Processes controlling silicon isotopic fractionation in a forested tropical watershed : Mule Hole Critical Zone Observatory (Southern India)

2018

Article référencé dans le Web of Science WOS:000428790200016

Riotte Jean, Meunier J. D., Zambardi T., Audry S., Barboni D., Anupama K., Prasad S., Chmeleff J., Poitrasson F., Sekhar M., Braun Jean-Jacques

Geochimica et Cosmochimica Acta, 2018, 228, p. 301-319 ISSN 0016-7037

Assessing the dynamics of the silica cycle in the critical zone remains challenging, particularly within the soil, where multiple processes are involved. To improve our understanding of this cycle in the Tropics, and more specifically the role played by vegetation, we combined elemental Si mass balance with the delta Si-30 signatures of the compartments involved in the waterplant-rock interactions of a tropical forested watershed, Mule Hole (Southern India). To accomplish this, we analysed (1) the d30Si values of present-day litter phytoliths from tree leaves and grass, as well as soil amorphous silica (ASi); (2) the Si isotope fractionation induced by phytolith dissolution; (3) the silicon mass balance inferred from isotopes at the soil-plant scale; and (4) the consistency between water sources and the delta Si-30 signatures in the ephemeral stream. The delta Si-30 values of present-day litter phytoliths and soil ASi vary within a narrow range of 1.10-1.40 parts per thousand for all samples, but two deep vertisol samples which likely trapped phytoliths from different vegetation growing under more humid conditions, as indicated by pollen analysis. A homogeneous signature of litter is a minimum condition for using delta Si-30 as a proxy for the litter/phytolith source of Si. However, litter-ash dissolution experiments demonstrate that the incipient dissolution of phytoliths fractionates Si isotopes, with the preferential dissolution of 28 Si over 30 Si yielding delta Si-30 values as low as -1.41 parts per thousand. Values close to the whole-sample signatures, i.e., above 1 parts per thousand, were recovered in the solution after a few hours of water-ash interaction. At the soil-plant scale, the average delta Si-30 value of soil-infiltrating solutions is slightly lighter than the average phytolith signature, which suggests phytoliths as the source of soil dissolved Si. The isotopic budget of dissolved Si within the soil layer, which was obtained based on previous elemental fluxes, is imbalanced. Equilibrating the isotopic budget would imply that up to 4100 mol ha(-1) yr(-1) of silica is taken up by vegetation, which is almost twice as large as that initially estimated from the elemental budget. The additional Si flux taken up, and likely stored in woody stems, was estimated assuming that Si isotopes followed a steady-state model for the whole Si plant uptake and then followed a Rayleigh model once in the plants. The delta Si-30 value of the additional Si flux taken up should be close to 0 parts per thousand, i.e., enriched in light Si isotopes compared to the litter. If steady-state conditions apply, the source could correspond to soil ASi dissolution or deep (saprolite) root uptake. At the outlet of the watershed, the stream exhibits low delta Si-30 values (0.28-0.71 parts per thousand) during peak flows and high delta Si-30 values (1.29-1.61 parts per thousand) during the recessions at the end of the rainy season. Heavy delta Si-30 signatures are consistent with the expected domination of seepage at the end of floods. The light delta Si-30 values during peak flow are slightly lower than the overland flow signature and reflect either a sampling bias of overland flow or a minor but significant contribution of another Si source within the stream, possibly the partial dissolution of phytoliths from the suspended load, with slight isotopic fractionation. This study confirms that vegetation controls the silicon cycle in this dry tropical forest. It also shows that silicon isotopes yield a better grasp of the mass balance and sources and potential mechanisms involved than the consideration of only silicon concentrations. However, this proxy still relies on working hypotheses, notably steady-state and/or Rayleigh fractionation models, which need to be confirmed in further studies.

Géologie et formations superficielles [064]

INDE ; ZONE TROPICALE

Fonds IRD [F B010072671]

fdi:010072671