4.7.1. Heterogeneity and Paleosol Representativeness

Fossilized soils (paleosols), which are present in the rock record as far back as at least 3.0 billion years ago, serve as important windows into Earth's terrestrial past [133]. They have been used to reconstruct ancient atmospheres (e.g., [134–138]), climates ([79] and refs. therein), and terrestrial biology and biogeochemical cycling (e.g., [139,140]). The geochemical composition of fossil soils is used to reconstruct a range of climatic and environmental changes; however, these tools are primarily based on small modern soil datasets. This work provides ranges of reasonable values for soil chemistries under known environmental and atmospheric conditions, providing critical background information to improve our paleoclimate and paleoenvironment reconstructions. Based on the highly variable geochemical results in this work, we urge researchers using paleosols for these types of reconstructions to be cautious in assuming a single paleosol profile to be representative of an entire landscape or basin. More work on how representative a paleosol profile is, chemically and climatically, should be done to incorporate landscape-scale variability in soils (e.g., [141]).

#### 4.7.2. Paleosol Fe, Atmospheric Oxygen Reconstructions, and Microbial Life

Because paleosols form on the Earth's surface and in direct contact with the atmosphere, their Fe chemistry has been used to reconstruct atmospheric oxygen levels [134,138,142]. Across a range of soil orders, environments, and climates, non-wetland soils in this work reflect oxidizing conditions (based on Fe<sup>3</sup>+/Fe2<sup>+</sup> ratios; Supplemental Table S4). While Fe3+/Fe2<sup>+</sup> ratios are no longer the primary tool for reconstructing paleo-oxygen levels, these results indicate that that tool is qualitatively robust in the absence of other observations such Cr or S isotopes.

A concern with using fossil soils for reconstructing atmospheric oxygen is that these soils likely hosted microbial life, which could have affected the redox signal ('biosignature') left behind. Indeed, redox-sensitive metals have been proposed as biosignatures because they can be mobilized by biologic processes and redeposited within a soil profile [143]. However, if it is suspected that the soils have variable redox states due either to high moisture input or poor drainage, how reliably might they have recorded the atmospheric oxygen signal as opposed to microscale oxic/reducing conditions due to microbial activity? Modern biological soil crusts (BSC; or, cryptogamic soils) are symbiotic communities of microbes (predominantly cyanobacteria), algae, and fungi; these communities have been proposed as analogues for early life on Earth (and Mars) [143]. Here, we found that the soil redox state being recorded by BSCs was oxic (again, using Fe<sup>3</sup>+/Fe2<sup>+</sup> ratios and dominant Fe speciation; Supplemental Table S4), despite BSCs' ability to retain water and temporarily shift to reducing conditions. It is unlikely that the microbial communities would have 'overprinted' the atmospheric oxygen signal even if they temporarily held water to cycle nutrients and leave behind biosignatures.

#### 4.7.3. Continental Weathering, Nutrient Fluxes, and the Atmosphere

Geologists are interested in constraining many of the processes described above in ancient soils and ecosystems. Weathered P, transported to the oceans via fluvial networks and continental drainage, is typically invoked as a first-order control on marine productivity, which is associated with organic C and organic-associated and inorganic carbonate sedimentation, atmospheric CO2 drawdown, and oxygen production [56]. Continental weathering, and therefore P flux, is a central control in many biogeochemical models of Earth's past. However, the continent-to-ocean transport of P is usually prescribed and not based on actual changes in fossil soil (paleosol) P values or weathering intensity data. Rather, it has been generalized based on crustal P values and fluxes controlled by large-magnitude changes, such as global glaciations [144,145] or limited modern observations [61,62].

Based on our results, continental weathering intensity and resulting elemental fluxes should vary with paleogeography (latitude effect), in addition to changes in climate and landmass area. Clay content through time could be taken into account and paired with CIA to reconstruct actual weathering intensity where those data are available in the rock record [146]. Forests also have higher CIA values than grasslands and shrublands (Figure 7), suggesting that the evolution and spread of different ecosystems could have changed the distribution of weathering intensity. Biogeochemical models should take these differences into account as opposed to assuming uniform behavior spatially.

Continental sulfide weathering in the Precambrian (pre-540 Ma) has been invoked as a major source of sulfate to the oceans, affecting organic matter respiration rates, marine alkalinity and oxidation state, and ultimately the concentration of atmospheric oxygen [147–150]. At the Great Oxygenation Event (GOE) 2.45 billion years ago, pyrite burial and increased ocean sulfate concentrations are thought to have decreased atmospheric methane through microbial sulfate reduction, lowering the greenhouse effect and leading to global glaciations [151,152]. Our results (Supplemental Table S5) constrain the number of Fe-bound sulfides in modern soils, forming under an oxic atmosphere, and suggest the potential for a 'false positive' for a reducing atmosphere if a paleosol is not correctly identified as waterlogged (and therefore reducing) or histic. Although it should be noted that using the mere presence of minerals, such as pyrite or uraninite, in the rock record as reducing indicators [153] is simplistic and no longer widely-used in the field.

#### 4.7.4. Vegetation and P in the Phanerozoic (542 Ma Onwards)

The rise of land plants during the Devonian period (419–349 Ma) has been suggested as a major driver of change to the Phanerozoic P and C cycles, which could have affected atmospheric oxygen [15,56,63,154]. While B horizons in Barren vegetation had increased P retention, the lack of variability between other vegetation domains suggests a somewhat binary response—either soils are vegetated and mobilize/accumulate P, or they are not vegetated and are relatively depleted in P. Only soils with vegetation accumulated high ranges of P, supporting the need for biological mediation of P in terrestrial settings.

#### **5. Conclusions**

From this large-scale analysis of modern soils' physical and chemical properties, we draw conclusions about the relationships between modern soil biogeochemistry and climate, and link those to terrestrial P transport. We also describe implications about past terrestrial biogeochemistry and interpreting the terrestrial rock record.

In modern soils, some of the most well-defined (and expected) relationships were between soil composition and weathering. Latitude, clay content, and weathering correlate as expected, with higher latitudes corresponding with lower weathering intensity and clay content. Contrary to our expectations, average P concentrations are relatively consistent across weathering intensities but show a slight peak at middle weathering intensities, which supports commonly applied models of soil age and P loss. Specifically, it helps to clarify the conflation of time and weathering intensity that can occur, pointing to the importance of soil age as a control on soil P depletion. Maximum terrestrial P transport, then, would still likely occur at a midpoint in the weathering-P spectrum, where a soil is mature enough to have mobilized P from apatite and be linked to high erosion rates, but before soil P is too depleted due to age (which would be a lower P flux overall). Contrary to expectations, P in soils was not strongly associated with Fe (oxyhydr)oxides. The scale of such relationships may be much smaller than continental (i.e., locally or per profile). Additionally, no strong, predictive relationships were present between Fe species and precipitation, soil moisture, or drainage, so predicting P transport based on climate or soil redox properties is not possible. Finally, the presence of vegetation (but not plant functional type or generalized ecosystem type) is important for P accumulation in soils, which has implications for how the rise of terrestrial plants may have changed P cycling through geologic history.

Based on modern relationships between soil P, weathering, and vegetation, the spread of land plants in the early Phanerozoic likely increased P accumulation in soils and on continents, rather than increasing the flux of P to the oceans. This would limit the role of terrestrial-marine P transport in marine productivity. Because of latitudinal variability in weathering intensity, biogeochemical models should take paleolatitude into account when using weathering intensity as a driver for P fluxes. Additionally, models should use density-normalized values for terrestrial P rather than bulk crustal values and should use the range of P in modern soils as a quantitative constraint. Finally, while we did not find the relationships between Fe species and climate that we had expected, the overall Fe signature in soils (including microbially-colonized) was oxic despite analyzing a range of soil moistures and drainages, which supports the use of paleosol Fe/redox geochemistry for paleo-atmosphere reconstruction.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2571-8789/4/4/73/s1, Figure S1: Map of soils used in this work; Figure S2: Maps of P2O5 and Fe concentrations by horizon from USGS database; Figure S3: Scatterplots of mean annual precipitation with P2O5 (a) and Fe (b); Figure S4: Scatterplot of Fe and P2O5, for all samples; Figure S5: Boxplots of the CIA and clay content, sorted by vegetation type; Figure S6: Scatterplots of organic carbon with the CIA, clay content, and P2O5.; Figure S7: Scatterplots of organic, inorganic, and total carbon with total Fe; Figure S8: Boxplot of Fe and P2O5, sorted by parent material; Figure S9: Scatterplot of Fe species and clay content; Figure S10: Principle components analysis biplot; Table S1: Description of the datasets and which has what horizons, geochemical info, etc.; Table S2: Sample locations and soil details for soils compiled in this work (not in the USGS database); Table S3: Soil geochemistry for soils compiled in this work (not USGS data); Table S4: Fe species and Fe3/2 ratios, including BSC samples; Table S5: Pyrite in soils, normalized to density.

**Author Contributions:** Conceptualization, R.M.D. and N.D.S.; methodology, R.M.D. and N.D.S.; formal analysis, R.M.D.; investigation, R.M.D.; resources, N.D.S.; data curation, R.M.D. and N.D.S.; writing—original draft preparation, R.M.D. and N.D.S.; writing—review and editing, R.M.D. and N.D.S.; visualization, R.M.D.; supervision, N.D.S.; project administration, N.D.S.; funding acquisition, R.M.D. and N.D.S. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by Lewis and Clark Astrobiology to R.M.D; Turner Award from the University of Michigan Department of Earth and Environmental Sciences to R.M.D and the NSF Award #1812949.

**Acknowledgments:** NRCS KSSL for soil sample storage, preparation, and shipping; USGS Southwest Biological Science Center for assistance in soil crust sampling; soils from Adrianna Trusiak/Rose Cory lab; Catherine Seguin for her thesis work on biological soil crusts and lab assistance; Sonya Vogel, Bianca Gallina, and Jordan Tyo for assistance in the lab; Emily Beverly for assistance in the field; and Kathryn Rico and Nikolas Midttun for helpful discussions. Our thanks to our anonymous reviewers for their helpful comments and suggestions to improve our manuscript.

**Conflicts of Interest:** The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.
