Climate change is happening, that much is certain. But in what way? How much? And with what impacts? Complex, non-linear models try to predict trends and patterns, but, inevitably, each model is different, parameters are difficult to estimate and the precise impacts remain uncertain. We live in an uncertain world, one where the knowledge about both likelihoods and outcomes remains uncertain.

The striving for increased predictive power over the consequences of climate change has yielded results in the past few decades. We clearly know a lot more than we did. But this is not enough to allow climate science to inform people to direct the future. While global circulation models and forecasting approaches will improve with better technology, more empirical data and faster numbercrunching capacities, the nature of climate–ecosystem interactions is such that non-linearity and the complexity of dynamic interactions means that uncertainty will always be present.

This article explores the implications of this, drawing lessons from the drylands of Africa, where non-equilibrium thinking has challenged conventional approaches to pastoral development.