally to geopetal peloid accumulations in lenticular voids.


The shape, size, branching style and polygonal meshworks of the Little Dal vermiform tubules closely resemble both spongin fibre networks of modern keratosan sponges (Fig. 2a–c) and vermiform microstructure either demonstrated or interpreted to be sponge-derived in diverse Phanerozoic microbial, reefal and non-reefal carbonate rocks6,7,8,10,24,25,27,37. The compositional and textural homogeneity of the microspar groundmass supports an origin through permineralization of a pre-existing biological substance9, rather than incremental accumulation of detrital sediment or microbial carbonate that passively incorporated complexly anastomosing tubular microfossils. Variable preservation and association with geopetal peloid accumulations are familiar aspects of Phanerozoic sponge taphonomy9,21,22,38. In previous work, detailed comparison of the three-dimensional characteristics of vermiform microstructure with branching cylindrical organism types yielded no convincing alternative to the sponge interpretation6.

The preference of Little Dal vermiform microstructure for environments that were not inhabited by photosynthetic calcimicrobes (reef flanks, depressions on active reef growth surface, and framework and shelter voids), versus its absence from filamentous calcimicrobial reef-framework components, suggests that (1) illumination may not have been a requirement and (2) the organism may have been unable to compete with reef-building photosynthesizers that grew and/or calcified rapidly. The interlayering of vermiform microstructure with calcimicrobe-free microbialite (microfacies iiib) in the high-energy, well-illuminated reef surfaces of reef stage V supports the interpretation that the vermiform-microstructured organism was not capable of competing with reef-building filamentous cyanobacteria, but instead occupied niches in which the filamentous calcimicrobes did not live owing to (1) poor illumination or (2) high hydrodynamic energy. The occupation of cryptic microniches (shelter and reef framework voids) by sponges (for example, microfacies i and ii), is well known in the Phanerozoic21,37,39,40.

The obligatory spatial association of vermiform microstructure with reefs built by oxygen-producing cyanobacteria may indirectly support a metazoan interpretation. Prior to the Neoproterozoic oxygenation event, marine dissolved oxygen was probably low41 except perhaps in the vicinity of photosynthesizing microbial communities; the metabolic requirements of metazoans may have limited early animals to localized, comparatively well-oxygenated (for the time) environments (oxygen ‘oases’). Given the approximately 890-Ma depositional age30,31, the vermiform-microstructured Little Dal organism may been tolerant of ‘low’ oxygen (that is, relative to modern levels), which is a characteristic of some modern and fossil sponges42.

If the vermiform-microstructured masses in the Little Dal reefs are accepted as early sponge body fossils, their approximately 890-Ma age would imply that (1) the evolutionary emergence of metazoans was decoupled from the Neoproterozoic oxygenation event41,42,43,44,45 and (2) early animal life was not catastrophically affected by the Neoproterozoic glacial episodes. If the Little Dal objects are truly sponge body fossils, they are older than the next-youngest undisputed sponge body fossils (Cambrian)15 by approximately 350 million years.

It would not be surprising to find that the earliest sponges were reef-dwellers; the history of Phanerozoic reefs is rich with reef-building and reef-dwelling sponges46. If the masses of vermiform microstructure in the Little Dal reefs were to be accepted as an early Neoproterozoic expression of sponge tissue preservation, their age and proposed identity would be compatible with (1) evidence that the opisthokont (animal + fungus) clade was already established by the time of the Mesoproterozoic–Neoproterozoic transition47,48, (2) possible evidence of 1-billion-year-old multicellular holozoans49, (3) molecular clock estimates for the emergence of the Porifera in the early Neoproterozoic1 and (4) a revised taxonomy of nonspiculate keratose sponges showing that they are a sister group to other demosponges19. The Little Dal vermiform microstructure is perhaps exactly what should be expected of the earliest metazoan body fossils: preservation through post-mortem calcification of sponge-grade soft tissue in the decaying bodies of small, shapeless, sessile, epibenthic and cryptic animals most closely affiliated with keratose sponges.


Field work was done on foot from two-person, backpacking-style camps placed at sites that are accessible only by helicopter. Samples were collected at various times between 1992 and 2018, under all required permits. Recording sample locations using GPS is not possible for most sites owing to the extreme topography of the exposures’ cliffs, pinnacles and canyons, and so sample location was documented using photographs and sketches. Several samples are from a mineral-exploration drill-core stored on-site in the field. Owing to the homogeneous grey weathering of reef surfaces, lithofacies cannot be identified in the field. Instead, hand samples were collected and later slabbed and thin-sectioned. Vermiform microstructure was identified in a small proportion of the samples collected. Repeat visits focused primarily on resampling the rare areas in which vermiform microstructure had been identified.

Standard 30-μm-thick thin sections were examined in plane-polarized transmitted light using a Nikon C-Pol binocular microscope fitted with digital camera and Luminera Infinity Analyze software (for lower-magnification images) and an Olympus BX-51 petrographic microscope equipped with Q-Imaging digital capture system (for higher-magnification images).

Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this paper.