Hypotheses have recently been formulated to elucidate the relationship between species diversity and ecosystem functioning. Using previously published mechanisms as a starting point we suggest that common mechanisms can be provided for this relationship by using the concepts of niche and trophic-level dynamics. The reasoning is the following: if remaining species within a trophic level can modify their niches as other species disappear, production within the level remains unchanged, whereas in the absence of niche modification production decreases. Decreased production within a trophic level affects biomass and production at other levels as predicted by trophic-dynamic models. Changes in biomass and production finally bring about changes in ecosystem functioning. In the redundant species hypothesis remaining species can modify their niches, and so functioning remains unchanged. In the predictable change hypothesis (our counterpart for the rivet hypothesis), and in the idiosyncratic response hypothesis, remaining species cannot modify their niches, leading to predictable and unpredictable changes in functioning, respectively. Unpredictable changes are due to differences in the characteristics of species and indirect interactions between populations.
We tested the hypotheses and the suggested mechanisms using a soil food web with three trophic levels: microbes, microbivorous nematodes and a predatory nematode. We established one diverse (3 bacterivores and 3 fungivores) and three simple (1 bacterivore and 1 fungivore) food webs and found that differences in trophic-level biomasses between the diverse and simple food webs were idiosyncratic. Unpredictability resulted from differences in microbivore characteristics - their efficiency in resource utilisation and vulnerability to predation and competition. Changes in microbial respiration and total mineralisation of C and N, i.e., system functioning, were also idiosyncratic rather than redundant or predictable when diversity was reduced, although idiosyncracy was not as clear as in the case of trophic-level biomasses. We conclude that predicting the influence of declining species diversity on trophic-level dynamics and ecosystem processes is difficult, at least in food webs with a small initial number of species, unless the characteristics of species and the nature of their interactions are known.