Spatial and temporal variation in interactions among plants, other species and the abiotic environment create context-dependency in vegetation pattern. We argue that we can enhance understanding of context-dependency by being more explicit about the kinds of direct interactions that occur among more than two living and non-living entities (i. e., third through nth parties) and formalizing how their combinations create context-dependency using simple conceptual models. This general approach can be translated into field studies of context-dependency in communities by combining: progressive sampling of local variation in vegetation pattern that encompasses variation in combinations of direct interactions; spatial and temporal measures of these direct interactions; locally parameterized versions of the conceptual models; and appropriately scaled experiments.