Deep Carbon through Deep Time

Robert M. Hazen, Yana Bromberg, Robert T. Downs, Ahmed Eleish, Paul G. Falkowski, Peter Fox, Donato Giovannelli, Daniel R. Hummer, Grethe Hystad, Joshua J. Golden, Andrew H. Knoll, Congrui Li, Chao Liu, Eli K. Moore, Shaunna M. Morrison, A.d. Muscente, Anirudh Prabhu, Jolyon Ralph, Michelle Y. Rucker, Simone E. Runyon, Lisa A. Warden, Hao ZhongCambridge 2019 · DOI · OpenAlex

Abstract

This chapter highlights the use of data-driven discovery to address remaining gaps in our understanding of deep carbon. By transitioning geology from a descriptive science to a predictive, data-driven discipline, researchers are uncovering the complex coevolution of Earth's geosphere and biosphere.

Introduction

For centuries, Earth science relied on the meticulous observations of individual researchers to identify trends in multivariate data. However, this manual approach often fails to capture the "big picture" of planetary evolution. The Deep Carbon Observatory (DCO) is leading a shift toward advanced data science techniques—such as network analysis and statistical modeling—to integrate disparate datasets, from zircon ages to microbial protein structures.

The Geochemical Uniqueness of Rodinia

One of the most significant insights gained from mineral evolution analysis is the identification of the supercontinent Rodinia (~1.3 to 0.9 Ga) as a geochemical outlier. While most supercontinent cycles are characterized by pulses of diverse elements, the assembly of Rodinia was uniquely enriched in Niobium (Nb), Yttrium (Y), and Zirconium (Zr), while remaining notably depleted in others, such as Gold (Au) and Copper (Cu).

Mineral-locality occurrences over time for various elements, highlighting the unique geochemical signature of the Rodinia supercontinent.

This "extrovert" assembly suggests aggressive erosion that stripped away near-surface ore deposits. The high weathering rates associated with Rodinia likely contributed to massive $CO_2$ sequestration, potentially triggering subsequent global glaciations.

Predictive Mineralogy: The Carbon Mineral Challenge

By applying Large Number of Rare Events (LNRE) modeling—a technique borrowed from linguistics—researchers have moved from cataloging known minerals to predicting undiscovered ones. The model estimated that Earth should host at least 548 carbon minerals, whereas only 403 were known at the start of the "Carbon Mineral Challenge." This predictive approach has already led to the discovery of over 30 new species, many of which are hydrous carbonates containing Sodium or Calcium.

Geochemical Evolution and the Biosphere

The coevolution of life and minerals is evident in the bioavailability of metals over time. Network analysis of oxidoreductase proteins reveals a transition from Iron (Fe) to Manganese (Mn) and finally Copper (Cu) as primary metal centers, mirroring the oxygenation of Earth’s oceans. Furthermore, 80% of all reported carbon mineral occurrences date to the Phanerozoic Eon, illustrating the profound impact of biological life on mineral diversity.

Trace element concentrations (Zr, Nb, Y, Ni, Co) in igneous rocks over 3 billion years, illustrating geochemical shifts across supercontinent cycles.

Challenges and Future Directions

Despite these advances, challenges remain. Sampling bias toward rare or aesthetically pleasing minerals persists, and metagenomic databases often lack the environmental metadata (e.g., pH, temperature) necessary for robust geo-bio correlations. Future research will focus on enhancing database quality, developing new analytical tools like affinity analysis, and applying these methods to other planetary bodies, such as Titan.

Summary updated

Abstract

This chapter highlights the use of data-driven discovery to address remaining gaps in our understanding of deep carbon.
BibTeX
@inproceedings{hazen2019deep,
  author = {Robert M. Hazen and Yana Bromberg and Robert T. Downs and Ahmed Eleish and Paul G. Falkowski and Peter Fox and Donato Giovannelli and Daniel R. Hummer and Grethe Hystad and Joshua J. Golden and Andrew H. Knoll and Congrui Li and Chao Liu and Eli K. Moore and Shaunna M. Morrison and A.D. Muscente and Anirudh Prabhu and Jolyon Ralph and Michelle Y. Rucker and Simone E. Runyon and Lisa A. Warden and Hao Zhong and Chao Liu},
  title = {{Deep Carbon through Deep Time}},
  booktitle = {Cambridge},
  year = {2019},
  doi = {10.1017/9781108677950.020},
}
Figure