Abstract: Aims With a few exceptions, most well-known field biodiversity experiments on ecosystem functioning have been conducted in plant communities (especially grasslands) in which different numbers of species are planted as treatments. In these experiments, investigators have either kept the total seed weight or seed number constant across treatment plots. However, although in some cases attempts have been made to randomly choose species for planting from a designated species pool, the issue of possible 'hidden treatments' remains unsolved. Particularly, the total and relative abundance among species and across treatments could still affect the results. This study aims to determine whether treatments related to planted seed abundance and seed size may contribute to observed productivity.
Methods We re-analyzed data from four biodiversity experiments based on a common seeding design (i.e. diversity treatments).
Important findings We show that diversity (richness) treatments usually involve a hidden treatment related to the planted seeds (i.e. weight, number and seed size) that ultimately affect plant density. Thus, the un-intended hidden treatment of seeding more seeds on more diverse plots contributes to the productivity to some degree. Such derivative but often neglected hidden treatments are important for further improvement of experimental design and have significant implications in ecological restoration.

Key words: experimental design, productivity, restoration, seed number, seed weight