A detailed study of the variation in productivity across a diversity gradie
nt in an experimental Mediterranean grassland examines the effects of a dom
inant perennial grass species upon the overall diversity-productivity relat
ionship. The experiment took place at the Greek site of the European-wide B
IODEPTH programme. The experimental design is characterized by the use of a
number of communities containing annuals and perennials within the total s
et of manipulated plots. The main results are: 1) a log-linear relationship
between diversity and productivity exists in Mediterranean grasslands synt
hesized by annuals only, 2) in mixed communities where multiple growth form
s coexist, the performance of a dominant or keystone species may reverse or
hide the diversity-productivity pattern of a functional or growth form gro
up of species taken separately, and 3) the introduction of the dominant gra
ss in the low-diversity mixtures creates an 'inverted' sampling effect whic
h can produce as an artefact a constant productivity response across the di
versity gradient.