The core principle shared by most theories and models of succession is that plant-environment (PE) feedback dynamics drive a directional change in the plant community, following a major disturbance. The most commonly studied feedback loops are those in which the regrowth of the plant community causes changes to the biotic (e.g., dispersers) or abiotic (e.g., soil nutrients) environment, which differentially affect species availability or performance. This, in turn, leads to shifts in the species composition of the plant community. However, there are many other PE feedback loops that potentially drive succession, each of which can be considered a model of succession.While plant-environment feedback loops in principle generate predictable successional trajectories, succession is generally observed to be highly variable. Factors contributing to this variability are the stochastic processes involved in feedback dynamics, such as individual mortality and seed dispersal, and extrinsic causes of succession, that are not affected by changes in the plant community but do affect species performance or availability. Both can lead to variation in the identity of dominant species within communities. This, in turn, leads to further contingencies if these species differ in their effect on their environment (priority effects). Predictability and variability are thus intrinsically linked features of ecological succession. We present a novel conceptual framework of ecological succession that integrates the propositions discussed above. This framework defines seven general causes: landscape context, disturbance and land-use, biotic factors, abiotic factors, differential species availability and performance, and the plant community. When involved in a feedback loop, these general causes drive succession and when not, they are extrinsic causes that create variability in successional trajectories and dynamics. The proposed framework provides a guide for linking these general causes into causal pathways that represent specific models of succession. Our framework represents a systematic approach to identifying the main feedback processes and causes of variation at different successional stages. It can be used for systematic comparisons among study sites and along environmental gradients, to conceptualize studies, guide the formulation of research questions and design of field studies. Mapping an extensive field study onto our conceptual framework revealed that the pathways representing the study’s empirical outcomes and conceptual model had important differences, underlining the need to move beyond the conceptual models that currently dominate in our specific fields and to find ways to examine the importance of and interactions among alternative causal pathways of succession. To further this work, we argue for integrating long-term studies across environmental and anthropogenic gradients, combined with controlled experiments and dynamic modeling.