*3.8. Principle Components Analysis*

Results from the principal components analyses (PCA) for the USGS data are shown in Supplemental Figure S10. Principle components 1 and 2 explain ~60% of the variance at a maximum. However, all eigenvalues are very low (<0.6) and data are essentially clustered around the origin, so while there are associations, they are very weak and should be interpreted conservatively. Latitude, P, Fe, and Al were associated with PC2 in the Top 5cm and A horizons, and vegetation and CIA were associated with PC1. Groupings changed in the C horizons, with Fe and Al associated with PC2, but P and latitude associated with Corg along PC1. In all horizons, CIA and vegetation vectors are opposite to each other, and were orthogonal to subparallel to most other vectors.

#### **4. Discussion**

In this section, we explore our results with a series of questions focused on constraining controls on Fe and total P in soils, defining their relationships on continental scales, and making inferences about how those relationships may impact terrestrial P fluxes. They are centered around key soil-forming factors, such as climate, vegetation, and weathering intensity, as well as soil redox-specific factors (i.e., precipitation, moisture and drainage, and Fe species). Because a majority of samples come from between 20 and 50◦N, the interpretations made here are most robust for those latitudes.
