Methods
Species
The five grass species we used are common prairie, grassland
and old field species but differ in their successional status (Tilman 1988)
and their response to nutrient addition (Tilman 1987). They are the five most
abundant grasses during old field succession at Cedar Creek Natural History
Area (CCNHA), where this work was performed. Agrostis scabra Willd. (a
native bentgrass) and Agropyron repens (L. ) Beauv. (quack grass,
introduced from Europe) are both early successional species at CCNHA. Poa
pratensis L. (Kentucky bluegrass, introduced) dominates mid-successional
fields at CCNHA. Schizachyrium scoparium (Michx. ) Nash-Gould (little
bluestem, formerly Andropogon scoparius Michx. ) and Andropogon
gerardi Vitm. (big bluestem) are native prairie species that dominate
older successional fields at CCNHA. For brevity, we will refer to each
species by its genus.
Experimental Design and Methods
This research was performed in an
experimental garden at CCNHA, which is located on a large glacial outwash
sandplain in Isanti and Anoka counties of Minnesota. Each species was grown
in both monoculture and mixed species plots across a range of total soil
nitrogens. In August, 1985, we used a bulldozer to remove the upper 60 to 80
cm of soil from a 3 m x 34m area of a 25 yr old field, thus exposing
subsurface sand. This substrate was 93% sand, 3% clay, and 4% silt, had a pH
of 6.6, contained 0.3% organic matter and had a total soil nitrogen, as N, of
90 mg/kg of soil. This area was divided into ten regions, 3 m x 12 m each,
with 1 m walkways between them. Each of the ten regions was randomly assigned
to be one of ten different soil mixtures. Nine of the soil mixtures were
established by adding an amount of top soil per plot such that, once top soil
and subsurface sand had been mixed to a depth of 22 cm, top soil would
comprise approximately 0% (soil mixture 1), 10%, 20%, 40%, 50%, 55%, 70%, 90%
or 100% (soil mixture 9) of the upper 22 cm of soil in the plot. To keep the
surface of all soil mixtures level with their surroundings, the sand in each
was excavated to a depth equal to the amount of top soil to be added. Soil
mixture 10 was a duplicate of the 100% top soil treatment, but also received
6.55 g/m2/yr of additional N as NH4NO3. This was added monthly in proportion
to normal N mineralization at Cedar Creek (Pastor et al. 1987; April = 9%,
May = 33%, June = 28%, July = 15%, August = 17%). After top soil was added,
each plot was thoroughly rototilled twice. This resulted in a uniform, 22 cm
deep layer of soil that was underlain by unmixed sand. Approximately 60 m3 of
top soil were used. This soil, obtained off-site, was a Duelm sandy loam of
the Hubbard-Isanti-Duelm Association. It averaged 72% sand, 4% clay, and 24%
silt, had 2.9% organic matter, total soil N of about 1200 mg/kg, and a pH of
7.2. It was chosen because it was a sandy outwash soil with total soil
nitrogen content similar to that of undisturbed upland soils at CCNHA.
To assure that no nutrients other than N would be limiting, we analyzed the
soils, and based on these analyses, added 5.6 g/m2 of P, 8.9 g/m2 of K, 42
g/m2 of MgSO4, 8 g/m2 of Ca, 0.68 g/m2 of Zn, 0.28 g/m2 of Cu, 0.11 g/m2 of B,
and 0.19 g/m2 of Mn in April 1986. These nutrients were thoroughly
rototilled, to a depth of 22cm, into all plots. These nutrients were also
added to the soil surface in subsequent years. Soil pH was stable, at 7.2, in
all plots except the pure sand mixture and the 10% top soil mixture which
received fine ground commercial lime in 1987 and 1988 to adjust their pH to
7.2. All plots were watered to receive at least 1.5 cm/wk in April, May, and
October, and 2.5 cm/wk in June through September.
In April, 1986, galvanized sheet metal (26 gauge) was used to divide each of
the 10 soil mixtures into 64 experimental plots, each of which was 0.75 m x
0.75 m. To do this, the sheet metal, which was 25 cm tall, was driven into
the soil to a depth of 23 cm. The entire garden was then fenced to a height
of 2 m and to a depth of 1.3 m belowground to exclude all mammalian
herbivores. Seeds of Agrostis, Agropyron, Poa, Schizachyrium and
Andropogon were planted in the monoculture plots in late May, 1986, at
a density, based on germination trials, to yield 3000 seedlings/m2. Seeds were
covered with 0.5 cm of sand. There were 2 replicate monocultures of
Andropogon, 3 replicate monocultures of Poa and
Schizachyrium, and 4 replicate monocultures of Agrostis and
Agropyron for each soil mixture, giving a total of 160 monoculture
plots. This design was caused by limited space and the need for a balance
between monocultures and competition plots in the various competition
experiments that were performed simultaneously in these gardens. Monoculture
and competition treatments were randomly assigned to the 64 plots within a
soil mixture, with each soil mixture being independently randomized. For further
information, see fertilization details.