VARIATION IN HERBIVORE EFFECTS ON GRASSLAND PRIMARY PRODUCTION: A TEST OF THREE MODELS

Grazing animals influence a wide range of plant and soil processes in the world’s grasslands. Ecologists have long understood that grazing can stimulate aboveground net primary production (ANPP), although this phenomenon has not been broadly generalizable across grasslands and grazing regimes. The mechanisms underlying grazer stimulation of ANPP are therefore of interest to a wide variety of stakeholders from ecologists to land managers. Three data-supported hypotheses offer differing explanations for the ways in which grazing interacts with resource availability to drive ANPP: the compensatory continuum hypothesis (CCH) implicates background resource availability, the limiting resource model (LRM) considers the direct effects of grazers on resource availability, and the episodic herbivory model (EHM) considers the indirect effects of grazers on resource availability brought about by the removal of standing biomass. No studies have yet compared these three models to measure their relative influence on ANPP. I conducted a paired-plot defoliation experiment in two distinct grassland community types (mesic and dry) in Yellowstone National Park to test how well each model explained variation in ANPP over two months of the growing season. I simulated an average Yellowstone grazing intensity (50% biomass removed monthly) and measured plant-available N and soil moisture as indices of resource availability. I also collected data on ANPP and resource availability in plots grazed by bison to study support for each model under a natural grazing regime. Clipping increased relative growth rates in each grassland type by over 100% but had no effect on N availability or soil moisture. Clipping in June increased ANPP in mesic but not dry grassland, supporting the CCH at the landscape scale. Within mesic grassland there was support for the EHM, LRMmoist and CCHN, but the EHM explained over twice as much variation in the clipping effect on ANPP than the next best model. In dry grassland, the EHM was the only model with support. The poor predictive power of alternative models in dry grassland may be due to their neglect of resource colimitation, which in a previous study was found to interact with grazing to influence ANPP in this system. Effects of ungulate grazing on ANPP varied between grassland types, and grazer-stimulation of ANPP was accompanied by increases in relative growth rate and N availability, supporting the LRM. Relative growth rate was negatively related to standing biomass in both clipped and grazed plots in accordance with the framework of the EHM. The clipping study empirically demonstrates for the first time a key tenet of the EHM, that herbivory can increase ANPP without any direct effect on resource availability. Both the clipping and grazing studies suggest that the three models considered are not independent and can theoretically operate simultaneously. Considered in full, this thesis highlights the diversity of ways by which grazing by wild, migratory, ungulates can stimulate ANPP