*3.4. Fe and P: Drainage and Soil Moisture*

Moderately poorly-drained soils had higher Fe and P concentrations than the other degrees of drainage. Perudic moisture regimes had the highest Fe and P values (Figure 8); however, these samples were dominated by basalt-parented soils in Iceland, which may skew the results. Aside from that, Aquic and Xeric moisture regimes had high Fe, and Ustic and Udic soils had high P (Figure 8). Some trends emerge with relatively high Fe/low P (Aquic) and vice versa (Udic).

**Figure 8.** Variability in Fe concentrations (**A**,**C**) and P concentrations (**B**,**D**) in different soil drainage and moistures regimes. Dashed gray lines represent crustal averages for Fe and P in respective plots. (**A**) Fe binned by drainage. (**B**) P binned by drainage. (**C**) Fe binned by moisture regime. (**D**) P binned by moisture regime. Some patterns emerge when comparing Fe and P between soil moisture regimes, e.g., Udic soils having low Fe/high P and Aquic soils having high Fe/low P. Perudic soils are dominated by basalt-parented soils, influencing their Fe and P concentrations and highlighting the importance of bedrock parent material. No data for Permafrost Fe or P concentrations. Bin sizes for (**A**,**B**): Poorly-drained (166), Moderately poorly drained (25), Moderately well-drained (16), Well-drained (185), Excessively well-drained (12). Bin sizes for (**C**,**D**): Unknown (111), Aridic (73), Aquic (15), Udic (82), Perudic (27), Ustic (54), Xeric (21), Permafrost (N.D.).

#### *3.5. Fe Speciation*

#### 3.5.1. Fe Speciation and P

Pyrite/Fe in sulfides, normalized to average soil order density, were above detection by titration in only 22 tested non-wetland soils out of 63 from around the continental U.S. (Supplemental Table S5), so that test was excluded from subsequent analyses. Density-normalized pyrite yields from all wetland sediments analyzed (n = 15) were measurable/above detection limit (>0.1 wt%), as opposed to non-wetland soils (Alfisols, Aridisols, and Mollisols), which typically fell below this limit after being

density-normalized. In a non-perennially-waterlogged soil (e.g., not wetlands), it is reasonable to assume that Fe contents in sulfides are negligible.

Contrary to expectations (see Section 4.3 below), Fedith and Feox pools did not display stronger or more robust correlations with P (Figure 9) than the other Fe pools. There were no clear correlations between P and any Fe species (for all, R<sup>2</sup> < 0.01). See Supplemental Table S4 for all Fe species results and Fe<sup>3</sup>+/Fe2<sup>+</sup> ratios.

**Figure 9.** Relationships between Fe species and P concentrations, all with R<sup>2</sup> values < 0.1. Arrows indicate off-plot values. (**A**) No strong correlation between Feasc and P (*p* = 10<sup>−</sup>6). (**B**) No significant correlation between Feacet and P (*p* = 0.2). Note the different *x*-axis scale. (**C**) No strong correlation between Fedith and P (*p* = 0.003), where a positive correlation had been expected due to P's affinity to sorb to Fe (oxyhydr)oxides. (**D**) No strong correlation is present between Feox and P (*p* = 10<sup>−</sup>9), again where a strong positive correlation might have been expected.

#### 3.5.2. Fe Speciation, Precipitation, Soil Moisture, and Drainage

Contrary to expectations (see Section 4.2.2 below), Fe speciation pools showed no strong predictive relationships (all R2 < 0.2) with mean annual precipitation (Figure 10), but there were differences between soil moisture regimes. Permafrost soils from the North Slopes of Alaska, high in organic matter and Fe, were dominated by labile Fe (Feasc). Aquic and Udic soils had high Fe in carbonates (Feacet), and Perudic soils (primarily from Iceland, with basalt parent material) had high Feox (Figure 11). Mean Fedith was consistent between soil moisture regimes but has an extremely high range in Aridic soils.

**Figure 10.** Relationships between mean annual precipitation (MAP; mm yr<sup>−</sup>1) and Fe species, colored by soil drainage. Arrows indicate off-plot values. (**A**) No correlation between MAP and Feasc (R2 < 0.1, *p* = 10<sup>−</sup>9). (**B**) No strong correlation between MAP and Feacet (R2 < 0.01; *p* = 0.0005); there could be a maximum MAP value beyond which Feacet is less kinetically favorable. Note the different *y*-axis scale. (**C**) No significant correlation between MAP and Fedith is present (R2 < 0.01, *p* = 0.7). (**D**) Modest significant correlation between MAP and Feox, (R2 = 0.16, *p* = 10<sup>−</sup>17). For all Fe species, there are no strong predictive trends between Fe species concentration and MAP, and no patterns in drainage.

**Figure 11.** Fe species binned by soil moisture regime. (**A**) Udic, Perudic, and Permafrost soils have the highest median Feasc. (**B**) Feacet is quite low in most soils. Note the different *y*-axis scale. (**C**) Aridic soils have high Fedith, but other moisture regimes show little variability. (**D**) Little variability can be seen in Feox across soil moisture regimes. Bin sizes (slight variability between Fe species is due to extraction yields): (A) Unknown (106), Aridic (73), Aquic (15), Udic (82), Perudic (26), Ustic (53), Xeric (21), Permafrost (21). (B) Unknown (111), Aridic (73), Aquic (15), Udic (82), Perudic (27), Ustic (54), Xeric (21), Permafrost (18). (C) Unknown (111), Aridic (61), Aquic (14), Udic (78), Perudic (26), Ustic (48), Xeric (15), Permafrost (21). (D) Unknown (106), Aridic (73), Aquic (15), Udic (82), Perudic (27), Ustic (54), Xeric (21), Permafrost (21).

Poorly-drained soils had a high range of Feasc and Fedith, while well-drained soils had high amounts of Feacet (Figure 12A). The other Fe species and drainages were consistent. The role of slope was considered, but local, small-scale topographic relief was not readily determinable for most samples.

**Figure 12.** Fe species binned by soil drainage. (**A**) Excessively well-drained soils have very low Feasc. (**B**) Note the different *y*-axis scale. Feacet is highest in well-drained soils, although poorly-drained soils show an unexpectedly high range as well. (**C**) Excessive soils show the lowest maximum Fedith, with the other drainages showing similar medians and ranges. (**D**) Feox has similar patterns to Fedith, with three very high values for moderate drainages. Bin sizes: (**A**) Poorly-drained (104), Moderately poorly drained (16), Moderately well-drained (11), Well-drained (140), Excessively well-drained (3). (**B**) Poorly-drained (109), Moderately poorly drained (16), Moderately well-drained (11), Well-drained (142), Excessively well-drained (3). (**C**) Poorly-drained (101), Moderately poorly drained (15), Moderately well-drained (9), Well-drained (123), Excessively well-drained (3). (**D**) Poorly-drained (109), Moderately poorly drained (16), Moderately well-drained (11), Well-drained (142), Excessively well-drained (3).

#### 3.5.3. Fe Speciation, Vegetation, and Soil Order

Forest soils had higher Feacet (Figure 13B) than the other vegetation groups, but Fe species concentrations were consistent otherwise. Oxisols had higher Fedith and Feox than other soil orders (Figure 14C), but Fe species had little variability between soil orders otherwise.

**Figure 13.** Fe species binned by vegetation type. (**A**) Forests and Grasslands have higher Feasc than other vegetations (except for Unknown). (**B**) Forests have the highest Feacet values (median and range). (**C**) No pattern seen in Fedith and vegetation, except that Barren soils have low Fedith. (**D**) No patterns in Feox and vegetation.

**Figure 14.** Boxplots of Fe species binned by soil orders, where centered red circles are the medians, the dark blue boxes are the 25th inner quartiles, the whiskers are the outer 25th quartiles, and the red circles are the statistical outliers (outside 3σ). (**A**) No differences in Feasc between soil orders. (**B**) Ultisols have the highest Feacet, although Alfisols have a high range of Feacet as well. (**C**) Oxisols have the highest Fedith, as expected. (**D**) Oxisols have the highest Feox, as expected.

#### *3.6. Organic Carbon*

Organic carbon (Corg) in A horizons ranged from 0 to 61 wt%, and in C horizons, ranged from 0 to 43 wt% [64]. The average Corg:P ratio for A horizons was 17.9:1, and for C horizons, 18.3:1. Relationships between Corg and CIA, P, Fe, and clay content were weak (Supplemental Figures S6 and S7). There were no strong correlations between inorganic C and Fetot, or between Ctot and Fetot (Supplemental Figure S7).

#### *3.7. Parent Material*

There is slight variability between Fetot and P concentrations and parent material, with igneous bedrock and ash/volcanics having the highest median values for Fe, with high P values as well (Supplemental Figure S8). While limestone parent materials show high Fe, there are only two samples in this group and they are not included in this discussion. The parent material does not appear to be a predictive control in either Fetot or P (Supplemental Figure S8). However, this could be due to the mixed provenance of many modern soils' parent material (i.e., alluvium/colluvium, glacial deposits, etc. as opposed to compositionally homogeneous bedrock). A more targeted exploration into bedrock-parented soils, Fetot, and P could well show more well-behaved relationships.
