In this paper, I present a method developed to detect ‘extra’, or novel, changes in diversity in a naturally fluctuating environment. When comparing samples in order to evaluate changes in community structure, the sampling procedure will inevitably induce randomness in observed species composition and abundance, so two samples may show considerable differences, even if they come from exactly the same community. Sometimes the effort put into the sampling varies as well, leading to the expectation of a further increase in difference. Finally, if there were a temporal distance between the samples, we would like to correct for variations in species abundance occurring naturally due to fluctuating environmental conditions. The test method presented here includes a model of the sampling procedure, and corrects for differential sampling efforts. The population dynamics is modelled by a diffusion process, its variance mimicking the effect a fluctuating environment has on species abundances. We are thereby able to test the null hypothesis of no unnatural change in diversity, against the alternative of additional changes in community structure due to for example human disturbances.