44-20 Reconstructing Phosphorus Dynamics in the Precambrian Ocean Using Independent Component Analysis of Shale Geochemistry

Abstract:

Phosphorus is widely regarded as the ultimate limiting nutrient for marine primary production over geological timescales, governing organic carbon burial and, ultimately, the accumulation of atmospheric oxygen. Because the atmosphere was persistently oxygen-poor in the Precambrian, the ocean is commonly inferred to have been depleted in phosphorus. A mechanistic basis for this inference exists: in a ferruginous water column, Fe(III) oxyhydroxides efficiently scavenge dissolved phosphate at the redox boundary, and P/Fe ratios in iron-rich rocks and bulk P concentrations in shales, compiled from the literature, are indeed low and comparatively constant throughout the Precambrian, rising sharply in the Neoproterozoic.

However, the compilations rest on bulk-rock chemistry, which represents a mixture of detrital, organic, carbonate, sulfide, and oxide components. Only the chemical precipitates were directly related to the coexisting seawater, so bulk measurements dilute and potentially bias the seawater signal. We, therefore, applied Independent Component Analysis (ICA) to a large shale geochemical database (SGP database) to isolate a Fe-oxide phase, potentially hosting marine phosphorus. Unlike PCA, ICA does not impose orthogonality among components, a constraint with no geochemical justification. ICA was applied independently to age-binned subsets spanning 2000–0 Ma using 13 elements (TOC, Mg, P, S, Ca, V, Cr, Mn, Fe, Zn, Mo, U, Zr), and 100 runs with different random seeds were clustered by cosine similarity to identify stable components because ICA is non-deterministic.

Geochemically interpretable components, including carbonate, organic, iron sulfide, and iron oxide phases, are resolved in most age bins. Rather than reading P/Fe directly from the component vector, which is not strictly quantitative, we selected samples scoring above the 75th percentile on the Fe-oxide component and obtained their median P/Fe. These ICA-filtered P/Fe ratios remain broadly constant from 2 Ga to the present, indicating that the Fe-oxide-bound P reservoir was not systematically depleted during the Precambrian. Given that elevated dissolved silica would have suppressed phosphate adsorption onto Fe-oxyhydroxides, the equivalent P/Fe ratios of iron oxides to the modern values imply an even larger seawater DIP reservoir than in the Phanerozoic. We suggest instead that low Mo availability restricted nitrogen fixation, making N rather than P the ultimate limiting nutrient.

© Copyright 2026 The Geological Society of America (GSA), all rights reserved.