Suding et al. (2004) demonstrate how conceptual advances in alternative ecosystem states theory have led to a greater understanding of why degraded systems are often resilient to restoration management. In their review they pose one (of several) ‘outstanding’ questions (Box 3 in Suding et al. 2004): “Are there predictable characteristics that indicate when a system will follow a successional pathway and/or that indicate the presence or absence of alternative ecosystem states?” We suggest that the persistence of alternative stable states might be predicted from simple consideration of assembly rules for systems structured along a gradient of environmental adversity. We raise the hypothesis that strongly abiotically- or disturbance-structured assemblages, with nonrandom trait under-dispersion (Weiher and Keddy 1995), are more likely to exhibit catastrophic phase shifts in community structure than assemblages which are weakly structured by environmental adversity.