This is a dataset comprises three samples randomly drawn from three simulated communities. Every community consists of 100 species with approximately 100,000 individuals following a log-normal distribution (mean = 6.5, SD = 1). Setting the first community as control group, the second and third community shared a total of 25 and 50 species with the control. A more detailed description of the control and scenario groups can be found in Zou and Axmacher (2021). The share dataset represents a random subsample of 100, 150 and 200 individuals from three three communities, containing 58, 57 and 74 species, respectively.