Using Markov models to examine mechanisms of interaction among multiple invasions on Robinson Crusoe Island, Chile

How can we best understand the responses of complex systems to environmental impact? Traditional snapshot approaches (e.g. ANOVA) identify differences in response, but may not identify the relative roles of underlying mechanisms. Here we propose an alternative approach; comparing Markov chain community models developed under impacted and control conditions. By comparing changes in transition patterns among treatments, Markov models may provide more sensitive indicators of functional change and functional pathways in communities. We illustrate this use of Markov models with our experiments on invasion on Robinson Crusoe Island, Chile. Because of seed dispersal, grazing, and disturbance, we predicted that the exotic European rabbit (Oryctolagus cuniculus) facilitates epizoochorous exotic plants at the expense of native plants. On the contrary, a standard MANOVA on our three-year field experiment found reduced exotic abundance and increased dominant native abundance in the presence of rabbits, but the MANOVA could not reveal underlying causes. To explore the mechanisms behind the change, we estimated parameters with Markov models by repeated intensive sampling of subplots through time. This modeling revealed a significant change in the pattern of transitions in response to treatment. Detailed examination of the changes in transitions between treatments shows grazing impacts all species groups, but surprisingly, the dominant native species colonized more bare space in the presence of rabbits whereas exotic plants colonized bare space less often in the presence of rabbits. Models manipulated to eliminate all exotics suggested that non-dominant natives were generally negatively affected by exotic invasions, even though they tended to be facilitated by rabbits in the presence of exotic plants. These results suggest that selective grazing, soil disturbance and competitive interactions overwhelm epizoochorous interactions. Consequently, analysis of Markovian chain community models provided us a rich and surprising insight into the mechanisms of this multiply invaded grassland community.