Diversity and Productivity in a Long-Term Grassland Experiment D. Tilman, P. B. Reich, J. Knops, D. Wedin, T. Mielke, C. Lehman

A field at Cedar Creek Natural History Area, Minnesota, was herbicided and burned in August, 1993, had 6-8 cm of soil removed to reduce the seed bank, was plowed and harrowed, then divided into plots of which 168 form this experiment. All plots received, in total, 10 g m^-2 of seed in May 1994 and 5 g m^-2 in May 1995, with seed mass divided equally among species. Two species that did not germinate in 1994 were replaced with different species in the same functional groups in the 1995, with their seeding proportionate to 10 g m^-2. Three species (Elymus canadensis, Poa pratensis, Panicum virgatum) did not establish in one of their two original monocultures, even after reseeding, and these three plots were abandoned after 1996. Plots were initially 13 m x 13 m, and reduced to 9 m x 9 m in 1998 by mown walkways. Treatments were maintained by weeding 3 or 4 times/year. Selective herbicides were also used through 1997. Plots were burned annually in the spring before growth began. Plots were sampled in mid-August for aboveground living plant biomass by clipping, drying, and weighing four 0.1 x 3.0 m vegetation strips per plot from 1996 through 1999, and eight strips per plot in 2000. Only non-mown areas were sampled. Different areas were sampled each year. Beginning in 1997, root biomass was sampled immediately after clipping by collecting three 5 cm diameter x 30 cm deep cores per clipped strip. Roots were washed free of soil, sorted from other organic material, dried and weighed. The sum of root mass and aboveground mass, on an aerial basis, is called total plant biomass, a measure of total living carbon.

Supplement B. Effects of observed plant diversity on total and aboveground biomass in 2000. Two indices of plant diversity are used. The first index, "observed species number," is the number of planted species observed in one or more of four 0.5 m² quadrats within each plot. The second, "effective species richness," is e^H', where H' is the Shannon diversity index based on abundances (percent cover) of planted species. Analyses, using general linear models, determined effects of plant diversity (continuous variable; entered first using Type I SS) and of functional group composition (categorical variable; entered second) on total biomass and on aboveground biomass. N=168. Overall model d.f.=28 and error d.f.=139, with plant diversity d.f.=1 and composition d.f.=27. The last columns show similar analyses, but with plant diversity entered second, after composition (i.e., Type III regressions).

					Plant Diversity		Functional Group		Plant Diversity
		Overall			(entered 1st)		Comp. (entered 2nd)		(entered 2nd)
Variable Analyzed	Year	R2	F Value	P	F Value	P	F Value	P	F Value	P
Observed Species Number:
Total Biomass	2000	0.68	10.6	<0.001	157.	<0.001	5.20	<0.001	13.3	<0.001
Aboveground Biomass	2000	0.63	8.40	<0.001	133.	<0.001	3.78	<0.001	17.7	<0.001
Effective Species Richness (eH'):
Total Biomass	2000	0.66	9.53	<0.001	136.	<0.001	4.84	<0.001	7.82	0.006
Aboveground Biomass	2000	0.61	7.73	<0.001	132.	<0.001	3.11	<0.001	14.0	<0.001

Supplement C. GLM Type III regressions (each variable considered after controlling for the other variable) for effects of species number and functional group composition on total biomass and on aboveground biomass in 2000, for two different subsets of plots. The first subset consists of all plots containing at least 1 of the 9 species with highest total or aboveground biomass in monoculture in 2000. The second subset consists of plots containing at least 2 of these top 9 species.

	Overall			Plant Diversity		Functional Group
Variable Analyzed	N	F-Value	P	F-Value	P	F-Value	P
At Least 1 of 9 Top Species:
Total Biomass	145	6.08	<0.001	10.5	0.002	3.33	<0.001
Aboveground Biomass	140	4.43	<0.001	8.85	0.004	2.72	<0.001
At Least 2 of 9 Top Species:
Total Biomass	95	3.47	<0.001	11.9	<0.001	2.67	0.002
Aboveground Biomass	88	2.34	0.005	5.04	0.028	1.87	0.030

Supplement D. Goodness of fit, as estimated by R2 values, for regressions of total biomass in 2000 on the log of the number of the N top species that were present in each plot. For instance, when N=4, the four species with the greatest percent cover in 2000 in the 16-species plots were considered. These four species were Schizachyrium scoparium, Lupinus perennis, Sorghastrum nutans and Andropogon gerardi. For each plot, we determined how many of these four species had been planted in the plot, and used this as a diversity index. Log [Diversity Index +1] was then used as the independent variable in a regression in which total biomass in 2000 was the dependent variable. Such regressions were repeated for the other diversity indices (values of N, number of top species considered, ranging from 2 to 18). Web fig. 1 shows that N=4 gave the best fit for total biomass.

Similar analyses were performed, but using aboveground biomass in 2000 as the dependent variable in a series of regressions. Web fig. 2 shows that the highest R² value occurred for N=9, i.e., when the number of the 9 most abundant species that were present in each plot was the independent variable. The next highest R² value occurred for N=13. The fitted curve had its peak at N=11. Each value of N from 7 to 15 had a higher R2 value than any N(6.

Supplemental Figure 1. Dependence of R² in Total Biomass on Number of Top Species Considered

Medium version | Full size version

Supplemental Figure 2. Dependence of R² in Aboveground Biomass on Number of Top Species Considered

Medium version | Full size version