100 yr of primary succession highlights stochasticity and competition driving community establishment and stability.

The study of community succession is one of the oldest pursuits in ecology. Challenges remain in terms of evaluating the predictability of succession and the reliability of the chronosequence methods typically used to study community development. The research of William S. Cooper in Glacier Bay National Park is an early and well-known example of successional ecology that provides a long-term observational data set to test hypotheses derived from space-for-time substitutions. It also provides a unique opportunity to explore the importance of historical contingencies and as an example of a revitalized historical study system. We test the textbook successional trajectory in Glacier Bay and evaluate long-term plant community development via primary succession through extensive fieldwork, remote sensing, dendrochronological methods, and newly discovered data that fills in data gaps (1940s to late 1980s) in continuous measurement over 100+ years. To date, Cooper's quadrats do not support the classic facilitation model of succession in which a sequence of species interacts to form predictable successional trajectories. Rather, stochastic early community assembly and subsequent inhibition have dominated; most species arrived shortly after deglaciation and have remained stable for 50+ years. Chronosequence studies assuming prior composition are thus questionable, as no predictable species sequence or timeline was observed. This underscores the significance of assumptions about early conditions in chronosequences and the need to defend such assumptions. Furthermore, this work brings a classic study system in ecology up to date via a plot size expansion, new baseline biogeochemical data, and spatial mapping for future researchers for its second century of observation.