Rethinking large-scale restoration experiments to improve ecological learning and on-the-ground application

Ideally, restoration should be informed by large “experiments” implemented over many different locations and time frames. This should help generalize outcomes and reveal ecological knowledge. We cautiously applied this approach to restoration of Phalaris arundinacea-invaded wetlands, and stretched institutional and funding capacity to implement as large an experiment as possible. Ecological learning was constrained by keeping the research practical for application, but valuable insight was gained. Exceptions encountered in the field prevented learning using traditional empirical approaches. We advocate increased funding, more integrated institutional support, and novel statistical techniques to “learn by doing” at various temporal and spatial scales.