Late Neoproterozoic seawater oxygenation by siliceous sponges

The Cambrian explosion, the rapid appearance of most animal phyla in the geological record, occurred concurrently with bottom seawater oxygenation. Whether this oxygenation event was triggered through enhanced nutrient supply and organic carbon burial forced by increased continental weathering, or by species engaging in ecosystem engineering, remains a fundamental yet unresolved question. Here we provide evidence for several simultaneous developments that took place over the Ediacaran–Cambrian transition: expansion of siliceous sponges, decrease of the dissolved organic carbon pool, enhanced organic carbon burial, increased phosphorus removal and seawater oxygenation. This evidence is based on silicon and carbon stable isotopes, Ge/Si ratios, REE-geochemistry and redox-sensitive elements in a chert-shale succession from the Yangtze Platform, China. According to this reconstruction, sponges have initiated seawater oxygenation by redistributing organic carbon oxidation through filtering suspended organic matter from seawater. The resulting increase in dissolved oxygen levels potentially triggered the diversification of eumetazoans.

This correlation with elements derived from clastic components (zircon, clay) suggests that ƩREE+Y scavenging occurs at the seawater-sediment interface rather than in the water column (see main text for explanations). ƩREE+Y are poorly correlated with P2O5 and TOC concentrations (D, E).

Supplementary Note 1: Constant silicon isotope fractionation during silica precipitation
The bulk sediment δ 30 Si could potentially be dependent on the composition of the sediment and thus sedimentary facies, because clay and organic matter can affect rates of silica precipitation 1,2 and thus Si isotope fractionation 2 . To rule out that the changing sedimentary facies towards more clay-and organic matter rich deposits affected the δ 30 Si trend upsection, we have compared the within-layer variability of Al2O3 and TOC with δ 30 Si. We have collected nine chert samples from two continuous stratigraphic layers at 'Longbizui' section, Hunan Province South China (28°30'0.00"N, 109°50'24.00"E). Individual samples were collected 0.7 to 1 m apart from one another, with a total lateral range between 2.7 to 4.1 m.
The bulk chert δ 30 Si of all samples within the chert layers is analytically indistinguishable despite a large range of Al2O3 and TOC ( Supplementary Fig. 3).
Therefore, we conclude that the fractionation factor during silica precipitation was

Supplementary Note 2: Geochemical mass balance
The sampled chert consists of silica inorganically precipitated from seawater, detrital minerals (clays: "detr" and quartz: "qtz"), authigenic illite ("auth"), and silica from sponge spicules. The chemical and Si isotope composition of bulk samples is a mixture between five end members (see below), the relative fractions of which are expressed here in Si mass fractions f. In order to calculate the contribution of sponge material in each sample, the following mass balance equations can be written.
The dominant detrital mineral of the siliceous shales and chert is illite (Supplementary Data 2) and thus for simplification we assume that Al is exclusively derived from illite. This assumption is verified by exemplary quantitative XRD analyses, showing that this approximation yields calculated amounts of illite that are within uncertainty identical to measured amounts (Supplementary Data 7). Furthermore, we assume that detrital illite (detr) and authigenic illite (auth) are present.
where ( Al Si ) j and δ 30 Si j are the molar Al/Si ratio and Si isotope ratio, respectively, of the bulk sample (j= chert) or end member (quartz: j= qtz, detrital clays: j= detr, authigenic clays: j= auth, inorganically precipitated silica: j= inorg, and sponge silica: j= sponge), and f(Si) j is the fraction of bulk Si that is present in the end member j, such that: The authigenic clay represents illite that formed during burial diagenesis from smectite. During the diagenetic transformation reaction of K-feldspar and smectite to illite two moles of SiO2 are released per mole illite formed 6 .
Additional constraints can be used for this mass balance. As shown in equation 4, we assume that detrital clays and authigenic clays have the same Si isotope composition (see discussion below). A fraction of the illite in the chert is likely of detrital origin rather than authigenic: with r 1 the ratio between the number of Si moles carried by detrital illite and that carried by authigenic illite. The rest of the detrital material is quartz: with r 2 the ratio between the number of moles Si carried by illite over that carried by quartz.
Our mass balance problem is thus a system of 5 equations (eqs. 1 to 5) and 5 unknowns (Si fractions f(Si)) that can be solved analytically provided that constraints are available for the remaining parameters (Si isotope composition and Al/Si ratios of chert samples and of end members, as well as for r1 and r2). and assume a 20 % uncertainty on this element ratio. For authigenic illite we use ( Al Si ) auth = 0.495 ± 0.099 (20 %), accounting for the two moles SiO2 that are generated during diagenetic smectite-to illite conversion 6 . As the SiO2 released in this reaction and in the authigenic illite is twice as high as that in detrital illite, r1 = 0.5 ± 0.5. Fine-grained clastic sediments also contain a fraction of detrital quartz, which is difficult to quantify. We assume that the clay/detrital quartz ratio is 1.7 8  compositions. We emphasise that uncertainty results were only marginally different using directly 1,000,000 runs over which all variables were considered independent, but believe that our approach is in theory sounder. Similar results were also obtained with lower number of simulations (10,000 in total), providing confidence that our simulations are statistically significant. Results are reported as the median and 25 th and 75 th percentiles (hence yielding a 50 % confidence interval) of the output distribution over the 1,000,000 runs.
Detrital quartz and clay have comparably well-known Si isotope compositions.
Because most detrital quartz has an igneous or metamorphic source, its isotope composition is relatively uniform. Clay δ 30 Si bears a greater variability, where clay δ 30 Si is typically negative [9][10][11] . The δ 30 Si qtz was set to -0.1 ± 0. Therefore, if organic carbon has a size-dependent carbon isotope composition sediments would become enriched in 12 C relative to bulk organic carbon by filter feeding through sponges. Alternatively, or additionally, a shift towards high δ 13 Corg (by 2 to 3 ‰) and low TOC can result from thermal alteration due to the extraction of isotopically light hydrocarbons 16 . This effect increases the δ 13 C of residual organic carbon, and would be pronounced in samples with low organic carbon concentrations. Overall, the isotopic difference between samples with low and high organic carbon concentrations is ambiguous with respect to discerning primary from secondary controls on δ 13 Corg. Regardless, the trend towards higher δ 13 Corg recorded in TOC-rich samples on the outcrop scale (main text Figure 1B) should be unaffected by potential diagenetic effects due to overall low postdepositional organic carbon losses -as suggested by the TOC-Ni and TOC-Cu correlations (main text figure 3).

Supplementary Note 4: Rare earth element geochemistry
Typical seawater REE patterns in bulk sediment ( Supplementary Fig. 1