Keratolite – stromatolite consortia mimic domical and branched columnar stromatolites

The term keratolite is proposed for keratosan sponge carbonate dominated by vermiform fabric that preserves the outlines of the original spongin skeleton. Thinly (<~2 cm) interlayered keratosan–microbial carbonate consortia in peritidal sediments near the Cambrian–Ordovician boundary in Newfoundland, Canada, are macroscopically indistinguishable from stromatolites. These carbonate domes and columns consist of approximately equal proportions of keratolite and stromatolite. The keratolite is characterized by pervasive microscopic vermiform fabric, which reflects the original spongin framework. The stromatolite is characterized by fine-grained carbonate with cross-cutting laminae, which primarily formed by sediment trapping. The intimate association of keratolite and stromatolite in these deposits indicates that the sponges and microbes involved shared similar environmental tolerances and requirements. Synchronicity of sponge colonization, followed by stromatolite regrowth, across adjacent columns suggests coordinated responses by both sponges and microbes to local ecophysiological stimuli. Due to their macroscopic similarity, keratolite and fine-grained stromatolite may commonly have been confused with one-another throughout the Phanerozoic, and possibly longer.

Here we propose the term keratolite for sediment dominated by the calcifiedtypically vermiformremains of keratosan spongin skeleton, and describe examples of intimately interlayered keratolite-stromatolite associations from peritidal carbonates near the Cambrian-Ordovician (485 Myr) boundary in western Newfoundland, Canada. In the field, these sponge-microbial consortia closely resemble branched and domical stromatolites, casting doubt on the ability of meso-macroscopic (e.g., field-based) studies, on their own, to confidently recognize finegrained stromatolites (Reitner et al., 1995). This supports the troubling realization thateven though bona fide Phanerozoic stromatolites are well-documentedkeratosan-microbial consortia could have been mistaken for stromatolites throughout the Phanerozoic (Luo and Reitner, 2016), and possibly also in the late Proterozoic, since it is widely thought that spongesincluding keratosanscould have originated in the Neoproterozoic Sperling and Stockey, 2018). On a more positive note, keratosan-microbial consortia demonstrate that metazoan-mat relationships were not exclusively competitive, and that these groups could cooperate in building calcified benthic communities (Luo and Reitner, 2016;Lee and Riding, 2021).

Field descriptions
We collected putative stromatolites (which we now recognize as keratolite-stromatolite consortia) from the uppermost Cambrian (Berry Head Formation) and lowermost Ordovician (Watts Bight Formation; ~485 Ma) successions in coastal exposures on the eastern and western sides of Isthmus Bay (Fig. 1). These localities were respectively named Port au Port and Green Head by Ji and Barnes (1994).

Locality 1, Port au Port (uppermost Cambrian, Berry Head Fm.)
A ca. 200 m thick northward-younging Cambrian-Ordovician boundary succession is exposed along the eastern shore of Isthmus Bay, south and southeast of Port au Port (Fig. 1). This section has also been referred to as East Isthmus Bay (Scorrer et al., 2019, fig. 1). Our sample horizon is ~30-33 m below the top of Berry Head Formation, and ~14-17 m below the Cambrian-Ordovician boundary, corresponding to thickness level 133-130 m of Scorrer et al. (2019, fig. 3).
We sampled branched columns that form broad low domes, up to ~1 m thick and a few meters wide, overlying thin bedded lime mudstone ( Fig. 2A) and laterally surrounded by intraclastic rudstone. The rudstones contain subrounded pebbles that include reworked columns, 1-3 cm in size, indicating synsedimentary lithification of the columns (Fig. 2E). The columns in the domes are 1-2 cm wide, short (~2-4 cm high), erect, closely spaced, and irregularly branched, with margins that can be smooth but are more commonly irregular. Their convex-up laminae indicate relatively low primary relief. Adjacent columns occur in subhorizontal horizons 5-10 cm thick, separated by thin (~1 cm) layers of medium to coarse intraclastic packstone-grainstone ( Fig. 2B-D). Similar sediment occupies intercolumn spaces. The columns often widen upward, and locally show bridging. Their margins are commonly ornamented by irregular lateral projections and protrusions (Fig. 2C). Branching ranges from parallel to slightly divergent (see Walter, 1972, fig. 3), primarily in response to influx of carbonate sediment. In overall appearance, these branched columns broadly resemble stromatolites that were widespread in the Neoproterozoic, ~800 Ma (e. g., Walter, 1972;Grey and Blake, 1999, fig. 6;Grey and Awramik, 2020, fig. 94a, d). The immediately associated sequence includes small steepsided putative stromatolite domes with up to 15 cm of primary relief, thrombolite domes, and carbonates ranging from locally bioturbated thinly bedded micrite to coarse intraclastic rudstone (Scorrer et al., 2019, fig. 3) and flat pebble conglomerate. These deposits suggest shallow water environments in which sediments were commonly reworked by waves and currents. Skeletal fossils are scarce in the sampled horizon, but brachiopods, trilobites and conodonts in the associated succession (Scorrer et al., 2019) indicate normal marine salinity.

Locality 2, Green Head (lowermost Ordovician, Watts Bight Fm.)
This northward-younging Cambrian-Ordovician boundary succession is exposed in coastal cliffs along the western side of Isthmus Bay, southwest of Port au Port (Fig. 1). It comprises sections termed Green Head (Pratt and James, 1982, fig. 2 and p. 558;Ji and Barnes, 1994, fig. 1) and Isthmus Bay (Pratt and James, 1986, fig. 2;Knight et al., 2008, fig. 3), as well as the Watts Bight and Boat Harbour reference sections of Knight and James (1987, fig. 3d). Our samples are early Tremadocian in age. They overlie thrombolite mounds with abundant chert nodules, ~5 m above the base of the Watts Bight Formation (Knight et al., 2008, fig. 4 column B).
We sampled a horizon of small, pale tan colored, planar to laterally linked low domes, in beds up to ~20 cm thick. Individual domes, ~5 cm wide with ~2-3 cm of primary synoptic relief, are underlain and overlain by bedded micrites (Fig. 3A). Samples were collected from the uppermost part of the bed, where thinly bedded micrite onlaps the domes (Fig. 3B). The immediately associated sequence includes laminated lime mudstones and bioturbated dolostones with abundant thrombolite mounds (Knight et al., 2008). A horizon of microbial domes with chert linings, slightly lower in the sequence, marks the base of the Watts Bight Formation (Knight et al., 2008). The "Green Head Mound Complex", with thrombolite and Amsassia (Pratt and James, 1982;Elias et al., 2021), occurs ~20 m above the sampled horizon. The overall depositional environment of the Watts Bight Formation has been interpreted as peritidal (Pratt and James, 1986). Macrofossils are generally scarce (Knight et al., 2008).

Port au Port branched columns
The keratolite-stromatolite columns either maintain their width vertically or, more commonly, slowly expand upward. They are generally closely spaced, and typically equal or exceed the volume of intercolumn sediment (Fig. 4). Each column consists of regular to irregular alternations of keratosan sponge and microbial carbonate (Figs. 4,5). The overall ratio of keratolite to microbial carbonate is 38% to 62% (Fig. 4C). Branching appears unrelated to whether the column is dominated by keratolite or microbial carbonate at the point of branching, and bridges that connect the columns can be formed by keratolite and/or microbial carbonate (Fig. 5C). Intercolumn matrix is dominated by subrounded-angular fine to coarse intraclast packstone-grainstone ( Fig. 5A-C), locally with small column fragments, and is commonly dolomitized (Fig. 4A, B).
The keratolite layers can be exceedingly thin, and typically range ~1-10 mm in thickness. They are characterized by pervasive "vermiform" fabric: microscopic sparry networks that traverse micritic groundmass ( Fig. 5A-D; see Section 5.3. Vermiform fabric, below). Keratolite layers generally maintain their millimetric thickness across the column width, which is typically ~1-4 cm, and can often be traced at the same level from column to column across entire hand samples (~10 cm; Figs. 2B-D, 4). These layers tend to have relatively even, welldefined bases, with slightly to moderately irregularoften less welldefinedtops ( Fig. 5A, B). Correspondingly, microbial carbonate layers have slightly to moderately irregular bases, and relatively even tops. Small scale lateral interfingering occurs locally between sponge and microbial carbonate ( Fig. 5A, B). Preservation is generally good in both microbial and keratosan fabrics, consistent with synsedimentary lithification, and the columns are only slightly dolomitized. In shape and size, the outlines of these Port au Port keratosan sponges closely resemble those of some present-day examples (e.g., Luo and Reitner, 2016, fig. 6F). Microbial fabrics show both even and cross-cutting fabrics (Fig. 5E) and locally incorporate carbonate silt-sand grains (e.g., lower part of Fig. 5A), suggesting agglutination of allochthonous sediment. Nonetheless, adjacent reworked keratolite-stromatolite clasts (Fig. 2E) also indicate synsedimentary lithification.

Green head domes
These small, laterally linked (see Logan et al., 1964) and welllaminated keratolite-stromatolite domes, with locally steeply angled margins, occur within sand-poor, very fine-grained, carbonate . Both form relatively even alternating layers, mostly up to ~2 cm thick, that tend to persist laterally across adjacent columns. mudstones (Fig. 3B). The keratolite-microbial carbonate ratio within the mapped domes is equal (50:50; Fig. 6C), and overall layer thicknesses, as well as the internal structures of both the keratosan sponges and the microbial carbonates, are similar to those of Port au Port columns (Fig. 7). The fine-grained micritic microbial carbonate is dominated by well-defined uneven laminae that commonly show crosscutting (Fig. 7). Allomicrite layers are commonly intercalated with sponge/microbial carbonate layers (Figs. 6, 7B). The keratolite layers range from very thin to 5 mm thick. They have smooth bases, but the tops can be even more irregular (Figs. 6, 7A) than those of the Port au Port columns, and small, isolated growths surrounded by allomicrite occur locally (Fig. 7B).

Comparisons
The branched columns (Port au Port) and small domes (Green Head) are both macroscopically laminated. Microscopically they consist of interlayered keratose sponge (keratolite) and microbial carbonate (stromatolite). Each of these components tends to form discrete laterally persistent bands that alternate with one-another on mm-cm scales. The matrix is coarse in the Port au Port samples, and fine in the Green Head samples. Nonetheless, these domes and branches both contain roughly similar proportions of keratolite and stromatolite. In both cases, the keratolite is dominated by vermiform microfabric and the stromatolite is typically finely layered micrite with thin, cross-cutting laminae that likely resulted from trapping and binding (Tosti and Riding, 2017). Reworked columns at Port au Port indicate synsedimentary lithification. In nearly all the cases observed, sponge bases are relatively smooth and sponge tops tend to be irregular. Thus, despite differences in immediately associated carbonate sediment (coarse at Port au Port vs. fine at Green Head), morphology (branched vs. domical), and age (Late Cambrian vs. Early Ordovician), in many respects the keratolite and stromatolite components are similar at the two localities.
In the Boat Harbour Formation, ~100 m higher in the succession than our Green Head samples, Pruss and Knoll (2017) found that animal trace fossils and microbialite only rarely co-occur, supporting an antagonistic relationship, whereas skeletons of benthic invertebrates commonly co-vary positively with thrombolites, suggesting facilitation between microbial bioherms and at least some animals, which is also inferred in the "Green Head Mound Complex" (Pratt and James, 1982). In these dynamic peritidal environments, it is possible that keratosans, in consortium with stromatolites, occupied transient habitats unsuitable for other sessile metazoans. The Port au Port branched columns are arranged in broad low domes, and abundant relatively coarse sediment separates the columns ( Fig. 2A). In contrast, small steep-sided domes at Green Head are surrounded by fine-grained carbonate (Fig. 3A). We infer that influx and movement of coarse sediment at Port au Port engendered and maintained branching, whereas small domes at Green Head developed relatively rapidly during intervals of slower sedimentation in less dynamic muddy environments. Irrespective of associated sediment, however, the stromatolite and keratolite fabrics in both situations are generally fine-grained. At Port au Port, sponge growth/colonization across adjacent closely spaced columns suggests a locally coordinated response to environmental and/or biotic triggers.
These interlayered sponges and mats, at Port au Port and Green Head, evidently shared similar environmental preferences and tolerances. Light could have promoted photosynthesis in both mat bacteria and sponge photo-endosymbionts, water movement would have brought food for sponges and nutrients for microbial mats, and current scour may have hindered burial by sediment. Microbial mats and sponges often tolerate fluctuations in temperature (although sponges are more sensitive to elevated temperature; Webster et al., 2008), and low oxygen levels could have favored both sponges  and mats (Des Marais, 1990;Gutiérrez-Preciado et al., 2018). Mats and sponges alike require stable substrates and sufficient relief to avoid over-burial by sediment. These requirements, together with ability to occupy similar environments, are typical of reef consortia in general (Riding, 2002).
Keratosan sponges appear to be preserved by synsedimentary calcification processes similar to those that affect siliceous sponges (Luo and Reitner, 2014): the skeletal scaffolding remains more-or-less intact, the associated soft-tissue is permineralized to CaCO 3 , most likely during microbial degradation, and the skeletal network (spongin in keratosans) is either replaced or infilled by microspar (Brachert, 1991;Reitner, 1993;Reitner et al., 1995;Warnke, 1995). Nonetheless, keratosans are generally macroscopically much less conspicuous as fossils than similarly calcified siliceous sponges. This is probably due to the delicate nature of keratosan spongin network in comparison to siliceous spicules, and to the commonly less distinctive overall morphology of keratosans. Recent studies have drawn attention to keratosans in Cambrian microbial carbonates  and show that keratose sponges can be significant components of structures long thought to be purely microbial in origin, such as Cryptozoön (Lee and Riding, 2021). In significant contributions, Luo andReitner (2014, 2016) identified "vermiform" microfabric as keratosan spongin network and demonstrated that keratosans likely have been widely overlooked in Phanerozoic shallow marine fine-grained carbonates generally (Table 1). The interlayered association with microbial carbonates characteristic of our Newfoundland examples is not unique to keratosans. Some lithistids and archaeocyaths, for example, also form relatively thin layers within and between microbial carbonate (Kruse and Reitner, 2014;Debrenne et al., 2015).

Vermiform fabric
The internal organic fibrous meshwork that supports keratose sponges  can be preserved in carbonates as "vermiform" fabric, a delicate microscopic sparry filamentous pattern created by the outlines of the original proteinaceous spongin network within fine-grained carbonate matrix (Luo and Reitner, 2014). Vermiform fabric is characterized by straight to slightly curved bifurcating filaments of moderately even thickness that create a somewhat irregular anastomosing network of curvilinear Y-shaped tubules (Figs. 5D, 8). In contrast, in siliceous spicular frameworks silica is deposited on proteinaceous filaments, forming regular spicules with distinctive shapes (Reiswig and Mackie, 1983;Weaver et al., 2007). For example,

Terminology
We propose the term keratolite for carbonate that preserves the vermiform outlines (e.g., Figs. 5D, 8) of the originally spongin keratosan skeleton. Keratolite characteristically consists of dense fine-grained millimetric to centimetric layers that represent synsedimentarily calcified organic tissue initially supported by a network of anastomosing spongin fibers. Internally, keratolite fabric can range from irregularly (Fig. 7A) to sub-millimetrically layered (Fig. 5A, B). Locally, it can incorporate fine sand grains (Lee and Riding, 2021). Keratolite can be closely associated with stromatolite (laminated microbial carbonate), as in our Newfoundland examples, as well as with diverse metazoan and other microbial reef fabrics (Table 1). In addition to vermiform fabric, cylindrical outlines and central cavities (spongocoel) typical of filterfeeding organisms can occasionally be recognized (Lee et al., 2014, fig. 7A-C).
Sponge calcification can involve microbial activity (Reitner, 1993;Reitner et al., 1995), and the conditions necessary for synsedimentary calcification to preserve keratosans were likely enhanced at times when microbial calcification in general was widespread. The secular distribution of keratolite compiled here (Table 1) broadly coincides with intervals when reefal microbial carbonates were generally abundant: Cambrian-Ordovician, Late Devonian-Mississippian, Early-Middle Triassic, and Late Jurassic-Cretaceous (Riding, 2006, fig. 6; Fig. 9). However, it is also possible that calcifying endosymbiotic bacteria might be involved in keratosan calcification, as in some present-day siliceous demosponges Garate et al., 2017).
The presence of fossil keratose sponges in carbonate sediment is only clearly revealed by recognition of vermiform fabric in thin-section (e.g., Lee et al., 2010Lee et al., , 2014Luo and Reitner, 2014). Current understanding suggests that, despite their local volumetric abundance (Hong et al., 2016), keratosans do not create morphotypes readily distinguishable from those of stromatolites, and their overall form probably broadly reflects environmental controls. This underscores the likelihood that keratose sponge-microbial associations have often gone unnoticed and may have been widely mistaken as purely stromatolitic (i.e., microbial) in origin. Many putative "stromatolites" of the past 500 or more million years may harbor substantial volumes of previously unrecognized keratolite.

Conclusions
1. Keratolite (new term) is defined as keratosan sponge carbonate dominated by vermiform fabric that preserves the outlines of the original spongin skeleton. 2. Branched columnar and domical morphotypes formed by keratolite-stromatolite consortia in the Late Cambrian-Early Ordovician of Newfoundland are macroscopically indistinguishable from similar morphotypes that are entirely stromatolite. 3. In these deposits, keratolite is intimately interlayered with stromatolite fabric and forms layers that can be traced across domes, and from column to column in branched forms. The keratosan sponge layers typically display smooth bases and irregular tops. They constitute ~38-50% of the combined keratolite-stromatolite volume. 4. Microscopically, keratolite is distinguished by pervasive vermiform fabric. This distinctive delicate, sparry, anastomosing filamentous network results from synsedimentary calcification of the supportive keratosan spongin framework within fine-grained carbonate. In contrast, associated fine-grained stromatolites in our samples lack vermiform fabric and often display evenly layered and also crosscutting laminae that likely resulted from agglutination of allochthonous sediment. Keratolite and stromatolite both experienced synsedimentary lithification, as shown by reworked fragments. 5. We suggest that spongiolite (originally siliceous sponge), and keratolite (originally proteinaceous sponge) are both varieties of  Kiessling, 2002;Riding et al., 2019).
synsedimentarily calcified sponge that, in close association with microbial carbonates, create Sponge-Microbial Consortia. 6. The macroscopic similarities of keratolite and stromatolite, together with the long geological range of keratosans, make it likely that keratolite has been mistaken for stromatolite throughout the Phanerozoic, and possibly in the late Proterozoic too.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.