*4.1. How do Latitude, Weathering, and Clay Content Associate with P and Fe Concentrations?*

Most of the expected relationships between soil order, Fetot and P, weathering, and latitude were supported. As expected, weathering generally decreases with latitude, dropping from a maximum CIA of >95 at 30–35◦ N to a CIA ~ 75 at 50◦ N (the northern limit of the large USGS dataset). Farther north, some B horizons deviate from that pattern, with CIA values higher than might be expected based on their climatic regime [80,81]. Interpreting the latitudinal trends here should include a caveat for latitudinal sampling bias and limitations of the dataset to, mostly, between 20◦ and 50◦ N. While the PCA results (Supplemental Figure S10) do not support a relationship between P and weathering, we interpret this discrepancy as being due to the complexities within the P-weathering relationship rather than negating the observations made between latitude, weathering, and P concentrations because the PCA eigenvalues are so low (most < ±0.2), and essentially, are clustered on the origin, which indicates a not-predictive value. P's relationship with weathering is complex and due to a variety of factors (climate, time, slope/erosion rate, etc.), and by looking only at the end product—which is what is left in the geologic record for analysis—we must inherently work around those limitations and draw the most robust conclusions possible from limited data on these large scales.

Fetot behaves as expected, accumulating in B horizons as soil weathering increases, while P accumulation in all horizons peaks at moderately-weathered soils (CIA ~ 60). The latter relationship is expected because a soil that has experienced moderate weathering and has developed somewhat (e.g., Inceptisol, Alfisol) has had sufficient time to have P mobilized from the bedrock/substrate and biotically cycled, but not so long or with such intense weathering that P is depleted from the substrate and removed via erosion and/or loss of biomass. An older and/or more intensely weathered soil (e.g., Ultisol) will have a substrate more depleted of P and will have lost more P. In terms of terrestrial P transport, a moderately-weathered soil is the most likely to have a large pool of potential P to transport. An interesting next step, building off this work, could be to test these expected correlations by mapping soil P with soil age and collecting fluvial P loads.

P concentrations are, on average, far below the crustal average of 870 ppm P. Even accounting for density differences between a typical soil (ρsoil ~ 1.62 <sup>±</sup> 0.2 g cm−3; n = 659) [72] and continental crust (e.g., granite; ρcrust ~ 2.7 g cm<sup>−</sup>3), all of the soil horizons are depleted in P relative to the crustal abundance. There is some variability between soil orders, but all but a small subset were below the crustal abundance (i.e., recent basalt-parented soils). When considering bulk density, the total Fe concentrations show the opposite, with the soil Fe average of 4.4% being greater than the crustal average of 3.5%, again with some variability between soil orders but overall greater than the crustal average (Supplemental Table S3). This has implications for biogeochemical modeling (see Section 4.7.3). While P concentrations could be expected to generally decrease with increased weathering intensity, a moderately-weathered soil is more likely to be at a mid-developmental stage (e.g., Inceptisol, Alfisol) and supporting vegetation that can cycle P without being P-depleted (e.g., [19]). This mid-CIA accumulation could be reflective of the vegetation's impact on soil nutrients (e.g., mycorrhizal P mobilization, P recycling through biomass/organic matter decomposition), pointing to a key balancing point in a soil's lifespan where the bedrock is being weathered enough to maximize fertility without yet being depleted. Forest and Altered/Cultivated soils have the highest CIA values (Figure 7), but P concentrations were relatively invariant between types of vegetation, suggesting that weathering intensity exerts a greater control on soil P than the type of vegetation.

Clay content (both percent of grain-size fraction and abundance of clay minerals) should increase with the degree of weathering a soil has undergone, and that is reflected in the results (Figure 4D). Consequently, there are secondary trends associated with clay content, i.e., clay content peaks at lower latitudes where CIA values are the highest. Fetot and clay content show a strong positive correlation, as expected (Figure 6E), but P shows only a weak potential decrease as clay content increases, supporting traditional thinking that P is depleted as a soil is increasingly weathered and/or older (Figure 6F). While it is kinetically favorable for P to sorb to Fe and Al oxides (which accumulate in B horizons as a soil ages), clays (often with surficial Fe-Al oxides) may be equally as important for immobilizing P in soils [82].

Interestingly, our results indicate no positive correlations between Fe oxides (Fedith and Feox) and clay content grain-size fraction (Supplemental Figure S9), as would be expected based on soil development patterns (i.e., higher weathering intensity leads to accumulation of clay minerals and Fe oxides), suggesting that while particle size may mediate Fe speciation due to varying degrees of reactivity, it is not a primary control. While the USGS database's clay content is based on mineralogy (determined by XRD), soils used in the Fe species work had clay content determined by settling, which would include both clay- and non-clay minerals that were clay-sized particles. This could explain at least partially the lack of the expected trend. Further exploration of clay mineralogy-specific correlations (or lack thereof) with Fe speciation pools, in the context of soil redox state, is warranted to explain this deviation from our expectations. Therefore, degree of weathering remains a primary control on soil P concentrations (e.g., [19]).

Although dust deposition is a source of P, it is typically secondary to bedrock weathering, not really applicable on anthropogenic-change timescales, and is not rapid in most of our study area (primarily the conterminous United States). Because it is not separately examined or included in this work, anyone using our results of soil P should take regional dust deposition and subsequent P replenishment into account when possible.

### *4.2. How Do Precipitation, Soil Moisture, and Soil Drainage A*ff*ect P and Fe Concentrations, and Fe Species?*
