1009 1019.1020

constant across treatments (10). The micro-cosms were kept under controlled environ- mental conditions for 8 weeks (10). The soilprocesses measured included leaf litter mass loss, leaf litter fragmentation, grossNO– productivity, and soil respiration (CO D. A. Heemsbergen,1,2 M. P. Berg,1 M. Loreau,3 J. R. van Hal,2 cies number on decomposition processes. Sat-uration in process rates occurred after more The loss of biodiversity can have significant impacts on ecosystem func- than one species was added (11), and mix- tioning, but the mechanisms involved lack empirical confirmation. Using soil tures showed a large variation in the mea- microcosms, we show experimentally that functional dissimilarity among sured soil processes. Net biodiversity effects detritivorous species, not species number, drives community compositional were calculated (10, 12) to assess whether effects on leaf litter mass loss and soil respiration, two key soil ecosystem positive or negative interactions among spe- processes. These experiments confirm theoretical predictions that biodiversity cies could explain the observed variation in effects on ecosystem functioning can be predicted by the degree of functional soil processes within a diversity treatment. We observed a range of negative, neutral, andpositive net diversity effects in two- and four- Functional redundancy of species is assumed expected on the basis of the mere additive species treatments (Fig. 1, A and B). For some to be a common feature in soils (1–4), and effects of single species. The nature (in- species combinations Efor example, Lumbricus experimental studies that manipulate species hibitory, neutral, or facilitative) of these interactions might be related to the degree combination E)^, the net diversity effect on of soil processes, in which the asymptote is in which species differ in their impact on soil respiration and leaf litter mass loss was reached at low levels of species number (5, 6).
soil processes. We hypothesized that species higher than expected, suggesting facilita- Even though species number per se does not mixtures that contain species with different tion. For other combinations Efor example, appear to be important, the functional di- effects on ecosystem processes (species that Polydesmus denticulatus and Oniscus asellus versity of the soil community (that is, the are functionally dissimilar as to these pro- (Fig. 1, combination C)^, a lower effect than range of species traits that determine their cesses) show facilitative interactions, ir- expected was observed, suggesting inhibition functional role) may affect ecosystem pro- due to interspecific competition. This shows cesses (5–7). Functional differences may re- taxonomic groups involved. Functional dis- sult in a variety of interactions among species.
number, but different species compositions, Because of the diverse and complex nature of effects of the various species on four eco- had very different effects on soil ecosystem these interactions, it may often be difficult to system process variables: leaf litter mass loss, leaf litter fragmentation, soil respira- when species are lost from or introduced into tion, and nitrification, all of which are re- the community. The central question exam- lated to the process of decomposition. Thus, ined in this paper is whether we can predict instead of focusing on ecological attributes the effects of changes in species composition of species that are associated with their on soil ecosystem processes if the functional functional impact (9), we directly measured dissimilarity of species in the community is their effect on ecosystem processes.
soil microcosms (10) with an increasing num- the decomposition of dead organic matter. It ber of macro-detritivores: zero, one, two, four, is known that species differ in their effects and eight species per microcosm (table S1).
species has a specific mode of affecting litter macrofauna community of a river floodplain fragmentation or nitrification, due to con- (10). Single-species treatments of all eight trasting functional attributes. Moreover, the species were included in the experimental effects of different species on a particular design to quantify their per-capita effects on process often differ in strength. These differ- soil process rates, and these were used to ences may lead to interspecific interactions quantify functional dissimilarity among spe- that result in species mixtures performing cies. To discriminate the effect of species better (facilitative interactions) or worse number from other compositional effects on Fig. 1. Net diversity effect on soil respiration (A) process rates, different two- and four-species and leaf litter mass loss (B) in relation to species number. Each dot represents a treatment mean (table S1). Each species was assigned ran- Vrije Universiteit, Institute of Ecological Science, (n 0 5 per treatment); error bars represent Department of Animal Ecology, de Boelelaan 1085, domly to multispecies treatments with the standard errors. Letters next to the dots refer to 1081 HV Amsterdam, Netherlands. 2Alterra, Wage- following constraints: (i) species were equally the actual species combination given in table S1.
ningen University and Research Centre, P.O. Box 47, represented, and (ii) both two- and four-species A nonsignificant regression between species 6700 AA Wageningen, Netherlands. 3Laboratoire number and soil respiration (linear regression, combinations contained taxonomic group di- 0 1.46, P 0 0.22) and leaf litter mass loss versity Eone versus two taxonomic groups and two versus three taxonomic groups, respec- *To whom correspondence should be addressed.
indicates that negative or positive net diversity E-mail: [email protected] tively (table S1)^. Total earthworm biomass effects were not related to species number.
species mixtures composed of species with strong differences in single-species effects fects on soil processes were explained by the (Fig. 1), indicating that species number per se (Table 1). The epigeic earthworm L. rubellus does not explain the observed net effects.
mixtures. These results suggest that it is Positive net diversity effects occurred in monoculture. Its strong effect on leaf litter not species number but the degree of func- mass loss is a consequence of its ability to tional differences between species that is transport litter to deeper soil layers. The ef- a driver of ecosystem processes, and this fect of the endogeic earthworm Aporrecto- effect in turn is due to facilitative interac- tions among species. The species-specific probably reflected changes in the physical contribution to the range of functional dis- conditions of the soil. Among arthropods, isopod O. asellus significantly fragmented sity generates positive interactions that en- leaf litter into smaller particles. These differ- hance ecosystem process rates. If we know ences in the way different species affect how species contribute to multiple species ecosystem processes are critical to under- interactions in the community, by an analysis of their functional dissimilarities, we may be able to predict the impact of local species loss or biological invasions on ecosystems.
should have stronger effects on process rates This may also have implications for ecosys- than communities consisting of functionally tem restoration, which may require the in- troduction of particular functionally dissimilar species or species combinations into impov- gression of both soil respiration and leaf litter mass loss against mean functional dis- ´n, J. Balandreau, Appl. Soil Ecol. 13, 105 (1999).
2. M. A. Bradford et al., Science 298, 615 (2002).
influence ecosystem processes tend to gen- 3. J. Laakso, H. Seta¨la¨, Oikos 87, 57 (1999).
Fig. 2. Net diversity effect on soil respiration erate facilitation. Facilitation was shown in 4. J. H. Faber, H. A. Verhoef, Soil Biol. Biochem. 23, 15 (A) and leaf litter mass loss (B) in relation to all combinations in which L. rubellus was mean functional dissimilarity (10) of species in 5. J. Mikola, R. D. Bardgett, K. Hedlund, in Biodiversity the community. Each series of dots represents present (Fig. 2, combinations A, E, F, G, and and Ecosystem Functioning, Synthesis and Perspec- a treatment (n 0 5 replicates per treatment; tives, M. Loreau, S. Naeem, P. Inchausti, Eds. (Oxford some dots overlap). Letters at the top of the chemical changes in the leaf litter by isopods Univ. Press, Oxford, 2002), pp. 169–180.
figure refer to the species combination given in 6. D. A. Wardle, O. Zackrisson, G. Ho¨rnberg, C. Gallet, or millipedes. Inhibition occurred between table S1 (10). A significant positive regression O. asellus and P. denticulatus (Fig. 2, com- 7. D. Tilman, J. Knops, D. Wedin, P. Reich, in Biodiversity between the mean functional dissimilarity of bination C). Both species have similar body and Ecosystem Functioning, Synthesis and Perspec- the communities and the net diversity effect forsoil respiration (linear regression, F sizes and showed the strongest comminuting tives, M. Loreau, S. Naeem, P. Inchausti, Eds. (Oxford Univ. Press, Oxford, 2002), pp. 21–35.
P 0 0.001) and leaf litter mass loss (linear activity (Table 1), suggesting possible com- 8. D. C. Coleman, D. A. Crossley, Fundamentals of Soil petition for leaf litter of a specific frag- Ecology (Academic Press, New York, 1996).
that positive net diversity effects are more ment size. Neutral net diversity effects were 9. B. Walker, A. Kinzig, J. Langridge, Ecosystems 2, 95 observed for species combinations lacking 10. Materials and methods are available as supporting functionally dissimilar species. Functional dissim- L. rubellus or O. asellus and P. denticulatus ilarity was related to neither species number nortaxonomic group number.
11. The relations between species number and soil process rates were best explained by exponentialregression (curve fits of linear regression and ex-ponential regression were compared using theirresidual sum of squares). Leaf mass loss: exponen- Table 1. The effect of single species (mean, n 0 5 per treatment) on four soil ecosystem processes related to decomposition. For each species, the total biomass (mean T SE, n 0 5 per species) added to the process rate after one species; leaf fragmentation: microcosms is given. Significant interspecific differences [one-way analysis of variance over all 0 3.81, P 0 0.026, saturation of process rate at treatments, with an a posteriori test on interspecific means using least-square differences (P e 0.05), one species; gross nitrate productivity: F with an unbalanced structure for all processes] between means within a column are marked with 11.21, P G 0.001, saturation of process rate after one different superscript letters. BC, Bray-Curtis; DW, dry weight.
12. M. Loreau, A. Hector, Nature 412, 72 (2001).
13. We thank D. Wardle for stimulating discussions.
M.P.B. was financially supported by an academy fellowship of the Royal Netherlands Academy of Science. The investigation was supported by the Arts and Sciences Research Council for Earth and LifeSciences (ALW), with financial aid from the Nether- lands Organization for Scientific Research (NWO) and from the Dutch Ministry of Agriculture, Nature and Food Quality through DWK program 384.
www.sciencemag.org/cgi/content/full/306/5698/1019/

Source: http://www.intrasense.nl/SENSE_INCLUDES/docs/verhoef_science.pdf