Explaining spatiotemporal variability in metacommunities is challenging because conducting empirical studies that link mechanisms across scales is difficult. We developed a stochastic model of epilithic algal succession in streams to better understand the underlying interactions that drive patch dynamics at three levels of spatial hierarchy: patch, channel habitat, and reach. The state of an algal patch is specified by one of nine community types, which are defined by growth form. Ninety-two empirical data sets of epilithic succession were used to estimate Markov transition probabilities among patch types as functions of inputs for light, nutrients, and current. Two additional probability matrices determine the effects of herbivory and disturbance based on user inputs. Testing model predictions of patch composition against independent empirical data indicated good correspondence based on resemblance metrics. Early in succession, most patches were occupied by prostrate and erect diatoms. These persisted under low light and nutrient conditions. Motile and stalked diatom patches characterized later stages when resources were moderate. Succession proceeded to mostly filamentous green algal patches under high resource levels. Sensitivity simulations indicated light most affected succession. Herbivory and disturbance increased the probability of prostrate diatom patches. We then used the model to examine how changes in abiotic and biotic parameters affect patch dynamics at different scales. Changes in nutrients or light affected patch diversity differently at the habitat vs. reach scale. Interactions with local resource levels within a channel habitat determined its alpha diversity. The proportion of channel habitat types within a reach, and their collective response to changes, determined beta diversity. Patch diversity within channel habitats generated by herbivory followed the predation hypothesis only when nutrients were low. At the reach scale, herbivory always increased patch similarity among channel habitats, lowering beta diversity. Diversity within channel habitats followed the intermediate disturbance hypothesis for a variety of habitats. Effects of disturbance on beta diversity at the reach scale depended on channel habitat heterogeneity. The stability of patches at the reach scale was highest when disturbance created high patch diversity, supporting the insurance hypothesis. The ideas generated can improve stream bioassessment methods and provide insight into mechanisms underlying patch dynamics in other ecosystems.