Plant community dynamics governed by red harvester ant (Pogonomyrmex barbatus) activities and their role as drought refugia in a semi-arid savanna

This study examined modifications made by Pogonomyrmex barbatus, by their processes of granivory and nest construction, to forb and grass dynamics under large-scale disturbances of fire, recent drought and long-term, large-mammalian herbivory using comparative studies, field experimental manipulations, and a simulation model on the Edwards Plateau, Texas. Ant nests are refugia for grass survival during extreme droughts as demonstrated during the drought of 1998 to 2002. Significantly greater cover of grasses and lower abundance and cover of forbs was found beside nests compared with surrounding habitat throughout the drought and recovery. Grasses near nests may be the seed source for surrounding habitats during recovery. Seeds were differentially collected among most forbs and grasses despite seed abundance. Harvest was significantly reduced in the fall relative to spring. During preference experiments, harvest differences were found between grazing treatments for two of four species, but only during the spring. High lipid content seeds were unpreferred in fall compared to high protein and carbohydrate content seeds. Granivory influences on seedling establishment were studied by comparing seedling recruitment among sown and naturally occurring seeds excluded and open to foragers. Exclosures were placed in three nest densities and two burn treatments. Seeds in exclosures produced significantly more seedlings than open arenas only during the first year of drought recovery. Densities of grasses and annual forbs were higher in open arenas the second year due to indirect effects of granivory. By reducing seeds ants release seedlings from competition. Sown seedling abundance was unaffected by colony density and fire. Colony density and distribution were influenced by topography, soil types, soil depth and woody cover, but not by historical grazing treatments. Cleared vegetation on nest disks impacted less than 1% of total surface area and losses were compensated by greater basal cover of grasses next to disks compared to surrounding habitats. Foraging areas influenced 17.3-73.6% of surface area and could diminish seed populations for potential seedlings. Model results agree with experimental observations that communities may be modified by P. barbatus presence due to differential responses of grass species to interaction between nests or granivory and rainfall amounts.