Recent research on biodiversity and ecosystem functioning has raised many questions, including disagreements about data interpretation and causes of observed results (e.g., Givnish 1994, Huston 1997, Aarssen 1997, Hodgson et al. 1998, Doak et al. 1998) and contradictions between experimental and observational studies (Wardle et al. 1997a, 1997b, Grime 1997). These critiques and alternative hypotheses have been met with various replies, but the best resolution to these debates may come from long-term field experiments in which diversity is directly controlled and detailed observations and related ecophysiological studies allow determination of underlying mechanisms. The Cedar Creek LTER began two such experiments, dubbed Biodiversity I and II, in 1994, and initiated a third, dubbed BioCON, in 1997.

Description of Experimental Methods

Biodiversity I (E123; http://www.lter.umn.edu/research/exper/e123; Tilman, Wedin and Knops 1996; Fig. 13 [pdf]) manipulates species diversity independently of composition. It has 147 plots, each 3 x 3 m , separated by 1-2 m paths. Plots were randomly assigned to have 1, 2, 4, 6, or 8 species (n=20 replicates each), 12 species (n=23), or 24 species (n=24). The species added, as seed, to each plot were chosen by separate, random draws from a pool of 24 grassland species.
Biodiversity II (E120; http://www.lter.umn.edu/research/exper/e120; Tilman et al. 1997a; Fig. 14 [pdf]) simultaneously manipulates species diversity, functional group diversity, and functional group composition. Its 245 plots, each 9 m x 9 m, were planted with 1, 2, 4, 8, or 16 species that were in either the C4 grass, C3 grass, legume, other forb, or woody functional groups. The composition and diversity of each plot was chosen randomly, but constrained to provide even coverage of all possible combinations of species and functional-group diversity.
BioCON (E141; http://www.lter.umn.edu/biocon/) directly controls plant diversity (1, 4, 9 or 16 perennial species randomly chosen from a pool of 16 species, planted as seed in 1997), N availability treatments (ambient soil vs. ambient soil + 4 g m-2 yr-1 N), and atmospheric CO2 concentrations (ambient vs 550 ppm, beginning in 1998) in a well-replicated split-plot experiment consisting of a full factorial combination of treatment levels in a completely randomized design. Its 296 individual plots, each 2 x 2 m, occur in six 20-m rings, three exposed to ambient CO2 and three to elevated CO2 using free-air CO2 enrichment (Lewin et al 1994 ). BioCON is an integral part of the LTER but mainly is funded by DOE.

Sampling methods for these experiments: Plant abundances are estimated via total plant cover (identified to species). The proportion of incident light intercepted by vegetation is measured. Biomass is measured once or twice per year by clipping, sorting, drying, and weighing vegetation in a 0.1 x 1.0 m strip per plot in Biodiversity I and BioCON, or four 0.1 m x 3.0 m strips per plot in Biodiversity II. Root mass is from three 5 cm diameter by 40 cm deep soil cores per clip strip, gently rinsed on a fine screen, dried and weighed. Soils are sampled for 0.01 M KCl extractable NO3 and NH4 and for total soil N using four 2.5 cm diameter by 20 cm deep soil cores per plot. Soil NO3 and NH4 at 40-60 cm depth are sampled to estimate leaching losses.

Results to Date

-Plant total cover and standing crop (correlates of productivity) were significantly increasing functions, and levels of unconsumed soil nitrate both in the rooting zone and below the rooting zone, were significantly decreasing functions of plant species diversity in both Biodiversity I (Fig. 9 [pdf], 15A [pdf]; Tilman et al.1996a; Tilman 1999a) and Biodiversity II (Fig. 15B [pdf]; Tilman et al. 1997a).
- Plant species diversity and plant functional group composition were approximately equally important determinants of plant community biomass and of levels of unconsumed soil nitrate.
-Our results in Biodiversity I and II provide tests of two alternative hypotheses for the effects of plant diversity on productivity and limiting nutrients: the sampling effect model (Huston 1997, Aarsen 1997, Tilman et al. 1997b) and the niche differentiation model (Swift and Anderson 1993, Naeem et al. 1994, 1995, Tilman et al. 1997b). As predicted by niche models but not by the sampling effect, results showed that (1) planted species coexisted but the sampling effect predicts displacement of all but one species; (2) many higher-diversity plots have greater standing crop than any plots of lower diversity (Fig. 15A, B [pdf]); (3) species significantly were inhibited mainly by species in their own functional group, and either not impacted by or positively impacted by species of other functional groups.
-In Biodiversity I, greater plant diversity was associated with a significantly lower number and biomass of invading weedy species (Fig. 10 [pdf]; Knops et al. 1999). Multiple regressions suggested that effects of diversity on invading species came mainly from lower soil NO3 in higher-diversity plots. Other analyses showed neighborhood distances to competitors and sizes of competitors also influenced the success of invaders (Naeem et al. 2000b).
-The species diversity of the arthropod community of Biodiversity II increased as both plant species and functional diversity increased (Fig. 11 [pdf]; Siemann et al. 1998, Knops et al. 1999). The diversity of herbivorous arthropods depended both on plant diversity and on the diversity of predatory and parasitoid arthropods, suggesting the importance of multi-trophic-level effects. Haddad et al. (in review) found that the effects of plant diversity on arthropod diversity seemed to result from greater plant biomass at higher diversity and the resulting greater number of insects.
- In BioCON, species diversity, elevated CO2 and N deposition all had significant effects and interactions on composition and ecosystem functioning during the 1998 and 1999 growing seasons (Fig. 3 D, E [pdf]; Reich et al, in prep.). Primary productivity and its response to added CO2 or N both increased with diversity, i.e., the community response was much greater than the average species response (Reich et al. in prep.). The stimulation of plant C stocks by elevated CO2 was four- to twenty-fold greater in 16-species plots than in monocultures. These data suggest that the reduction of diversity occurring globally may profoundly influence the ability of ecosystems to store C in response to increasing atmospheric CO2 and N deposition, and hence on the ability of ecosystems to be C sinks in the face of increasing atmospheric CO2 and N deposition.

Mechanistic, Synthetic and Related Studies:

The biodiversity experiments are the focus of many related studies. G. Burt-Smith and P. Grime are growing the Biodiversity I species under controlled conditions in Sheffield, UK to determine which plant traits are best correlated with observed species abundances in Biodiversity I. Joe Craine, a Ph. D. student of F. S. Chapin, is studying the roles of root and leaf longevity, respiration rates, photosynthetic rates, and other traits on abundance patterns in the chronosequence, in Biodiversity I and II, and in BioCON. We are determining effects of plant diversity on the decomposer foodweb by measuring microbial densities (DAPI epifluorescent enumeration) (Porter and Feig 1980), sole source C use profiles (Ecolog plates, Biolog Corp.)(Garland and Mills 1994), and microbial biomass(Islam and Weil 1998a, 1998b) in Biodiversity I.

Future Research:

These are the best replicated and most long-term biodiversity experiments in existence. They already have yielded insights, but their greatest contributions are likely to come in the next 5 to 10 years because they are just attaining asymptotic biomasses and compositions. For the next 6 years we will continue gathering the variables listed above and the related mechanistic studies. In addition, we will pursue a series of related questions:
-How does diversity influence population and ecosystem stability? We explored this with the N addition experiment (Tilman and Downing 1994, Tilman 1996a), but it lacks direct control of diversity, necessitating control of confounding variables by multiple regression. However, measurement of year-to-year fluctuation in total biomass and species abundances in plots in all three biodiversity experiments will directly assess the effect of diversity and composition on stability. Given normal climatic variation, 6 added years of data will provide a strong test of the diversity-stability hypothesis.
-Why does diversity influence productivity? Current analyses support niche differentiation over sampling effects, but give no insight into the type of differentiation. We will work to determine the interspecific differences that cause diversity, thus addressing the mechanisms of coexistence.
-How do plant diversity and composition impact the rest of the food chain? We will continue annual arthropod sampling (counted, to species; begun in 1996) in Biodiversity II (Siemann et al. 1998) to see the effects of plant diversity and composition on arthropod abundances, food chain structure, and stability, and possible feedback effects of arthropods on plant dynamics
-How do plant diversity and composition impact microbial diversity and composition and microbially-mediated processes? Ecosystem processes are ultimately a function of microbial regulation of N, C and P cycles, but the significance of the extraordinary microbial diversity is poorly understood (Wall and Moore 1999). In Section 2. C. we describe how we will use the biodiversity experiments to explore the effects of plant diversity and composition on the diversity and composition of the soil microbial community.
-How do plant diversity and composition influence the susceptibility of a community to invasion by exotic species? Biological invasion is widespread, but the mechanisms controlling susceptability to invasion are debated (Robinson et al. 1995, Levine et al. 1995, Planty-Tabacchi et al. 1996, Wiser et al. 1998, Levine and D'Antonio 1999, Stohlgren et al. 1999, Tilman 1999a). We propose a multi-investigator, multi-pronged approach to this question. Our team includes C. Brown, S. Naeem, M. Davis (in collaboration with P. Grime and K. Thompson), J. Knops, and D. Tilman. We will explore mechanisms of neighborhood interaction (Harper 1977, Pacala and Silander 1985, Goldberg 1987, Naeem et al. 2000b), the role of limiting resources (Tilman 1999a, Davis et al., in review), the effects of disturbance and climatic variation, and the relations between the traits of potential invaders versus established plants.