Background
The functioning and sustainability of ecosystems may depend on their biological diversity (Ehnich and Ehnich 1981, Wilson 1992, NcNaughton 1997, Ewel et al 1991, Vitousek and Hooper 1993, McNaughton 1993, Tilman and Downing 1994, Naeem et al 1994b). Elton's (Elton 1958) hypothesis that more diverse ecosystems are more stable has received much attention (Ehnich and Ehnich 1981, McNaughton 1997 1993, Tilman and Downing 1994, May 1973, Goodman 1975, King and Pimm 1983, Pimm 1984, Tilman ), but Darwin's proposal (McNaughton 1993, Darwin 1859) that more diverse plant communities are more productive, and the related conjectures (Ewel et al 1991, Vitousek and Hooper 1993, Ehnich and Mooney 1983, Swift and Anderson 1993) that they have lower nutrient losses and more sustainable soils, are less well studied(Ewel etal 1991, Vitousek and Hooper 1993, McNaughton 1993, Naeem et al 1994b, Swift and Anderson 1993, Naeem et al 1995).
The diversity-productivity hypothesis is based on the assumption that interspecific differences in the use of resources by plants allow more diverse plant communities to utilize more fully limiting resources and thus attain greater productivity (McNaughton 1993, Naeem et al 1994b, Swift and Anderson 1993, Naeem et al 1995). A related hypothesis is that nutrient leaching losses from ecosystems should be a decreasing function of plant diversity because of greater nutrient capture and/or immobilization in more diverse ecosystems (Ewel et al 1991, Vitousek and Hooper 1993, Swift and Anderson 1993). Taken together, these lead to the diversity-sustainability hypothesis; that the sustainability of soil nutrient cycles and thus of soil fertility depends on biodiversity. But the relationships between biodiversity and ecosystem functioning remain controversial (Vitousek and Hooper 1993, May 1973, Pimm 1984, Naeem et al 1995, Lawton and Brown 1993, Givnish 1994, Tilman et al 1994, Andre et al 1994, Naeem et al 1994a) because existing field data are from studies that lacked direct experimental control of biodiversity or sufficient replication. Only direct control of diversity allows attribution of responses to diversity rather than to other correlated factors (Naeem et al 1994b, Naeem et al 1995, Givnish 1994, Tilman et al 1994, Andre et al 1994, Naeem et al 1994a). Moreover, the species composition of plots must be randomly determined to avoid biases caused by the traits of particular species (Givnish 1994, Tilman et al 1994).
E123 is a well-replicated field experiment, in which species diversity has been directly controlled.