Herbivory is a major energy transfer within ecosystems; an open question is under what circumstances it can stimulate aboveground seasonal primary production. Despite multiple field demonstrations, past theory considered herbivory as a continuous process and found stimulation of seasonal production to be unlikely. Here we report a new theoretical model that explores the consequences of discrete herbivory events, or episodes, separated in time. We discovered that negative density (biomass) dependence of plant growth, such as might be expected from resource limitation of plant growth, favors stimulation of seasonal production by infrequent herbivory events under a wide range of herbivory intensities and maximum plant relative growth rates. Results converge to those of previous models under repeated, short-interval herbivory, which generally reduces seasonal production. Model parameters were estimated with new and previous data from the Serengeti ecosystem. Patterns of observed frequent and large magnitude stimulated production in these data agreed generally with those predicted by the episodic herbivory model. The model thus may provide a new framework for evaluating the sustainability and impact of herbivory.

Biomass measurements made once, on average 28 days apart, in each of consecutive months April, May, and June in 2000, 2001, and 2002 in three 4 x 4 m fenced plots at each of seven sites from the Long-Term Grazing Exclosure (LTGE) experiment. Four 25 x 25 cm quadrats were clipped to the ground surface in each 4 x 4 m plot, with green material sorted to stem versus leaf and green separated from litter (gray plant material) Different quadrats were clipped from different locations within plots in each month. Green material was air-dried at 45^oC for a week and then weighed. Sites varied in mean annual rainfall from 490 – 890 mm/yr , soil N from 0.05% to 0.22% and soil P from 0.005% to 0.15%