Savannas currently occupy a fifth of the earth's land surface and are predicted to expand in the next few centuries at the expense of tropical forests, mainly as a result of deforestation and human fires. Can such a vegetation trend impact, through changes in plant Si cycling, the lithogenic silicon (LSi) release into soils (through chemical weathering) and the net dissolved Si (DSi) outputs from soils to stream water (through chemical denudation)? The first step of an investigation requires quantifying the net Si fluxes involved in the plant/soil system. Here, a schematic steady-state Si cycle, established for a tropical humid savanna (Lamto, Ivory Coast) that developed on a ferruginous soil and is subjected to annual fires, is presented. Erosion was assumed to be insignificant. LSi and biogenic Si (BSi under the form of phytoliths) pools were measured, and Si fluxes were estimated from Si concentrations and mass balance calculation. Identification of plant and soil phytoliths indicated that the soil BSi pool is in equilibrium with the current BSi input by the savanna. In the soil column, mixing between a young rapidly recycled BSi pool and an old stable BSi pool is attested by a mixing line equation. Storage of the old BSi pool is assimilated as a BSi output from the plant/soil system. A BSi output additionally occurs after annual fires, when ashes are exported. Both BSi outputs decrease as much the BSi dissolution. In order to uptake constant DSi flux, the savanna increases by three to eight times the net LSi release, depending upon the post-fire ash exportation scenario. A comparison between savanna and rainforest Si cycles that maximizes the differences in plant/soil systems and minimizes differences in climate is presented. The comparison revealed that BSi storage is higher in the savanna soil than in the rainforest soil, mainly due to BSi production that is twice higher in the savanna (127 vs 67 kg/ha/yr). The resulting LSi release that is enhanced by plant uptake is more than 1.5 higher in the savanna than in the rainforest (from 33 to 85 kg/ha/yr in the savanna vs 21 kg/ha/yr in the rainforest). On the contrary, DSi output from soils to stream water, which is not controlled by plant Si cycling but more likely by the soil hydrological regime (or meteoric weathering), is close to twice as high in the rainforest/ferrallitic soil ecosystem (16 vs 9 kg/ha/yr). This case study suggests that the predicted expansion of savannas at the expense of forests should significantly increase DSi uptake by plants, BSi storage in soils, BSi output with ash exportation, and, hence, LSi release through chemical weathering, without direct impact on DSi outputs from soils to stream water. Tracks for further assessing the role of plant Si cycling on chemical weathering, Si and C cycles were suggested: 1) estimates of BSi fluxes that were wrongly based on the assumption that the amount of DSi leached out from soils is linked to the magnitude of plant Si cycling and/or to BSi concentration in soils should be reappraised and 2) changes in the magnitude of plant Si cycling should be accounted in geochemical carbon cycle models, for one of the plant-induced weathering mechanisms.
Read full abstract