Over two decades of ecosystem research suggest that there are fundamental differences in the way that temperate and tropical ecosystems cycle nitrogen (N) and phosphorus (P), and that foliar and litter N:P decrease with latitude. Commonly, two hypotheses are invoked to explain these differences: 1) tropical soils are older than temperate soils, and 2) leaching is more intense in the tropics. However, it has been difficult to test the relative importance of these drivers. While  substrate ages are straightforward to determine and interpret on geomorphically stable (uneroded) surfaces that formed at a particular time, in eroding landscapes  substrate “ages” reflect the rates of rock weathering, erosion and mixing of dust into soil. Here we build a  framework for quantifying P losses from soil relative to bedrock by estimating P inputs (rock weathering and dust deposition) and outputs (P leaching). We parameterize our model with spatially-explicit estimates of global erosion and dust deposition, which together drive the mean residence time of P in soil.  To test the relative importance of residence time and leaching rate, we initially hold leaching rate constant everywhere precipitation exceeds evapotranspiration.
Results/Conclusions

Surprisingly, we find only a modest latitudinal gradient in soil P depletion if we hold leaching rate constant.  Mean depletion values in the humid tropics are <2x greater than in the previously unglaciated humid temperate zone.  Dust inputs increase total soil P relative to identical conditions without dust, but not markedly more in the temperate zone than in the tropics.  However, when we allow the leaching rate to vary,  the retention of P in soils changes dramatically.  The relatively small latitudinal gradient in P residence time, and the large predicted effect of leaching rate suggest that soil age differences are unlikely to be the main driver of latitudinal gradients in ecosystem stoichiometry.  Several studies suggest that leaching rates increase (often non-linearly) increase with precipitation.  While our model does not explicitly incorporate this effect, we hypothesize that instead of soil residence time, large latitudinal gradients in infiltration may cause leaching rates to vary as well, which provides a candidate for explaining the inferred latitudinal gradients in soil P depletion.