A pertinent debate in plant ecology centers around the generality of the self-thinning rule. However, studies focused on highly simplified settings such as even-aged monospecific populations or optimal conditions. This neglects the fact that most natural communities, to which the classical self-thinning slope is often applied, are age-structured, composed of multiple species and exposed to various types of abiotic stress.

With the help of an individual-based model, we relax these simplified assumptions and systematically test for changes in the biomass–density relationships of uneven-aged, functionally diverse plant communities across a complete stress gradient, using excessive to insufficient soil water as a case study.

We show that frequent recruitment, which resulted in an uneven-aged community, and stress intensity caused predictable changes in the entire biomass–density trajectory. Increasing stress resulted in steeper (more negative) slopes and increased the intercept in the classical self-thinning section irrespective of excessive or insufficient soil water as a stress type. Recruitment steepened the slope, too and enabled a novel section in the biomass–density trajectory. This novel section represented a quasi-steady state of the density-dependent dynamics of new generations which occurred locally within patches of recruitment. At the community level, the slope of the biomass–density relationship at quasi-steady state had a significantly flatter slope of −1.1 under optimal soil water conditions. Functional diversity showed little impact on density-dependent mortality. Namely, it resulted in an earlier onset of mortality but not in changes in the values of the slope and intercept.

We conclude that the classical −3/2 slope is not useful to describe the biomass–density relationship in natural and semi-natural plant communities. The magnitude and direction of variation in the slope are related to the age–structure and abiotic stress intensity in the plant community.

Mean ± SD skewness of the biomass distribution of nineteen soil water scenarios across ten replicates over 2000 years. Changes in the biomass-density trajectory over time can be tracked by means of the skewness of the biomass distribution (Berger et al. 2002). The skewness of the biomass distribution subdivided the biomass-density trajectory into maximum four distinct sections. Logarithm of mean plant biomass of survivors w (mg/m²) and the logarithm of plant density N (individuals/m²) of nineteen soil water scenarios for all ten replicates over 2000 years.