The dynamics of pervasive ecosystems are typically highly unpredictable, and therefore self-organising approaches are often exploited to make their applications resilient to changes and failures. The SAPERE approach we illustrate in this paper aims at addressing this issue by taking inspiration from natural ecosystems, which are regulated by a limited set of "laws" evolving the population of individuals in a self-organising way. Analogously, in our approach, a set of so-called eco-laws coordinate the individuals of the pervasive computing system (humans, devices, signals), in a way that is shown to be expressive enough to model and implement interesting real-life scenarios. We exemplify the proposed framework discussing a crowd evacuation application, tuning and validating it by simulation.