It has been widely hypothesised that the productivity of an ecosystem affects the number of species that it can support. Despite decades of study, the nature, extent, and underlying mechanisms of this relationship are unclear. One suggested mechanism is the “more individuals” hypothesis (MIH). This proposes that productivity controls the number of individuals in the ecosystem, and that more individuals can be divided into a greater number of species before their population size is sufficiently small for each to be at substantial risk of extinction. Here, we test this hypothesis using REvoSim: an individual-based eco-evolutionary system that simulates the evolution and speciation of populations over geological time, allowing phenomena occurring over timescales that cannot be easily observed in the real world to be evaluated. The individual-based nature of this system allows us to remove assumptions about the nature of speciation and extinction that previous models have had to make. Many of the predictions of the MIH are supported in our simulations: rare species are more likely to undergo extinction than common species, and species richness scales with productivity. However, we also find support for relationships that contradict the predictions of the strict MIH: species population size scales with productivity, and species extinction risk is better predicted by relative than absolute species size. Furthermore, we show that the scaling of species richness with productivity depends upon the ability of species to partition niche space. Consequently, we suggest that the MIH is applicable only to ecosystems where niches are not already saturated with species, and where partitioning is therefore still possible. Some hypotheses regarding patterns of biodiversity implicitly or explicitly overlook niche theory in favour of neutral explanations, as has historically been the case with the MIH. Our simulations demonstrate that niche theory exerts a control on the applicability of the MIH. --

This datasets consists of data collected from the text files output by the REvoSim eco-evolutionary system. These text files were processed using several scripts written in the Python programming language, which are available as supplementary information alongside the related manuscript.

The term "Evenness", referred to in these data, describes the Shannon's Evenness Index of the population sizes of all species sampled in the simulation in question.

Some values are missing at low energy levels from the analysis of mean individual fitness. This is because, as noted in the data, these simulations contained no sampled species and, as such, did not have a mean individual fitness value.