Cooperation peaks at intermediate disturbance

Explaining cooperation is a challenge for evolutionary biology [1, 2]. Surprisingly, the role of extrinsic ecological parameters remains largely unconsidered. Disturbances [3, 4] are widespread in nature and have evolutionary consequences [5]. We develop a mathematical model predicting that cooperative traits most readily evolve at intermediate disturbance. Under infrequent disturbance, cooperation breaks down through the accumulation of evolved cheats. Higher rates of disturbance prevent this because the resulting bottlenecks increase genetic structuring (relatedness [6-8]) promoting kin selection for cooperation. However, cooperation cannot be sustained under very frequent disturbance if population density remains below the level required for successful cooperation. We tested these predictions by using cooperative biofilm formation by the bacterium Pseudomonas fluorescens [9, 10]. The proportion of biofilm-forming bacteria peaked at intermediate disturbance, in a manner consistent with model predictions. Under infrequent and intermediate disturbance, most bacteria occupied the biofilm, but the proportion of cheats was higher under less frequent disturbance. Under frequent disturbance, many bacteria did not occupy the biofilm, suggesting that biofilm dwelling was not as beneficial under frequent versus intermediate disturbance. Given the ubiquity of disturbances in nature, these