Soil phosphorus (P) is assumed to influence primary production and decomposition in many tropical forests, yet the drivers of soil P status in these biogeochemically diverse ecosystems remain poorly understood. We evaluated the relative importance of parent material, topographic position and forest type for determining three metrics of soil P status in the Luquillo Mountains of Puerto Rico: 1) total soil P concentrations, 2) P forms as determined by Hedley Fractionation, and 3) the loss of P relative to bedrock. We sampled soils from four replicate watersheds across a factorial combination of two parent materials (quartz diorite and volcaniclastic) and two forest types (Tabonuco and Colorado). Within each of these sixteen watersheds we dug nine soil pits to a depth of 80cm (3 each on ridges, slopes and valleys), and collected 61 parent material samples. We analyzed total major elements by lithium borate flux fusion and XRF, and trace elements by ICP-MS. We determined P forms by sequential extraction with NaHCO3, NaOH, and HCl, and analyzed both inorganic and organic P in these chemically-defined forms. We used multivariate regression trees to partition the variance in these datasets and elucidate the relative importance of our independent variables.
Results/Conclusions
Volcaniclastic rocks had higher P (~600 ppm) than did quartz diorites (~300 ppm; P<0.001). Similarly, total soil P concentration was two times higher in volcaniclasitic than quartz diorite-derived soils (P < 0.001). For soils on a given rock type, valleys had higher (≤40%) total P than slopes or ridges (P < 0.001). Volcaniclastic-derived soils also had slightly higher NaHCO3 and NaOH-extractable P concentrations, but this difference was small relative to the 2.5 times higher concentrations in valleys than on ridges. The fraction of recalcitrant P (P not removed by NaHCO3, NaOH, or HCl) was 2 times higher on the clay-rich volcaniclastic-derived soils (P < 0.01) and this fraction did not change significantly with slope position (P>0.05). Finally, the loss of P relative to parent material, as measured by indexing to an immobile element, was higher in volcaniclastic than quartz diorite-derived soils (P < 0.001). These three metrics of soil P status indicate that even in strongly weathered tropical soils, parent material can directly influence soil P concentration and shape the soil characteristics (e.g. mineralogy, iron content) that control P losses during soil development. We suggest that the role of parent material in driving differences in P status across tropical ecosystems merits additional investigation.