Moderate patchiness optimizes heterogeneity, stability, and beta diversity in mesic grassland.

Heterogeneous disturbance patterns are fundamental to rangeland conservation and management because heterogeneity creates patchy vegetation, broadens niche availability, increases compositional dissimilarity, and enhances temporal stability of aboveground biomass production. Pyrodiversity is a popular concept for how variability in fire as an ecological disturbance can enhance heterogeneity, but mechanistic understanding of factors that drive heterogeneity is lacking. Mesic grasslands are examples of ecosystems in which pyrodiversity is linked strongly to broad ecological processes such as trophic interactions because grazers are attracted to recently burned areas, creating a unique ecological disturbance referred to as the fire-grazing interaction, or pyric herbivory. But several questions about the application of pyric herbivory remain: What proportion of a grazed landscape must burn, or how many patches are required, to create sufficient spatial heterogeneity and reduce temporal variability? How frequently should patches burn? Does season of fire matter? To bring theory into applied practice, we studied a gradient of grazed tallgrass prairie landscapes created by different sizes, seasons, and frequencies of fire, and used analyses sensitive to nonlinear trends. The greatest spatial heterogeneity and lowest temporal variability in aboveground plant biomass, and greatest plant functional group beta diversity, occurred in landscapes with three to four patches (25%-33% of area burned) and three- to four-year fire return intervals. Beta diversity had a positive association with spatial heterogeneity and negative relationship with temporal variability. Rather than prescribing that these results constitute best management practices, we emphasize the flexibility offered by interactions between patch number and fire frequency for matching rangeland productivity and offtake to specific management goals. As we observed no differences across season of fire, we recommend future research focus on fire frequency within a moderate proportion of the landscape burned, and consider a wider seasonal burn window.