8. Biodiversity and Ecosystem Processes
127
Relationships consistent with Fig. 1b, and the models discussed in Section 2.1.2,
between plant species richness (or the variety of plant functional types) and ecosystem processes within a site have now been demonstrated in major field experiments
at Cedar Creek (Tilman et al. 1996, 1997b). A similar experiment in California
(Hooper and Vitousek 1997, 1998) varied functional group richness from 1-4. Although some combinations of plant functional groups were complementary, group
composition was more important than the number of functional groups, consistent
with the idiosyncratic hypothesis (Fig. 1c). Hooper and Vitousek did not vary plant
species richness, and used only 2 or 3 species per functional group. Manipulating a
small number of species and functional groups in this way is statistically likely to
yield idiosyncratic responses, for the reasons explained in Section 2.1.3.
The first experiment to directly test the relationship between species richness
and ecosystem processes in ecological communities used a controlled environment
facility (the Ecotron at Silwood Park) (Naeem et al. 1994, 1995; Lawton et al.
1998), and manipulated both plant and animal diversity. In the Ecotron experiment, only primary production (above-ground biomass) increased with species richness. The sampling effect contributes strongly to this relationship (Naeem et al.
1995), but more subtle processes may also be involved (Naeem et al. 1996). Other
ecosystem processes were not consistent with the rivet hypothesis (presumably for
the reasons laid out in Section 2.1.3), but either showed no effect of richness (litter
decomposition) or conformed to the idiosyncratic hypothesis (there were significant differences between treatments in nutrient cycling, but not simply related to
spec;es richness).
Most recently, again using experimentally synthesised plant communities in
laboratory and field microcosms, van der Heijden et al. (1998) have moved the
problem to another trophic level, and examined the influence of mycorrhizal fungal
diversity on ecosystem processes. Once again, exactly as in Fig. 1b, plant biomass
rises curvilinearly with increasing species richness of arbuscular mycorrhizal fungi
(AMF); soil phosphorus concentrations do the opposite, and decline with increasing fungal richness. The authors explicitly explain their results in terms of niche
complementarity: "increasing AMF biodiversity resulted in more efficient exploitation of soil phosphorus and to better use of the resources available in the system"
(page 70).
Species richness has also been shown to buffer and stabilise ecosystem processes, exactly as predicted by the insurance hypothesis (Section 2.1.4), in plant
communities at Cedar Creek (Tilman and Downing 1994; Tilman et al. 1998) and
the famous Park Grass plots at Rothamsted (Dodd et al. 1994), in zooplankton
assemblages (Frost et al. 1995) and in pioneering, three-trophic level experimental
communities made up of bacteria and protists (McGrady-Steed et al. 1997; Naeem
and Li 1997).
Finally, as predicted in Section 2.1.5, correlations between ecosystem processes
and plant species richness based on across-site comparisons differ from the withinsystem results. On a series of Baltic islands, ecosystem processes declined with
increasing plant diversity (Wardle et al. 1997a) (e.g. Fig. 3), because of over-riding
inter-island differences in fire regimes and post-fire histories.
127
Relationships consistent with Fig. 1b, and the models discussed in Section 2.1.2,
between plant species richness (or the variety of plant functional types) and ecosystem processes within a site have now been demonstrated in major field experiments
at Cedar Creek (Tilman et al. 1996, 1997b). A similar experiment in California
(Hooper and Vitousek 1997, 1998) varied functional group richness from 1-4. Although some combinations of plant functional groups were complementary, group
composition was more important than the number of functional groups, consistent
with the idiosyncratic hypothesis (Fig. 1c). Hooper and Vitousek did not vary plant
species richness, and used only 2 or 3 species per functional group. Manipulating a
small number of species and functional groups in this way is statistically likely to
yield idiosyncratic responses, for the reasons explained in Section 2.1.3.
The first experiment to directly test the relationship between species richness
and ecosystem processes in ecological communities used a controlled environment
facility (the Ecotron at Silwood Park) (Naeem et al. 1994, 1995; Lawton et al.
1998), and manipulated both plant and animal diversity. In the Ecotron experiment, only primary production (above-ground biomass) increased with species richness. The sampling effect contributes strongly to this relationship (Naeem et al.
1995), but more subtle processes may also be involved (Naeem et al. 1996). Other
ecosystem processes were not consistent with the rivet hypothesis (presumably for
the reasons laid out in Section 2.1.3), but either showed no effect of richness (litter
decomposition) or conformed to the idiosyncratic hypothesis (there were significant differences between treatments in nutrient cycling, but not simply related to
spec;es richness).
Most recently, again using experimentally synthesised plant communities in
laboratory and field microcosms, van der Heijden et al. (1998) have moved the
problem to another trophic level, and examined the influence of mycorrhizal fungal
diversity on ecosystem processes. Once again, exactly as in Fig. 1b, plant biomass
rises curvilinearly with increasing species richness of arbuscular mycorrhizal fungi
(AMF); soil phosphorus concentrations do the opposite, and decline with increasing fungal richness. The authors explicitly explain their results in terms of niche
complementarity: "increasing AMF biodiversity resulted in more efficient exploitation of soil phosphorus and to better use of the resources available in the system"
(page 70).
Species richness has also been shown to buffer and stabilise ecosystem processes, exactly as predicted by the insurance hypothesis (Section 2.1.4), in plant
communities at Cedar Creek (Tilman and Downing 1994; Tilman et al. 1998) and
the famous Park Grass plots at Rothamsted (Dodd et al. 1994), in zooplankton
assemblages (Frost et al. 1995) and in pioneering, three-trophic level experimental
communities made up of bacteria and protists (McGrady-Steed et al. 1997; Naeem
and Li 1997).
Finally, as predicted in Section 2.1.5, correlations between ecosystem processes
and plant species richness based on across-site comparisons differ from the withinsystem results. On a series of Baltic islands, ecosystem processes declined with
increasing plant diversity (Wardle et al. 1997a) (e.g. Fig. 3), because of over-riding
inter-island differences in fire regimes and post-fire histories.