This paper presents distributed adaptation techniques for use in wireless sensor networks. As an example application, we consider data routing by a sensor network in an urban terrain. The adaptation methods are based on ideas from physics, biology, and chemistry. All approaches are emergent behaviors in that: (i) perform global adaptation using only locally available information, (ii) have strong stochastic components, and (iii) use both positive and negative feedback to steer themselves. We analyze the approaches’ ability to adapt, robustness to internal errors, and power consumption. Comparisons to standard wireless communications techniques are given.