Phytolith indexes: proxy of grass-dominated biomes in tropical Africa
by
Laurent Bremond
CEREGE, Europôle de Méditerranéen de l'Arbois, BP45, 13545 Aix-en-Provence Cedex 04,France
Coauthors: A. Alexandre (CEREGE), J. Guiot (CEREGE)

Because grasslands are the most widespread continental floral associations, reconstruction of grassland past dynamics would improve global environmental change simulations. Recent vegetation models simulate grass-dominated biomes with difficulties. For example, BIOME 3 (Haxeltine and Prentice, 1996) is not successful at differentiating moist versus dry savannas. In BIOME 6000 (Prentice and Webb, 1998) different grassland biomes needed to be grouped in one biome : the PFTs "warm grass/shrub" and "cold grass/shrub" are merging into a same "steppe" biome as well as "semidesert" and "hot desert" into "desert".

Recent research on phytoliths allows a definition of tropical grass-dominated biomes based on four modern phytolith indexes which are related to the tree cover density and the dominant grass subfamily: - index 1 indicates the proportion of lignous dicotyledone element versus grasses; - index 2 characterizes the proportion of C4-short grasses (Chloridoideae subfamily) versus C4-tall grasses (Panicoideae subfamily); - index 3 characterizes the proportion of C3- versus C4-grasses; - index 4 indicates the proportion of silicified bulliformes cells of grass epidermis related to a water stress. These vegetation features are also linked to climate constrains (e.g. rainfall, temperature, partial pressure of CO2), edaphic constrains and/or human impact. In order to integrate phytolith data in vegetation models we first calibrate the relation between phytolith indexes, vegetation structure and climate parameters. Then, new PFTs are defined and grassland biomes are desegregated.

In this study we work on 63 surface soil samples collected in West Africa along a latitudinal and rainfall gradient (Mauritania and Senegal, collected by A-M. Lézine). Firstly phytolith indexes are confronted with vegetation structure and their bioclimatic boundaries. Secondly, an equation is calibrated by a multiple regression between annual rainfall, a (Priestley-Taylor coefficient, or ratio of annual actual evapotranspiration to the annual equilibrium evapotranspiration) and phytolith indexes. The validity of the equation is attested by high correlation coefficients between estimated and measured values on an independent verification data set.

Such a calibration should help to make phytolith assemblage data more useful for comparison between data and outputs of coupled atmosphere-terrestrial biosphere models.

Date received: April 3, 2001

Copyright © 2001 by the author(s). The author(s) of this document and the organizers of the conference have granted their consent to include this abstract in Atlas Conferences Inc. Document # cagc-94.