The inevitability of surprise in agroecosystems

Many critical transformations of ecosystems contain advanced signals of their imminence, but it is also true that many critical transformations can be shown to contain no such signal, at least with the sorts of data normally available to field workers. This paper explores some generalized theoretical structures and distinguishes between those that may provide a signal that could be used to predict a critical transformation and those that, by their very nature, do not provide such a clue. I argue that it is almost certain that such completely unpredictable structures will sometimes emerge from systems that tend to be as complicated as agroecosystems, in both their natural and social sciences manifestations. Precaution should thus be taken when designing agroecosystems.