Phytolith signal of aquatic plants and soils in Chad, Central Africa

To identify the phytolith signal of lacustrine environments, which are prone to preserving faunal remains including hominins, we analyzed the phytolith content of 46 grass and sedge species, and of 26 soil and mud samples. The samples were collected in Chad (Central Africa), in the Sudanian and Sahelian phytogeographical zones, near temporary and permanent water-bodies (including Lake Chad) and in grass-dominated biomes on well-drained soils. Altogether, we observed and counted separately 80 different phytolith types, including 38 grass silica short cells (GSSCs). Phytolith type diversity and relative abundances were analyzed in the botanical specimens to improve the phytolith taxonomic resolution. For the Poaceae, we used a value-test analysis to identify significant cohorts of phytoliths to characterize aquatic, mesophytic, and xerophytic species. Our results show that the abundance of Cyperaceae in swampy areas may be deduced from the combined abundance of blocky and elongate phytolith types, but not by the typical silicified Papillae phytoliths, which were barely found preserved in the soil/mud. The abundance of aquatic Poaceae near water-bodies is inferred from the presence and abundance of a cohort of eight GSSC types (including notably several trapeziform GSSCs within the bilobate, cross, and saddle categories), which averages 42% in the mud samples, but only 23% and 14% in the samples from the Sudanian and Sahelian zones, respectively. The characterization is unclear for mesophytic grasses, but obvious for xerophytic grasses whose abundance in the Sahelian grasslands is inferred from the presence and abundance of a cohort of five GSSC types (mainly tabular saddles), which averages 50% in the soil samples from the arid Sahelian zone, and <19% in the more humid Sudanian and swamp samples. In conclusion, considering the full morphological diversity of grass silica short cell phytoliths (rather than just the broad morphological categories) allows greater discrimination of the aquatic environments. Such approach is therefore required for analyzing vegetation distribution at a local scale.