Building on the concept of plants as ecosystem engineers, and on publi
shed information on effects of particular plant species on soils, we r
eview the evidence that such effects can provide a positive feedback t
o such plants. Based on case studies involving dune formation by Marra
m grass, N supply by N-2-fixing plants, depression of N availability b
y ericaceous plants, 'islands of fertility' in deserts, mull- and mor-
forming temperate forest trees, and formation of peatbogs, as well as
similar other cases, we conclude that there is strong evidence for pla
nt-soil feedbacks in a variety of ecosystems. We argue, moreover, that
these feedbacks could have played a role in the evolution of the plan
t species in question. These ideas are based mainly on correlative obs
ervations, and need further testing.