Biogeochemical variation among tropical forests spanning the global range of denudation rates

Background/Question/Methods

The biogeochemistry of tropical forests changes predictably as soils age.  Over time, nitrogen (N) comes to cycle in excess, phosphorus (P) availability declines, and other rock-derived elements are leached from the soil.   However, quantifying soil age is difficult in most landscapes, since age is a function of denudation rate, soil thickness, and soil production rate.  In such cases, we hypothesized that denudation rate may serve as a proxy for soil age, and posited that across a gradient in denudation rate we would see similar patterns to those observed across chronosequences.  Specifically, we predicted that more rapidly denuding soils would appear functionally younger than their more slowly-denuding counterparts.  To test this, we used ¹⁰Be in river sands to measure catchment-scale denudation rates in six sites in the neo and old-world tropics.  Denudation rate differs by three orders of magnitude among these sites, and spans nearly the range found on Earth.  All sites host tropical or subtropical forests atop soils derived from granitic parent material.  We sampled soils from ridge crests in the same catchments and measured soil d¹⁵N (a proxy for long-term nitrogen availability), phosphorus (P) fractions, and the depletion of rock-derived elements relative to parent material.   

Results/Conclusions

Denudation rate varied from 0.004 mm yr^-1 in Imataca, Venezuela to 1.9 mm yr^-1 in northeastern Taiwan.  Denudation rates in the Mata Atlantica, Brazil (0.049 and 0.015 mm yr^-1 in different areas), Luquillo, Puerto Rico (0.1 mm yr^-1), and the Sierra De Las Minas, Guatemala (0.17 mm yr^-1) fell between these extremes.  Thus reasonable estimates of soil residence time suggest a 10,000-fold difference between these sites. Soil d¹⁵N consistently increased with decreasing denudation rate from 3.5‰ in Taiwan to 8‰ in Imataca, Venezuela (p=0.03).   The fraction of rock-derived P and potassium lost relative to parent material increased with decreasing denudation rate (p=0.004 and 0.009, respectively). However, other rock-derived elements showed no such trend (p>0.1), nor did P fractions vary consistently with denudation rate.  For example, the fraction of P in recalcitrant form (not extractable with bicarbonate or sodium hydroxide) ranged between 30-50% but showed no consistent pattern across this extreme gradient in soil residence time. Nevertheless, taken together, these data are cause for cautious optimism that denudation rates can be used to predict biogeochemical properties in landscapes where soil age is not known.