S CIENCE’ S C OMPASS ● REVIEW REVIEW: ECOLOGY Biodiversity and Ecosystem Functioning: Current Knowledge and Future Challenges M. Loreau,1* S. Naeem,2 P. Inchausti,1 J. Bengtsson,3 J. P. Grime,4 A. Hector,5 D. U. Hooper,6 M. A. Huston,7 D. Raffaelli,8 B. Schmid,9 D. Tilman,10 D. A. Wardle4 primary production in grassland ecosystems The ecological consequences of biodiversity loss have aroused considerable interest (20–23). Because plants, as primary produc- and controversy during the past decade. Major advances have been made in describing ers, represent the basal component of most the relationship between species diversity and ecosystem processes, in identifying ecosystems, they represented the logical functionally important species, and in revealing underlying mechanisms. There is, place to begin detailed studies. Several, al- however, uncertainty as to how results obtained in recent experiments scale up to though not all, experiments using randomly landscape and regional levels and generalize across ecosystem types and processes. assembled communities found that primary Larger numbers of species are probably needed to reduce temporal variability in production exhibits a positive relationship ecosystem processes in changing environments. A major future challenge is to with plant species and functional-group di- determine how biodiversity dynamics, ecosystem processes, and abiotic factors versity (Fig. 1). interact. These results attracted a great deal of interest, not only because they were novel, but also because they seemed counter to pat- he relationship between biodiversity the current debate in which scientists dis- terns often observed in nature, where the and ecosystem functioning has emerged agree about the relative importance of func- most productive ecosystems are typically T as a central issue in ecological and tional substitutions and declining species characterized by low species diversity (26, environmental sciences during the last de- richness as determinants of changes in eco- 27). The controversy over the interpretation cade. Increasing domination of ecosystems system functioning. Comparative studies of these results started with the realization by humans is steadily transforming them into have begun to reveal the extent to which that they can be generated by different mech- depauperate systems (1, 2). Because ecosys- functional substitutions alter ecosystem anisms. The mechanisms discussed so far tems collectively determine the biogeochemi- properties such as productivity, decompo- may be grouped into two main classes. First cal processes that regulate the Earth system, sition rates, nutrient cycling, and resistance are local deterministic processes, such as the potential ecological consequences of and resilience to perturbations (17, 18). On niche differentiation and facilitation, which biodiversity loss have aroused considerable the other hand, a new wave of experimental increase the performance of communities interest (3–9). studies has manipulated species richness by above that expected from the performance of Recent experimental and theoretical using synthesized model ecosystems in individual species grown alone. We will sub- work in this area has also led to animated both terrestrial and aquatic environments sume them here under the term “complemen- debates and controversies (10–14). Human (19–25). Both approaches suggest that a tarity” for convenience’s sake. Second are impacts on the environment from local to large pool of species is required to sustain local and regional stochastic processes in- global scales cause not only a general de- the assembly and functioning of ecosys- volved in community assembly, which are cline in diversity, but also predictable func- tems in landscapes subject to increasingly mimicked in experiments by random sam- tional shifts as sets of species with partic- intensive land use. It is not yet clear, how- pling from a species pool. Random sampling ular traits are replaced by other sets with ever, whether this dependence on diversity coupled with local dominance of highly pro- different traits (15, 16). This has resulted in arises from the need for recruitment of a ductive species can also lead to increased few key species from within the regional average primary production with increasing 1Laboratoire d’Ecologie, UMR 7625, Ecole Normale species pool or is due to the need for a rich diversity, because plots that include many Supe«rieure, 46 rue d’Ulm, FÐ75230 Paris Cedex 05, assortment of complementary species with- species have a higher probability of contain- France. 2Department of Zoology, University of Wash- ington, 24 Kincaid Hall, Box 351800, Seattle, WA in particular ecosystems. ing highly productive species (10, 11, 28). 98195Ð1800, USA. 3Department of Ecology and Crop In this article, we seek to set a common Two issues are involved in this controversy: Production Science, Swedish University of Agricultural framework to understand these issues, to Are stochastic community assembly process- Sciences, Box 7043, SEÐ750 07 Uppsala, Sweden. move beyond past differences of opinion, and es relevant? And what is the relative impor- 4Department of Animal and Plant Sciences, University of Sheffield, Sheffield, S10 2TN, UK. 5NERC Centre for to define new perspectives, after a recent tance of the two classes of mechanisms? Population Biology, Imperial College at Silwood Park, conference held in Paris. We do not attempt There are diverging views on the rele- Ascot, Berks, SL5 7PY, UK. 6Department of Biology, to comprehensively review these issues, ele- vance of the sampling component of biodi- Western Washington University, 516 High Street, ments of which can be found elsewhere (3– versity effects. As sampling processes were 7 Bellingham, WA 98225Ð9160, USA. Environmental 9). Rather, we focus on major regions where not an explicit part of the initial hypotheses, Sciences Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831Ð6400, USA. 8Environment De- recent advances have been made. they have been viewed by some as “hidden partment, University of York, York, YO10 5DD, UK. treatments” (10), whereas others have viewed 9Institut fu¬r Umweltwissenschaften, Universita¬t Zu¬- Experimentally Altered Diversity them as the simplest possible mechanism rich, Winterthurerstrasse 190, CHÐ8057 Zu¬rich, Swit- Although the first study that experimentally linking diversity and ecosystem functioning zerland. 10Department of Ecology, Evolution and Be- havior, University of Minnesota, St. Paul, MN 55108, manipulated diversity did so across several (28). This debate should be resolved through USA. trophic levels (19), later studies focused increasing knowledge about the patterns and *To whom correspondence should be addressed. E- mainly on effects of plant taxonomic diversi- processes of biodiversity loss in nature, mail: [email protected] ty and plant functional-group diversity on which are still poorly known overall. If dom- 804 26 OCTOBER 2001 VOL 294 SCIENCE www.sciencemag.org S CIENCE’ S C OMPASS Fig. 1. Responses of total (A) or aboveground (B and C) plant biomass (in ported elsewhere (23). Filled squares and line 1, Germany; filled circles grams per meter squared) to experimental manipulations of plant species and line 2, Portugal; filled triangles and line 3, Switzerland; solid dia- richness (A and B) or functional-group richness (C) in grasslands in monds and line 4, Greece; open squares and line 5, Ireland; open circles Minnesota (A) (31), across Europe (B) (23), and in California (C) (22). and line 6, Sweden; open diamonds and line 7, Sheffield (UK); open Points in (A) and (B) are data for individual plots. In (B) different diamonds and line 8, Silwood Park (UK). Symbols in (C) correspond to regression slopes are shown for the eight sites to focus on between- functional groups and their combinations: B, bare ground; E, early-season location differences rather then the general log-linear relationship re- annuals; L, late-season annuals; P, perennial bunchgrasses; N, N fixers. inant species control ecosystem processes Fig. 2. Hypothesized mecha- and mostly rare species go extinct, the vagar- nisms involved in biodiversity ies of community assembly or disassembly experiments using synthetic may have little relevance. But environmental communities. Sampling effects are involved in community as- changes and landscape fragmentation could sembly, such that communi- prevent recruitment of appropriate dominants ties that have more species (29). Also, climate change could lead to grad- have a greater probability of ual losses of species as abiotic conditions containing a higher pheno- begin to exceed species’ tolerance limits. typic trait diversity. Pheno- Such losses could be random with respect to typic diversity then maps onto ecosystem processes species effects on any given ecosystem pro- through two main mecha- cess, leading to patterns of process response nisms: dominance of species to changes in diversity similar to those ob- with particular traits, and served in randomly assembled communities. complementarity among spe- It should be emphasized that recent experi- cies with different traits. In- ments were not intended to reproduce any termediate scenarios involve complementarity among par- particular sequence of species loss; they re- ticular species or functional flect potential patterns, unaffected by corre- groups or, equivalently, dom- lations between diversity loss and composi- inance of particular subsets tional changes, rather than actual predictions of complementary species.