The idea that species diversity is causally related with ecosystem performance has stimulated a growing number of manipulative experiments in terrestrial and marine systems. These experiments may confound effects of species number and community composition. The effects of diversity can be due to the number of species per se (or other functional/taxonomic categorisation), to the probability of including species with particularly important traits (sampling effect), or to an interaction between both attributes. We test these hypotheses in a field experiment conducted in Helgoland island (German Bight, North Sea), in which we manipulated the number of functional groups and assemblage composition. Artificial substrata (PVC) made up of 16 (2.5 χ 2.5 cm) settlement panels were exposed at a range of depths and sites to natural colonisation for 6 months. Colonisers were categorised into 4 functional groups and panels rearranged to create assemblages of defined functional diversity. Diversity treatments (2 levels) consist of (i) 4 monocultures (one per functional group) and (ii) 1 mixed culture including all groups. Diversity treatments were fully crossed with two density treatments (100 and 50 % species cover). Clearance rate, biomass production, and community composition are measured from January until June 2007 to assess temporal stability of the dependent variables. Our result will help to identify the mechanisms associated with ecosystem responses to changes in biodiversity.