There is currently a great deal of interest in science curricula in which student learning is organized around a single overarching task. One claim is that this approach can lead to a better understanding of scientific content, where content is understood in its original, narrow sense. As researchers, we would like to know if these claims about student content learning actually hold. Ideally, we would be able to map out students' changing conceptual ecologies. However, such work is theoretically and methodologically difficult. To make this task feasible, we focus on the recurrent functional patterns of knowledge, which we refer to as modes. Using "mode-sensitive" clinical interviews with middle school students working with the "I, Bio" curriculum, we demonstrate how analysis at the level of modes can capture the landscape of a conceptual ecology, and provide a language for describing the broad sweep of conceptual changes that result from task-structured instruction.