Linking chronosequences with the rest of the world: predicting soil phosphorus content in denuding landscapes

Results from chronosequences from the arctic to the tropics show that phosphorus (P) availability, total P, and the fraction of bedrockderived P remaining in soil diminishes as soils age. Thus we predict that ecosystems mantling old substrates are likely to have low available P. Yet there are myriad examples in the biogeochemical literature where the results from chronosequences are used to argue the reverse, and ecosystems observed to be P poor are assumed to mantle an old substrate. This premise is difficult to test, for while the concept of substrate age is useful on uneroded surfaces that formed at a particular time, it becomes obscured in denuding landscapes, where substrate ages instead reflect the rates of rock weathering, denudation and mixing of dust into soil. Here we explore this premise as it relates to one of the most ubiquitous assumptions in the biogeochemical literature: that the differences in nitrogen (N) and P cycling between temperate and tropical regions are driven by gradients in substrate age. We build a conceptual framework for quantifying the fraction of parent material P remaining in soil ([SoilP]/[RockP]), by estimating P inputs (rock weathering and dust deposition) and outputs (P leaching). We parameterize our model with spatially explicit (0.5 ) estimates of global denudation, weathering zone thickness, and P deposition. To test the assumption that latitudinal gradients in P status are the result of soil age, we apply a single P loss rate, derived from a humid tropical system in the Hawaiian Islands, to our spatially explicit map of soil residence times. Surprisingly, in this formulation, we find only a modest latitudinal gradient in soil P depletion, with mean depletion values in the humid tropics \29 greater than in the previously unglaciated humid temperate zone. This small latitudinal gradient in P depletion is unlikely to be sufficient to drive the observed differences in tropical vs. temperate ecosystem stoichiometry (e.g. trends in foliar and litter N:P). Thus our results suggest that, to the extent P depletion is greater in the tropics, the appropriate conceptual model for attributing causation may not be one of a chronosequence where time is the primary driver of P loss. We hypothesize that the covariation of inferred P availability with latitude may be strongly controlled by latitudinal changes in rates of P leaching and occlusion, rather than gradients in