Trinidadian streams are naturally divided by waterfalls. Below them live pike cichlids and other predators; above them, only much smaller fish. John Endler and later David Reznick used this geography as a ready-made experiment, transplanting guppies from high-predation to low-predation reaches and returning to sample the descendants.
The results appeared quickly. Where predation was relaxed, males developed larger and more numerous colour spots, since female preference for bright males was no longer offset by the cost of being conspicuous. Life history shifted too: transplanted populations matured later, at larger size, and produced fewer, larger offspring. Changes of this magnitude were detected within roughly two to eleven years, and were confirmed as genetic by rearing later generations in common laboratory conditions.
The remaining question is generality. It is not clear how much these rates of change, measured in small isolated streams with strong selection, transfer to large populations under weaker or more variable selective pressure.