Increases in reef size, habitat and metacommunity complexity associated with Cambrian radiation oxygenation pulses

The rise of animals (metazoans) is a seminal event in the history of life. The Cambrian Radiation ~540 Ma marks the appearance of abundant and diverse metazoans and increasing ecosystem complexity in the fossil record¹. A causal relationship between the redox and fossil records is proposed, where oxygen provision reached a threshold, or series of thresholds, which allowed for the diversification of metazoans with increasing metabolic demands². Global geochemical data, however, suggest that oxygenation was not a simple, linear process, but rather occurred episodically via a series of short-lived pulses (1–3 Myr), or ‘oceanic oxygenation events’ (OOEs)^3,4. Early and even later Cambrian seas likely had shallower, and more dynamic, oxygen minimum zones (OMZs) than modern oceans^5,6. Such pulses of increased oxygenation (and related changes in productivity) are hypothesised to have increased the extent of shallow-ocean oxygenation and hence to have promoted diversification⁷. But what remains unquantified is the community-wide response of metazoans to such redox cycles, an insight into the evolutionary processes involved, and hence whether these pulses were indeed a driving force for the Cambrian Radiation.

In order to test the hypothesis that oxic pulses led to diversification and potentially ecological development, a correlation between increased oxygenation, rates of origination, and metrics of metazoan ecosystem complexity needs to be demonstrated. Early Cambrian marine environments were heterogeneous with respect to oxygen provision and nutrient load at a regional scale, so in order to investigate potential correlations, we require the integration of global and local redox proxies, and biotic records in the same stratigraphically well-constrained geological successions.

During the early Cambrian, the Siberian Platform was a vast isolated, tropical continent almost entirely covered by an epicontinental sea (Fig. 1a)^8,9. The platform supported a single metacommunity, i.e. a species pool with many local, interacting communities e.g.¹⁰, representing a third of total early Cambrian metazoan benthic diversity with widespread metazoan (archaeocyath sponge) reefs that formed bioherms (Fig. 1b)^7,11. Dynamic and synchronous changes of body size in archaeocyath sponges, hyoliths, and helcionelloid molluscs through the early Cambrian on the Siberian Platform have been quantified, which coincide with elevated biodiversity and rates of origination: these have been proposed to follow OOEs¹². Here we consider temporal changes in both the position of archaeocyath sponge reefs as a function of relative water depth, and in individual reef size (diameter), as well as the ecological complexity of the reef-building and dwelling communities by quantification of changing reef community membership of sessile archaeocyath sponge, coralomorph, and cribricyath species, on the Siberian Platform.

To quantify ecological complexity, we used metacommunity analyses, which compare the structure between communities in terms of taxa (generally species) compositions spatially and temporally¹⁰ (see Methods). The ‘Elements of Metacommunity Structure’ framework used here is a hierarchical analysis that identifies properties in site-by-species presence/absence matrices that are related to the underlying processes, such as species interactions, dispersal, and environmental filtering that shape species distributions¹⁰. Application to various marine and terrestrial palaeocommunities has demonstrated the robustness of these methods to fossil data and sample size variations^13,14. There are fourteen different types of metacommunity structure which are determined by the calculation of three metacommunity metrics: Coherence, Turnover, and Boundary Clumping, which reveal different controlling processes of underlying metacommunity structure^{10,15,16,17,18}.

The most ecologically complex metacommunities are classified as Clementsian, and have positive coherence, turnover and boundary clumping¹⁶. Clementsian metacommunities contain groups of taxa with similar range boundaries that respond to the environment synchronously as taxa have physiological or evolutionary trade-offs within the communities associated with environmental thresholds¹⁹. By contrast, when taxa respond individualistically to the underlying environment, without accounting for other taxa within the community, the structure is Gleasonian, and is defined by positive coherence and turnover but no significant boundary clumping¹⁶. When coherence is positive, but turnover is not significantly different from random, then the resultant metacommunity structures are known as quasi-structures (e.g. quasi-Clementsian), which reflect weaker underlying structuring processes.

We determined the metacommunity structure for archaeocyath sponge species on the Siberian Platform throughout their early Cambrian record using an entire previously published data set¹¹ then on a sub-set of metacommunities which had a sufficient number of reef sites to be suitable for analyses, i.e. with a sufficient number of sites to be statistically significant. Further, to investigate the effects of water depth on metacommunity structure, we used Spearman rank correlations to test whether the metacommunity ranking (as determined by reciprocal averaging, a type of correspondence analysis which ordinates the sites based on their species composition¹⁷), is significantly correlated to water depth. Finally, to quantify how pairwise associations between taxa change between the three temporally different metacommunities, we determined which pairwise taxa co-occurrences are significantly non-random using a combinatorics approach, and whether any non-random co-occurrences are positive or negative²⁰.

Species richness estimates are highly sensitive to differences in sampling. When comparing species richness of assemblages from several time intervals, it is advisable to standardise sampling across those assemblages to ensure that changes in species richness are not attributable to sampling differences. One approach is to subsample each time interval down to a standardised number of individuals (size-based rarefaction), but this approach can underestimate changes in richness because it tends to sample low-richness assemblages more completely than high-richness ones²¹. Coverage-based rarefaction, where each sample is down-sampled to a standardised level of taxonomic completeness, avoids this potential issue. The coverage of a sample is the proportion of species in the assemblage which are represented in that sample, and it can be estimated by subtracting the proportion of singletons in a sample from 1 (e.g.²²; see also²¹ for details). We used the estimateD function from R package iNEXT²³ to produce coverage-standardised species richness estimates with 95% confidence intervals, by repeatedly down-sampling the sampled assemblage from each time interval to match the coverage of the lowest-coverage interval. We did this by setting datatype = “abundance”, base = “coverage” and leaving all other arguments as default.

In sum, we test the biotic response to OOEs by compiling metrics of archaeocyath reef size, location, and metacommunity complexity, integrated with existing data on archaeocyath individual size, species richness and origination and extinction rates¹² and high-resolution geochemistry^4,7 recalculated to the same stratigraphic scale, on the Siberian Platform over 11 Myr through Cambrian stages 2–3 (mid-Tommotian to early Botoman on the Siberian stratigraphic scale; 525–514 Ma). These results are used to quantify the community-wide response of metazoans to extrinsic redox cycles, and hence gain insight into the evolutionary processes involved.

Geological setting and evolution of redox

During the early Cambrian shallow marine carbonates associated with evaporites and siliciclastics dominated the inner Siberian Platform, passing to shallow marginal carbonates of transitional facies known as the transitional zone (or the Anabar-Sinsk), thence to deep ramp and slope settings that accumulated organic-rich limestone and shale (Fig. 1a)^24,25,26. Archaeocyathan reefs or bioherms were almost entirely restricted to the transitional facies. Such reefs appeared and proliferated during Cambrian stages 2 and 3 (Tommotian, Atdabanian and earliest Botoman), disappeared at the beginning of Stage 4 (middle Botoman) and re-appeared briefly at the end of this stage (Toyonian).

We integrate palaeontological (archaeocyath species number and individual size), palaeoecological (reef size and palaeodepth location) and chemostratigraphic information (carbon isotope cycles 5p, 6p, and II–VII) for sections of the Aldan, Selinde and Lena rivers with sub-metre-scale lithostratigraphic subdivisions^{27,28,29,30,31,32,33} (Figs. 1c, 2a–c, 3a). This results in negligible uncertainty associated with sample heights, which are fixed relative to a consistent datum within each section.

Throughout Cambrian stages 2 and 3, high-amplitude positive δ¹³C carbon isotope excursions show a strong positive covariation with the sulphur isotope composition of carbonate-associated sulphate (δ³⁴S_CAS) in sections from the Lena River (Fig. 3b)⁷. The rising limbs of these excursions are interpreted as intervals of progressive burial of reductants under anoxic bottom water conditions, and a progressive increase in atmospheric oxygen⁷. Coincident δ¹³C and δ³⁴S_CAS peaks (numbered II–VII) correspond with a pulse of atmospheric oxygen into the shallow marine environment (creating an OOE), followed by a corresponding decrease in reductant burial under more widespread marine oxia (falling limbs of δ¹³C and δ³⁴S_CAS), and leading to gradual de-oxygenation over Myr⁷. In addition, phosphorous retention might have occurred under oxic shallow marine conditions, acting to reduce primary productivity and further oxygenate the shallow marine environment in the short-term (<1 Myrs). The OOEs during δ¹³C peaks IV and VI–VII occur ~520.5 Ma and ~515 Ma, respectively (Fig. 3b).

The carbonate uranium isotope (δ²³⁸U) records from the Sukharikha, Bol’shaya Kuonamka, and Kotuykan rivers of Siberia, the Laolin and Xiaotan sections of South China, and the Oued Sdas section of Morocco⁴ are calibrated relative to the δ¹³C record and archaeocyath biostratigraphy (Siberia only), and show a consistent pattern with δ¹³C and δ³⁴S_CAS (Fig. 3c), whereby decreasing values of δ²³⁸U represent global expansions of anoxic, or more specifically euxinic, conditions conducive to widespread reductant (e.g. pyrite) burial⁴. Values of δ²³⁸U reach a nadir during δ¹³C peak 6p, and recover to a more positive mean value during the subsequent terminal Stage 2 to lower Stage 3 interval, which may reflect a gradual transition to a less reducing (or less euxinic) global deep ocean characterised by continued redox stratification⁴.

Reef evolution and habitat occupation

Archaeocyath species diversity on the Siberian Platform increased progressively from their first appearance on the Selinde River in the basal Tommotian Nochoroicyathus sunnaginicus Zone (Fig. 2c), when they were represented by only 3 non-reef building species, to reach 50 species by the early Atdabanian Retecoscinus zegebarti Zone (Fig. 3d)^11,34. After a middle Atdabanian decline, raw species diversity increased again to reach a maximum of 60 species in the early Botoman¹¹. In parallel, origination rates of archaeocyaths increased in two steps, during the Tommotian–early Atdabanian and middle Atdabanian-early Botoman (Fig. 3e). Archaeocyath extinction rates show peaks in the middle Atdabanian and middle Botoman, the latter being the Sinsk Extinction event which resulted in the near extinction of this group and many other metazoans with calcareous skeletons³⁵.

In the middle Aldan and Lena rivers’ area, and on the northern slope of the Aldan Anticline, the first archaeocyath reefs occur at the base of the Pestrotsvet Formation (basal Tommotian Nochoroicyathus sunnaginicus archaeocyathan Zone) (Fig. 1c)³⁶. In the Dvortsy and Ulakhan-Sulugur sections (Fig. 2a and b) they were hosted within ooid and skeletal grainstone, and packstone indicating shallow water conditions at or above fair-weather wave base^36,37. The measured framework of the archaeocyath-calcimicrobial (clotted Renalcis) reef is 1.68 m wide and 1.12 m high with the archaeocyath assemblage consisting of 7 species³⁶.

The basal Pestrotsvet Formation accumulated at the beginning of a transgressive systems tract and deposition continued to the end of the Tommotian (Dokidocyathus regularis and D. lenaicus-Tumuliolynthus primigenius zones) when patch reefs were widespread across the present Lena River area³⁸. Reefs were low biconvex to planoconvex mud mounds, commonly stacked together with sharp boundaries surrounded by peribiohermal sediment, typically argillaceous mudstone and wackestone with skeletal floatstone, rudstone or grainstone. The principal consortia were ramose archaeocyaths (produced by modular branching Archaeolynthus, Tumuliolynthus and Cambrocyathellus spp.), archaeocyath-Renalcis framestone (built by massive modular Dictyocyathus translucidus and Spinosocyathus maslennikovae), bindstone (composed of plate-like Okulitchicyathus discoformis), cementstone, and various archaeocyath-rich mudstones accumulated in subtidal environments between fair-weather and storm-wave base^39,40. Individual reefs contained between 4 and 20 archaeocyath species including, in places, a distinctive large narrow-conical coralomorph Cysticyathus tunicatus¹¹.

The largest (0.5–2.0 m in length and 0.2–0.5 m in height) and most diverse reefs and community types are found in the D. regularis Zone⁴⁰. The belt of reef distribution narrowed in the second part of the zone and reefs disappeared due to regression reducing habitable shelf space by the end of Tommotian 3 (D. lenaicus-T. primigenius Zone)³⁸. Both individual bioherm size and community type diversity also reduced.

Although scattered archaeocyaths re-appeared at the very beginning of the following Atdabanian Stage (Retecoscinus zegebarti Zone), patch reefs only re-appeared towards the end of this zone during a transition from a highstand to the succeeding transgressive tract, and rapidly formed the largest archaeocyath reef belt known on the Siberian Platform, the Oy-Muran reef massif (Fig. 1b and Supplementary Table 2). This reached over 2 km wide along the middle Lena River and was surrounded by mostly argillaceous limestone of the Pestrotsvet Formation. This reef massif was constructed of a number of large low biconvex to planoconvex mud-mounds (1.0–6.0 m in diameter), which like their middle-late Tommotian predecessors, developed on the shallow ramp between fair-weather and storm-wave base and did not produce either prominent palaeorelief or any complex zonation typical of Cenozoic reef systems^41,42. In addition to Renalcis, two larger calcimicrobe species (chambered Tarthinia and dendritic Tubomorphophyton) participated in framework building. Individual bioherms contained 16–21 archaeocyath species including modular individuals, two coralomorphs (large Khasaktia vesicularis bowls and tiny Hydroconus mirabilis cones) and a single cribricyath¹¹.

Archaeocyath buildups also appeared in slightly deeper muddy sediments leeward of the Oy-Muran reef massif where the Negyurchene biohermal massif accumulated (in the vicinity of the Zhurinsky Mys section) (Fig. 1b). Here smaller, lensoid bioherms (about 0.5 m in diameter) grew below storm-wave base within photic zone. This framework was constructed by branching Khasaktia bowls, Renalcis and Epiphyton with accessory solitary archaeocyaths; rare reefs up to 1.5 m in diameter were formed by clones of branching Dictyocyathus bobrovi^11,43. These reef palaeocommunities consisted of 4–10 archaeocyath species. Further to the West (the Malykan section), ephemeral archaeocyath settlements occasionally formed within agitated intertidal conditions, preserved as grainstones and packstones composed mostly of small fragmented cups of 4 species and the filamentous calcimicrobe Batinevia¹¹. In total, 12 different archaeocyathan reef palaeocommunities existed during that time and community diversity reached a peak (Fig. 3h and Supplement Data 1).

This reef-building episode terminated at the end of the transgression. Although the Oy-Muran reef massif re-appeared later, severe dolomitization and faulting within this area obscures all palaeobiological detail. Better preserved, however, is the windward area of the transitional facies zone facing the open sea and represented by limestones of the upper Pestrotsvet which are conformably overlain by the lower Perekhod Formation. These accumulated during the following highstand systems tract (from the Bachyk Creek to the Sinyaya River). Here reefs are regularly stalked dendrolites built mainly by Tubomorphophyton calcimicrobes with 9–21 species of solitary archaeocyaths restricted to muddy depressions and small primary cavities¹¹. These reefs often show a mushroom shape varying in diameter from 0.1 to 0.5 m at different stratigraphic levels. The largest of such bioherms formed wide, laterally contiguous patch reefs or biostromes extending for several tens of square kilometres (e.g. the Bachyk biostrome in Pestrotsvet Formation strata of the Carinacyathus pinus archaeocyath Zone). Host sediments of mainly wackestone, rudstone and mudstone have abundant calcimicrobes indicating reef growth near or slightly below storm-wave base within the photic zone. In general, such reefs occurred through the Atdabanian C. pinus, Nochoroicyathus kokoulini and Fansycyathus lermontovae archaeocyath zones. Examples from the N. kokoulini Zone strata deposited at the end of a transgressive episode are very rare, but a few ephemeral settlements are present within a muddy floatstone in the Bachyk Creek and some other sections.

Deeper water mud mounds without calcimicrobes up to 0.3 m in diameter, probably developed below storm-wave base and below the photic zone. These occur in the easternmost area of the uppermost Pestrotsvet Formation (F. lermontovae Zone in the Ulakhan- and Achchagy-Tuoydakh sections), and were built by 7–12 species of small solitary archaeocyaths and siliceous spiculate sponges surrounded by mudstone¹¹.

Early Botoman archaeocyaths are known from the upper Perekhod Formation to the East of the Oy-Muran massif and from the Mukhatta Unit to the west of this complex (Fig. 1b). Here rich assemblages are described from reworked perireefal facies represented by oolite and skeletal grainstones. Reefs are known from all relative water depths, reaching 1 m in diameter from above fair-weather base to below storm-wave base. The Sinsk Event, ~513 Ma, represented by the Sinsk Formation, is an inferred anoxic event, and led to a complete disappearance of reef communities during the early Botoman Bergeroniellus gurarii trilobite Zone³⁵. There was a brief middle Toyonian reef-building episode before final extinction of archaeocyaths on the Siberian Platform³⁵.

Here we show that individual archaeocyath reef size and extent of habitat, total species diversity, rates of origination and metacommunity complexity increase with phases of oxygenation and decrease as anoxia increases, in pulses, not in a linear fashion, demonstrating patterns of biodiversity and ecosystem integrity were discontinuous, during the Cambrian Radiation.

Source: Ecology - nature.com