Grasses and dung were collected in the Serengeti National Park and analyzed for silica content by wet ashing. Grasses from grasslands differing in the grazing intensities experienced were grown in the laboratory in a factorial experiment to determine factors controlling tissue silicification. Concentrations of silica in tissues of plants collected in the field were higher than have been reported for any other plants abundant in grazing ecosystems. Silica contents in the field were higher in more heavily grazed grasslands and in tissue produced earlier in the growing season. Animal dung contained substantial quantities of silica. Laboratory experiments indicated that tissue silicification was increased by defoliation, was higher in plants from more heavily grazed grasslands, varied in different organs and species in patterns confirming current hypotheses about plant defense, and was affected by the availability of soluble silica in the nutrient medium. Silica in the nutrient medium promoted the yield of unclipped plants substantially. Total yield was 18% higher than that of control plants, although hydroponic experiments with solutions prepared and handled in plastic indicated that silica was not a growth requirement, except, perhaps, at the ecologically unrealistic concentrations that might result from reagent contamination. Yield stimulation by silica was differentially distributed among organs, tending to promote photosynthetically active tissues and crowns. Flowering of one species was promoted by silica. Leaves of silica-fed plants were larger. Leaf blade chlorophyll concentrations were 15% higher in silica-fed plants from the more heavily grazed grasslands. The results suggest that complex patterns of grass silicification had a role in the radiation of grazing animals and grasses and may contribute to maintaining the biotic diversity of contemporary grassland-savanna ecosystems by influencing the partitioning of forage species and organs among grazers. Growth promotion by silica may be due to the substitution of mineral support for carbon-based support associated with the deposition of silica in the intercellular spaces of aerial tissues. Since soils of the Serengeti region commonly have pH levels above neutrality, where the availability of silica is low, silica supply could influence primary productivity and resultant energy and nutrient flow through the trophic web in the native environments of the plants.

Plants from four populations of three species of African grasses were collected from grasslands in Tanzania's Serengeti National Park that differ in the grazing intensity that they experience. Plants were grown in the laboratory in a factorial experiment in which variables were plant origin, species identification ofplants, defoliation intensity, and supply of soluble silicate in the nutrient medium. All plants accumulated silica in leafblades in the absence of soluble silicate from the nutrient medium. Plants native to the more heavily grazed grassland. accumulated more silica in their leaf blades than did plants from the less heavily grazed site. Blade silica content was higher-when plants were defoliated, indicating that silicification is an inducible defense against herbivores. The quantitative heterogeneity of this qualitatively homogeneous plant defense system may have contributed to the evolution of high species diversity in the grazing fauna.Grasses and grazing mammals have had an extended, intimate evolutionary history from their simultaneous appearance in the Eocene to their spectacular adaptive radiations during the Pliocene and Pleistocene (1-4). The evolution of hypsodont dentition during the faunal radiations provides presumptive evidence that plant leafsilicification was a major evolutionary trend in the Poaceae (5,6). In addition to silica's deleterious effects upon herbivores due to the mechanical abrasion of enamel surfaces (7), ingestion of plant silica is involved in the etiology of several pathological conditions in animals (8, 9). Adaptive radiation in grazing mammals and grasses was fundamentally different from the radiations of dicotyledonous plants and herbivorous insects (10). The latter were characterized by structural and chemical complexity ofthe plants and sophisticated feeding modes, chemical tolerances, and detoxification mechanisms within the animals (11)(12)(13)(14). Grasses, in contrast, are comparatively simple structurally and lack the diverse secondary chemistry characteristic of dicotyledons. They appear to be a comparatively uniform food source by comparison with dicotyledons. How did such comparatively simple anti-herbivore defenses as silicification of grass leaves lead to the adaptive radiation of a highly diverse grazing mammal fauna? We here-report an experiment with grasses native to an unmanaged ecosystem supporting some semblance of the Pleistocene megafauna. The results indicate that the diversity ofgrasses and the mammals that feed upon them may have been influenced by the quantitative heterogeneity ofa qualitatively homogeneous plant defense system. MATERIALS AND METHODSThe experimental design was a 24 factorial with the following variables: (i) Origin: plants were obtained from the heavily grazed (15-17) short grasslands of the southeastern Serengeti National Park in Tanzania and from a comparatively lightly grazed (15)(16)(17) Silica in leaf blades was determined by wet ashing and weighing (20). Data are expressed as a percentage of leaf dry weight. ...