Experimental Design



Biotic-Abiotic Interaction Experiment


a) Establishment of experimental ecosystems (mesocosms).

Design of the ecosystem unit began in 1994 with library research on terrestrial mesocosms and lysimeters (e.g., Beyers and Odum 1993, Knight and Will 1977, Minderman and Leeflang 1968, Joffe 1933), and discussions with members of the Hubbard Brook Sandbox Experiment (Bormann et al. 1993). Field mesocosms can be an effective tools for testing ecological hypotheses (Bormann et al. 1987, Teuben and Verhoef 1992). In the summer and fall of 1994, two sites were selected; they were prepared for the experiment (drainage, installation of utilities, etc.); 49 tanks were placed in a randomized block design; and 48 tanks were filled with one of two soil substrates, carefully mixed in total to insure uniformity among mesocosms. The tanks were constructed of linear polyethylene, 2.44 m in diameter, and 1 m in depth to the bottom of the soil fill. A conical section below the soil was filled with nutrient-poor coarse granite to allow drainage. A fine geotextile cloth was used to keep the soil from infiltrating the gravel layer. A 15.2 cm diameter PVC stand pipe was placed vertically to the bottom of the tanks and outfitted with a vacuum extraction system to permit removal and analysis of drainage waters.

b) Description of biotic and abiotic factors

To test the interaction hypotheses, three main variables were selected for study of whole ecosystem interactions:

Soil - Two unweathered, glacial lake deposit sand differing in physical and chemical properties

  1. medium sand, low calcium: from Milton, VT
  2. coarse sand, high calcium: from Kullman, VT

    • Table showing size fractions & chemistry

      > 250 µm (%) < 250 µm (%) Ca (ppm*) Mg (ppm*)

    Milton, VT 48 52 39.3 30.3

    Kullman, VT 77 23 3450 38.9

          * ammonium acetate extractable

Community - four assemblages of tree species

  1. red pine and red maple (Pinus resinosa & Acer rubrum)
  2. red pine and grey birch (Pinus resinosa & Betula populifolia)
  3. eastern white pine and red maple (Pinus strobus & Acer rubrum)
  4. eastern white pine and grey birch (Pinus strobus & Betula populifolia)

Plant material was obtained from a variety of sources. Gray birch was obtained from a Champlain Valley (Chittenden Co.) and a northern Vermont site (Lamoille Co.). White pine was collected from a population in the Champlain Valley and northern Vermont as well. Red pine was obtained from an upper pennisula location (Wexford Co.) and a lower penninsula site (Chippewa Co.) in Michigan. Red maple seed stock was grown from seed from northern and southern populations provided by the Pettawawa Forest Experiment Station in Canada.

Location - the environmental characteristics associated with two sites:

  1. Sub-boreal (northerly, cold): at the University experimental forest at Wolcott, VT
  2. Temperate (southerly, warmer due to a lake effect): at the U.S. Forest Service site in the Champlain Valley, S. Burlington, VT

Differences between locations include: precipitation volume and chemistry, dry deposition, air and soil temperature, solar radiation. To quantify the these potential differences between locations, air temperature, precipitation, solar radiation (PAR), and windspeed are collected/monitored continuously at both sites using automatic dataloggers (Campbell Scientific 21X, ADC-2, LiCor 1000) and other collection devices.

c) Experimental design

In order to the quantify the interaction components, a rigorous experimental design was critical. Three blocks of eight mesocosms each were established at the southern site and two blocks of eight at the northern site. A block consisted of each of two soils planted with the four commnity assemblages. Eight additional non-vegetated mesocosms (four at each site), and one tank with granite only were established as controls. The total number of experimental units is thus 49. In addition, six nursery plots containing the soil x location x tree species assemblages were established to provide a source of material for replacements during the establishment phase and for intermediate destructive harvests needed for nutrient and dimension analysis. Although we recognize that inclusion of single species experimental units would have provided useful additional information, it would result in either a substantial expansion of the project or a scaling down of the multi-species community focus. We felt that multi-species component was critical, and that we could gain some individual species information though orthogonal contrasts by partitioning of the community component of variation (Table 4). In fact, this approach has provided a powerful mechanism for examining the contribution of species specific associations (biotic x biotic interactions) in the development of community-driven variation during ecosystem development.

d) Description of response variables

To assess the importance of the three variables and their interaction, several population, community, and ecosystem-level dependent variables will be assessed. These include:

Community, species, population, half-sib family relevant variables:

  • growth (height, diameter)
  • living biomass accumulation and allocation (aboveground, belowground)
  • nutrient acccumulation and allocation (Ca, Mg, K, P, N)
  • nutrient use efficiency (nutrient/biomass relationship)

Whole ecosystem (mesocosm) relevant variables:

  • organic matter accumulation and distribution (soil, litter, living biomass)
  • nutrient accumulation and distribution (soil, litter, living biomass; Ca, Mg, K, P, N)
  • nutrient conservation (inputs/outputs)
  • degree of hydrologic control (drainage/precipitation, storage)

The mesocosm design permits measurement of a wide variety of population, species, community and ecosystem-level variables. The overall experimental design allows for detailed measurement of stocks and fluxes of water and nutrients.