The worldwide degradation of coral reef systems often is associated with shifts of the benthic community from coral- to macroalgal-dominance. Transitions to macroalgal-dominance, especially those that are maintained by stabilizing feedbacks, can have profound societal implications, prompting considerable interest in the predictability and reversibility of these undesired shifts. Herbivory by fishes is a key ecological process that can prevent an undesired shift to macroalgae and/or facilitate a return to coral-dominance. Herbivorous fishes are harvested in many small-scale fisheries, setting up the potential for fishers to alter coral-macroalgae interactions and the resilience of alternative reef states. However, in this context, macroalgae, herbivorous fishes, and fishing are often considered as monolithic groups despite the potential for substantial, species-specific differences in their influence. For example, macroalgae species vary in traits that influence growth rates, herbivores vary in their dietary preferences, and fishers can differentially target species of herbivorous fishes. My dissertation research examines the intricacies of these influences in the context of abrupt (non-linear) ecosystem shifts and the resilience of coral reefs. I first applied the competition – palatability paradigm from plant – herbivore theory to assess how the attributes of macroalgae influence their vulnerability to trophic and competitive interactions, and in turn, can strengthen or weaken feedbacks stabilizing a macroalgae state. My second research focus was motivated by the recognition that different functional groups of herbivores play two distinct but complementary roles in controlling macroalgae on coral reefs; grazers can prevent the establishment of macroalgae, but only browsers can remove mature macroalgae. Consequently, I explored how spatial covariation in grazing and browsing rates is linked to the prevention and reversibility of shifts to macroalgae at the local reef scale. Notably, grazing and browsing herbivores have different life history attributes that make them differentially susceptible to overexploitation from fishing. Thus, as a third research focus, I modeled how selective fishing on grazing or browsing herbivores influences the non-linear dynamics of shifts between coral and macroalgae states. My findings provide deeper understanding into key ecological processes and attributes that shape non-linear dynamics of the benthic community on coral reefs and can inform spatially-explicit management strategies to enhance coral resilience in our rapidly changing world.