The ‘classic stromatolite’ Cryptozoön is a keratose sponge‐microbial consortium

Animal evolution transformed microbial mat development. Canonically inferred negative effects include grazing, disturbance and competition for space. In contrast, ancient examples of cooperation between microbial mats and invertebrates have rarely been reported. Late Cambrian (~485 million years) Cryptozoön is widely regarded as the first stromatolite to have received a taxonomic name and has been compared with present‐day examples at Shark Bay, Australia. Here, we show that Cryptozoön is an interlayered consortium of keratose (‘horny’) sponge and microbial carbonate in roughly equal proportions. Cryptozoön's well‐defined layering reflects repeated alternation of sponge and microbial mat. Its distinctive lateral growth is due to the ability of keratosans to colonize steep and overhanging surfaces. Contrary to the perception of Phanerozoic stromatolites as anachronistic survivors in a eukaryotic world, Cryptozoön suggests mutualistic behaviour in which sponges and microbial mats cooperated to gain support, stability and relief, while sharing substrates, bacteria and metabolites. Keratosan‐microbial consortia may have been mistaken for stromatolites throughout the record of the past 500 million years, and possibly longer.


| INTRODUC TI ON
Microbial carbonates, such as stromatolites, provide the longest fossil record of life on Earth, from at least 3.4 Gyr ago (Allwood et al., 2006;Hofmann, 2000;Lowe, 1980) to the present day (Black, 1933;Logan, 1961). Yet they guard their secrets well and continue to pose persistent research challenges, especially regarding the organisms and processes involved in their formation (Javaux, 2019). These key questions are not restricted to Precambrian examples. Here, we show that late Cambrian Cryptozoön (Hall, 1883), the first stromatolite to receive a formal name (Burne & Moore, 1993) and the focus of an early life controversy (Schopf, 2000), is composed of an intimate association of keratose sponge and microbial carbonate. It has long been suggested that the rise of animals in the late Proterozoic and early Palaeozoic significantly impacted stromatolite development (Awramik, 1971;Grotzinger & Knoll, 1999;Riding, 2006) and restricted their abundance (Garrett, 1970). Recognition of Cryptozoön as a sponge-microbial consortium highlights a very different, mutualistic, relationship between animals and stromatolites. This flies in the face of conventional wisdom that animals outcompeted stromatolites and demonstrates that, in some cases at least, animals and microbial mats developed mutually beneficial associations. In addition to prompting reassessment of exactly how and why stromatolites have changed through time, this raises the intriguing possibility that many supposed stromatolites of the past 500 or more million years may not be what they seem.

| G EOLOG I C AL S E T TING
Late Cambrian and early Ordovician shallow water platform carbonates of the Beekmantown Group were deposited on a passive margin south-east of the Proterozoic Adirondack massif (Landing, 2012). Within this succession, the uppermost Cambrian Little Falls Formation is unconformably overlain by the Lower Ordovician Tribes Hill Formation (Landing et al., 2003). The Little Falls Formation (~40 m thick according to core from Pallette Quarry; Mazzullo et al., 1978) is largely dolomitic, but has limestones at it base (Hoyt Limestone) and top (Ritchie Limestone), both of which are latest Cambrian (Furongian, ~485 Ma) in age based on conodonts (Landing et al., 2011) (Figure 1). The Hoyt was deposited in a shallow, tidally influenced, restricted but normal marine, inner-shelf environment (Mazzullo et al., 1978).

| ME THODS
Vertically orientated slabs and corresponding thin sections (7.6 × 5.2 cm size) were prepared from each sample. Polished slabs were scanned using a flatbed scanner. Thin sections were observed and photographed with a binocular microscope coupled with a digital camera. Dimensions of samples were measured using ImageJ.

| Cryptozoön
Cryptozoön was described and named by Hall (1883) from the upper Cambrian Hoyt Limestone near Saratoga Springs, NY, USA, ~300 km north of New York City ( Figure 1). Goldring (1938) recognized at least four thin (≤1 m) layers containing innumerable closely packed Cryptozoön, within and immediately above the Hoyt Limestone, each layer being associated with oolite and quartz sand and dominated by a particular species: C. proliferum, C. ruedemanni and C. undulatum. Friedman (2000) interpreted Cryptozoön to have formed in a lagoon in the 'lee of an oolite shoal' within a shallowing peritidal sequence.
Present-day exposures include Petrified Sea Gardens, a 20th-century tourist attraction 5 km west of central Saratoga Springs, and Lester Park together with Hoyt Quarry, 1 km north of Petrified Sea Gardens (Friedman, 2000) (Figure 1).

Steele (1825) described distinctive layered domes in the Hoyt
Limestone as 'calcareous concretions', but Hall (1847) recognized them to be organic and suggested they were 'sea plants'. Much later, however, from thin sections, Hall (1883) noted laminae traversed by 'numerous, minute, irregular canaliculi which branch and anastomose without regularity', which also contained 'extraneous and inorganic substances between the concentric laminae'. He named these large concentrically layered deposits Cryptozoön and contrasted them with stromatoporoids (now considered hypercalcified sponges, but at that time often likened to hydrozoans). Hall (1883) realized that although Cryptozoön and stromatoporoids both have 'concentric structure', they differ in significant detail. Macroscopically, stromatoporoids grow convex-up 'from a broad base which is covered by an epitheca', whereas Cryptozoön expands 'from a point below' with the convex surface 'on the lower side' and is 'made up of irregular, concentric laminae of greater or less density and of very unequal thickness' (Hall, 1883).

Glaciated bedding surfaces at Lester Park and Petrified
Sea Gardens display thousands of closely spaced layered 'cabbage-like' domes, typically < 1 m wide, that form a very thin (<1 m) Cryptozoön proliferum biostrome (Friedman, 2000;Goldring, 1938) F I G U R E 1 Locations and geological setting. (a) Cryptozoön is currently exposed in the Hoyt Limestone at localities northwest of Saratoga Springs, New York State, USA: Petrified Sea Gardens (43°05'02"N 73°50'36"W), Lester Park (43°05'32"N 73°50'52"W) and adjacent Hoyt Quarry (43°05'29"N 73°50'55"W). (b) Stratigraphy of the upper Cambrian-lower Ordovician succession near Saratoga Springs (after Landing, 2012;Mazzullo et al., 1978)  Montana (Fenton & Fenton, 1931). Noting Cryptozoön's tendency for lateral expansion, Burne and Moore (1993) compared its surface appearance to the tops of present-day thrombolite domes at Lake Clifton, Australia, and to 'flat-topped mounds' in the Great Salt Lake, Utah, attributed to the cyanobacterium Aphanothece (Carozzi, 1962). They suggested that Cryptozoön proliferum 'grew laterally without any restriction other than that provided by neighbouring structures' and are 'depositional forms in which upward growth was restricted' (Burne & Moore, 1993  grains. (f) Small grain-filled depression (centre) between sponge and microbial layers that also incorporate sand grains sponge tissue (see Reitner, 1993). Sponge fabric also occurs attached to the walls of sediment pockets and rarely within bivalve shells.

| Sponge-microbe mutualism
Present-day microbial mats and keratosans can both colonize and accumulate allochthonous sediment, and it is estimated that 10% of sponges, particularly Demospongiae and Hexactinellida, are welladapted to life with and within allochthonous sediment (Hoffmann et al., 2007;Strehlow et al., 2017). Despite the risk of clogging and burial, sediment incorporation in sponges can deter predation, reduce the need for spicule formation, provide strength, stability and protection (Schönberg, 2016), and increase competitiveness (Biggerstaff et al., 2017). Coarse sediment increases stability (Cerrano et al., 2004), and some sponges selectively discriminate between silica and carbonate grains (Bavestrello et al., 1998). Laminar keratosans cope better with high sediment levels than cup-and hat-shaped forms (Bell et al., 2015;Schönberg, 2016). Our Cryptozoön measurements show that keratosan laminae are shorter and thicker than the microbial carbonate laminae (Figure 8). We performed one-sided t tests [P(T <= t)] to assess these differences. The derived p-value for length (.038) is less than the standard significance level of .05, indicating that sponge and microbial lamina lengths are statistically significantly different.
The difference in thickness is even larger, with p = .00000072, <<.05.  In Cryptozoön, agglutinated sand-size sediment is more common in keratose sponge layers than in microbial carbonate (Figure 7b, d).

F I G U R E 8
This, together with faster innate sponge growth, could account for generally greater thickness of the keratosan layers.
Microbial and keratosan layers repeatedly, and relatively thinly, alternate in Cryptozoön (Figure 3). This is unlikely to be fortuitous. In addition to having similar environmental preferences and behaviours, the microbial mats and keratosans that constructed Cryptozoön could have been mutualistic in providing substrates, bacteria and organic matter.
Stress due to elevated temperature and salinity may have been common in the shallow water environments occupied by Cryptozoön during late Cambrian 'greenhouse' conditions (Lee & Riding, 2018). We suggest that frequent alternation of microbial mat and sponge layers reflects cooperation in Cryptozoön construction as follows: 1. Mats/biofilm surfaces provided sponges with favourable substrates for larval settlement and possibly contributed bacterial symbionts.
2. Keratosans relatively rapidly created extensive enveloping layers but were prone to mortality events that tended to be dome-wide.
3. Microbial mats colonized dead keratosan surfaces, benefited nutritionally from tissue decay and recreated a substrate suitable for sponge larval settlement.
Repetition of this cycle created Cryptozoön's distinctive alternating and laterally extensive layering. At the same time, key macromorphic features of Cryptozoön, such as oblate basal growth, lateral expansion and overhanging margins, reflect the ability of keratosans to encrust steep and overhanging surfaces. Cryptozoön is a sponge-microbial consortium and can therefore be considered an early example of inter-laminated reef construction by differing organisms. Its overall organization broadly resembles that of laminar stromatoporoid and bryozoan consortia in the Middle Ordovician (Hong et al., 2018, fig. 5). A simpler example might be the Early Ordovician sponge Pulchrilamina, which typically is interlayered with fine-grained sediment (Toomey & Ham, 1967, p. 987 and pl. 128).
From this perspective, Cryptozoön can be considered a forerunner of the laminar intergrowth that was a key strategy of reef formation in the Mid-Late Ordovician (Fagerstrom, 1987, p. 344;Kröger et al., 2017, p. 597) and continues to the present day (Riding, 2002b).

| Recognizing keratosan-microbial consortia
It is likely that sponges widely cooperated with microbial mats in the construction of layered calcified sediments since the Cambrian In lithistids, the primary silica spicules have been secondarily replaced by Ca-carbonate. Anthaspidellids are characterized by regular spicule networks that consist of thin dendroclones and thick trabs (fused dendroclones). The resulting structure is polygonal in transverse section (a) and ladderlike in longitudinal section (b). In keratosans, original spongin protein network has been secondarily replaced by Ca-carbonate and is preserved as vermiform fabric poorly oxygenated seafloors that were dominated by bacteria (Gingras et al., 2011). Many putative 'stromatolites' since then may harbour substantial volumes of previously unrecognized keratosan sponge fabric (Luo & Reitner, 2014, 2016. In contrast to the stereotype of Phanerozoic stromatolites as victims of animal competition, Cryptozoön provides persuasive evidence of mutualistic behaviour by which bacterial mats and sponges both benefited. Cooperation was favoured by convergent environmental requirements and tolerances (Lee & Riding, 2018), as well as by similar abilities to incorporate sediment (Schönberg, 2016), that allowed sponges and mats to share substrates, possibly nutrients and bacteria, thereby both gaining relief, stability and support.
Cryptozoön is evidently a sponge-microbial consortium.
Paradoxically, therefore, this 'earliest named stromatolite' is only partly stromatolitic. The ability of keratosans to create laminar morphology and incorporate sand grains, together with the superficial similarity of spongin network to algal and cyanobacterial filaments, and its resemblance-when calcified-to clotted peloidal fabric, have made it easy to confuse encrusting keratosans with microbial carbonate, as pointed out by Luo and Reitner (2016). Fossil keratosans have also been mistaken for other sponges such as lithistids. Nonetheless, calcified keratosans are distinguished by their distinctive vermiform fabric and may also leave macroscopic clues to their presence, as in Cryptozoön's oblate morphology and overhanging lateral expansion.
It is likely that keratosan-microbial consortia are widespread in seafloor carbonate accumulations hitherto regarded as essentially microbial (Friesenbichler et al., 2018;Lee et al., 2014;Luo & Reitner, 2016) and do not all necessarily resemble Cryptozoön. This can be tested by scrutiny of well-preserved putative microbial carbonates of all ages.
A search for keratosans could be particularly rewarding in the late Proterozoic where the suspected origins of sponges have long defied unambiguous confirmation (Botting & Nettersheim, 2018) and where vermiform-like fabrics have been observed (Vologdin, 1962).

ACK N OWLED G M ENTS
We are indebted to Dick Lindemann for help and guidance in the field and for generously providing additional samples, and to Liyuan Liang for statistical analysis and Figure 8. We thank three reviewers for helpful comments that improved the final manuscript, and Kurt Konhauser and Russell Shapiro for expert editorial advice.
J-H.L. was supported by the National Research Foundation of Korea (2018R1A4A1059956 and 2019R1A2C4069278).

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interests.

DATA AVA I L A B I L I T Y S TAT E M E N T
Data available on request from the authors.